diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,14033 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.1953125, + "eval_steps": 500, + "global_step": 2000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 9.765625e-05, + "grad_norm": 79.3869857788086, + "learning_rate": 3.3333333333333337e-06, + "loss": 7.7773, + "step": 1 + }, + { + "epoch": 0.0001953125, + "grad_norm": 80.6998291015625, + "learning_rate": 6.6666666666666675e-06, + "loss": 7.7617, + "step": 2 + }, + { + "epoch": 0.00029296875, + "grad_norm": 67.86173248291016, + "learning_rate": 1e-05, + "loss": 7.6523, + "step": 3 + }, + { + "epoch": 0.000390625, + "grad_norm": 31.295886993408203, + "learning_rate": 1.3333333333333335e-05, + "loss": 7.3242, + "step": 4 + }, + { + "epoch": 0.00048828125, + "grad_norm": 27.488962173461914, + "learning_rate": 1.6666666666666667e-05, + "loss": 7.2773, + "step": 5 + }, + { + "epoch": 0.0005859375, + "grad_norm": 16.267820358276367, + "learning_rate": 2e-05, + "loss": 7.1641, + "step": 6 + }, + { + "epoch": 0.00068359375, + "grad_norm": 37.23747253417969, + "learning_rate": 2.3333333333333336e-05, + "loss": 7.2812, + "step": 7 + }, + { + "epoch": 0.00078125, + "grad_norm": 17.467384338378906, + "learning_rate": 2.666666666666667e-05, + "loss": 7.0508, + "step": 8 + }, + { + "epoch": 0.00087890625, + "grad_norm": 9.459285736083984, + "learning_rate": 3e-05, + "loss": 6.9375, + "step": 9 + }, + { + "epoch": 0.0009765625, + "grad_norm": 8.56839656829834, + "learning_rate": 3.3333333333333335e-05, + "loss": 6.8789, + "step": 10 + }, + { + "epoch": 0.00107421875, + "grad_norm": 24.074443817138672, + "learning_rate": 3.6666666666666666e-05, + "loss": 6.8672, + "step": 11 + }, + { + "epoch": 0.001171875, + "grad_norm": 11.778696060180664, + "learning_rate": 4e-05, + "loss": 6.7969, + "step": 12 + }, + { + "epoch": 0.00126953125, + "grad_norm": 14.282492637634277, + "learning_rate": 4.3333333333333334e-05, + "loss": 6.8047, + "step": 13 + }, + { + "epoch": 0.0013671875, + "grad_norm": 8.801307678222656, + "learning_rate": 4.666666666666667e-05, + "loss": 6.7188, + "step": 14 + }, + { + "epoch": 0.00146484375, + "grad_norm": 8.514760971069336, + "learning_rate": 5e-05, + "loss": 6.7148, + "step": 15 + }, + { + "epoch": 0.0015625, + "grad_norm": 9.48074722290039, + "learning_rate": 5.333333333333334e-05, + "loss": 6.6641, + "step": 16 + }, + { + "epoch": 0.00166015625, + "grad_norm": 7.557260036468506, + "learning_rate": 5.6666666666666664e-05, + "loss": 6.6172, + "step": 17 + }, + { + "epoch": 0.0017578125, + "grad_norm": 6.613265037536621, + "learning_rate": 6e-05, + "loss": 6.5977, + "step": 18 + }, + { + "epoch": 0.00185546875, + "grad_norm": 13.777167320251465, + "learning_rate": 6.333333333333335e-05, + "loss": 6.6328, + "step": 19 + }, + { + "epoch": 0.001953125, + "grad_norm": 6.775121688842773, + "learning_rate": 6.666666666666667e-05, + "loss": 6.5391, + "step": 20 + }, + { + "epoch": 0.00205078125, + "grad_norm": 9.053375244140625, + "learning_rate": 7.000000000000001e-05, + "loss": 6.543, + "step": 21 + }, + { + "epoch": 0.0021484375, + "grad_norm": 5.147669792175293, + "learning_rate": 7.333333333333333e-05, + "loss": 6.5039, + "step": 22 + }, + { + "epoch": 0.00224609375, + "grad_norm": 9.595170974731445, + "learning_rate": 7.666666666666667e-05, + "loss": 6.5, + "step": 23 + }, + { + "epoch": 0.00234375, + "grad_norm": 6.9720964431762695, + "learning_rate": 8e-05, + "loss": 6.4883, + "step": 24 + }, + { + "epoch": 0.00244140625, + "grad_norm": 9.805245399475098, + "learning_rate": 8.333333333333333e-05, + "loss": 6.4414, + "step": 25 + }, + { + "epoch": 0.0025390625, + "grad_norm": 5.9478044509887695, + "learning_rate": 8.666666666666667e-05, + "loss": 6.4297, + "step": 26 + }, + { + "epoch": 0.00263671875, + "grad_norm": 8.962249755859375, + "learning_rate": 8.999999999999999e-05, + "loss": 6.4297, + "step": 27 + }, + { + "epoch": 0.002734375, + "grad_norm": 5.202462673187256, + "learning_rate": 9.333333333333334e-05, + "loss": 6.375, + "step": 28 + }, + { + "epoch": 0.00283203125, + "grad_norm": 5.034671306610107, + "learning_rate": 9.666666666666667e-05, + "loss": 6.3555, + "step": 29 + }, + { + "epoch": 0.0029296875, + "grad_norm": 7.446832656860352, + "learning_rate": 0.0001, + "loss": 6.3477, + "step": 30 + }, + { + "epoch": 0.00302734375, + "grad_norm": 5.554169654846191, + "learning_rate": 0.00010333333333333333, + "loss": 6.3086, + "step": 31 + }, + { + "epoch": 0.003125, + "grad_norm": 5.340260028839111, + "learning_rate": 0.00010666666666666668, + "loss": 6.293, + "step": 32 + }, + { + "epoch": 0.00322265625, + "grad_norm": 6.529757976531982, + "learning_rate": 0.00011, + "loss": 6.2812, + "step": 33 + }, + { + "epoch": 0.0033203125, + "grad_norm": 8.140447616577148, + "learning_rate": 0.00011333333333333333, + "loss": 6.2656, + "step": 34 + }, + { + "epoch": 0.00341796875, + "grad_norm": 5.15455961227417, + "learning_rate": 0.00011666666666666667, + "loss": 6.2266, + "step": 35 + }, + { + "epoch": 0.003515625, + "grad_norm": 7.219542980194092, + "learning_rate": 0.00012, + "loss": 6.2539, + "step": 36 + }, + { + "epoch": 0.00361328125, + "grad_norm": 5.522006988525391, + "learning_rate": 0.00012333333333333334, + "loss": 6.2305, + "step": 37 + }, + { + "epoch": 0.0037109375, + "grad_norm": 4.108240127563477, + "learning_rate": 0.0001266666666666667, + "loss": 6.2109, + "step": 38 + }, + { + "epoch": 0.00380859375, + "grad_norm": 5.117391586303711, + "learning_rate": 0.00013000000000000002, + "loss": 6.1992, + "step": 39 + }, + { + "epoch": 0.00390625, + "grad_norm": 9.84037971496582, + "learning_rate": 0.00013333333333333334, + "loss": 6.2227, + "step": 40 + }, + { + "epoch": 0.00400390625, + "grad_norm": 5.794247150421143, + "learning_rate": 0.00013666666666666666, + "loss": 6.1953, + "step": 41 + }, + { + "epoch": 0.0041015625, + "grad_norm": 5.9350810050964355, + "learning_rate": 0.00014000000000000001, + "loss": 6.1836, + "step": 42 + }, + { + "epoch": 0.00419921875, + "grad_norm": 6.219013214111328, + "learning_rate": 0.00014333333333333334, + "loss": 6.168, + "step": 43 + }, + { + "epoch": 0.004296875, + "grad_norm": 4.501729488372803, + "learning_rate": 0.00014666666666666666, + "loss": 6.1055, + "step": 44 + }, + { + "epoch": 0.00439453125, + "grad_norm": 6.2689948081970215, + "learning_rate": 0.00015, + "loss": 6.1328, + "step": 45 + }, + { + "epoch": 0.0044921875, + "grad_norm": 4.54839563369751, + "learning_rate": 0.00015333333333333334, + "loss": 6.1055, + "step": 46 + }, + { + "epoch": 0.00458984375, + "grad_norm": 4.619740009307861, + "learning_rate": 0.0001566666666666667, + "loss": 6.082, + "step": 47 + }, + { + "epoch": 0.0046875, + "grad_norm": 7.260074615478516, + "learning_rate": 0.00016, + "loss": 6.0938, + "step": 48 + }, + { + "epoch": 0.00478515625, + "grad_norm": 5.270298480987549, + "learning_rate": 0.00016333333333333334, + "loss": 6.0859, + "step": 49 + }, + { + "epoch": 0.0048828125, + "grad_norm": 6.634472846984863, + "learning_rate": 0.00016666666666666666, + "loss": 6.0625, + "step": 50 + }, + { + "epoch": 0.00498046875, + "grad_norm": 4.172934532165527, + "learning_rate": 0.00017, + "loss": 6.0469, + "step": 51 + }, + { + "epoch": 0.005078125, + "grad_norm": 5.719873905181885, + "learning_rate": 0.00017333333333333334, + "loss": 6.0273, + "step": 52 + }, + { + "epoch": 0.00517578125, + "grad_norm": 4.851602554321289, + "learning_rate": 0.00017666666666666666, + "loss": 6.0234, + "step": 53 + }, + { + "epoch": 0.0052734375, + "grad_norm": 4.784290313720703, + "learning_rate": 0.00017999999999999998, + "loss": 5.9883, + "step": 54 + }, + { + "epoch": 0.00537109375, + "grad_norm": 5.738783359527588, + "learning_rate": 0.00018333333333333334, + "loss": 6.0078, + "step": 55 + }, + { + "epoch": 0.00546875, + "grad_norm": 4.525517463684082, + "learning_rate": 0.0001866666666666667, + "loss": 5.9844, + "step": 56 + }, + { + "epoch": 0.00556640625, + "grad_norm": 5.545010566711426, + "learning_rate": 0.00019, + "loss": 6.0078, + "step": 57 + }, + { + "epoch": 0.0056640625, + "grad_norm": 5.725442409515381, + "learning_rate": 0.00019333333333333333, + "loss": 5.9688, + "step": 58 + }, + { + "epoch": 0.00576171875, + "grad_norm": 4.413902282714844, + "learning_rate": 0.00019666666666666666, + "loss": 5.9414, + "step": 59 + }, + { + "epoch": 0.005859375, + "grad_norm": 4.734899520874023, + "learning_rate": 0.0002, + "loss": 5.9609, + "step": 60 + }, + { + "epoch": 0.00595703125, + "grad_norm": 4.493259906768799, + "learning_rate": 0.00020333333333333333, + "loss": 5.9648, + "step": 61 + }, + { + "epoch": 0.0060546875, + "grad_norm": 4.654330253601074, + "learning_rate": 0.00020666666666666666, + "loss": 5.957, + "step": 62 + }, + { + "epoch": 0.00615234375, + "grad_norm": 5.410141944885254, + "learning_rate": 0.00021, + "loss": 5.9766, + "step": 63 + }, + { + "epoch": 0.00625, + "grad_norm": 3.376593589782715, + "learning_rate": 0.00021333333333333336, + "loss": 5.9062, + "step": 64 + }, + { + "epoch": 0.00634765625, + "grad_norm": 5.334528923034668, + "learning_rate": 0.00021666666666666668, + "loss": 5.9141, + "step": 65 + }, + { + "epoch": 0.0064453125, + "grad_norm": 4.844014644622803, + "learning_rate": 0.00022, + "loss": 5.9062, + "step": 66 + }, + { + "epoch": 0.00654296875, + "grad_norm": 4.3729939460754395, + "learning_rate": 0.00022333333333333333, + "loss": 5.9023, + "step": 67 + }, + { + "epoch": 0.006640625, + "grad_norm": 6.744809627532959, + "learning_rate": 0.00022666666666666666, + "loss": 5.9648, + "step": 68 + }, + { + "epoch": 0.00673828125, + "grad_norm": 4.010498046875, + "learning_rate": 0.00023, + "loss": 5.8945, + "step": 69 + }, + { + "epoch": 0.0068359375, + "grad_norm": 5.671877384185791, + "learning_rate": 0.00023333333333333333, + "loss": 5.9414, + "step": 70 + }, + { + "epoch": 0.00693359375, + "grad_norm": 3.9647581577301025, + "learning_rate": 0.00023666666666666668, + "loss": 5.8672, + "step": 71 + }, + { + "epoch": 0.00703125, + "grad_norm": 4.879064083099365, + "learning_rate": 0.00024, + "loss": 5.8672, + "step": 72 + }, + { + "epoch": 0.00712890625, + "grad_norm": 5.130377292633057, + "learning_rate": 0.00024333333333333336, + "loss": 5.8711, + "step": 73 + }, + { + "epoch": 0.0072265625, + "grad_norm": 5.8605122566223145, + "learning_rate": 0.0002466666666666667, + "loss": 5.875, + "step": 74 + }, + { + "epoch": 0.00732421875, + "grad_norm": 5.346503257751465, + "learning_rate": 0.00025, + "loss": 5.9023, + "step": 75 + }, + { + "epoch": 0.007421875, + "grad_norm": 5.273228645324707, + "learning_rate": 0.0002533333333333334, + "loss": 5.8867, + "step": 76 + }, + { + "epoch": 0.00751953125, + "grad_norm": 3.8530828952789307, + "learning_rate": 0.00025666666666666665, + "loss": 5.8477, + "step": 77 + }, + { + "epoch": 0.0076171875, + "grad_norm": 5.188773155212402, + "learning_rate": 0.00026000000000000003, + "loss": 5.8633, + "step": 78 + }, + { + "epoch": 0.00771484375, + "grad_norm": 3.4620485305786133, + "learning_rate": 0.0002633333333333333, + "loss": 5.8398, + "step": 79 + }, + { + "epoch": 0.0078125, + "grad_norm": 4.68696403503418, + "learning_rate": 0.0002666666666666667, + "loss": 5.8359, + "step": 80 + }, + { + "epoch": 0.00791015625, + "grad_norm": 4.243693828582764, + "learning_rate": 0.00027, + "loss": 5.8555, + "step": 81 + }, + { + "epoch": 0.0080078125, + "grad_norm": 5.6479363441467285, + "learning_rate": 0.00027333333333333333, + "loss": 5.832, + "step": 82 + }, + { + "epoch": 0.00810546875, + "grad_norm": 3.7417171001434326, + "learning_rate": 0.00027666666666666665, + "loss": 5.8281, + "step": 83 + }, + { + "epoch": 0.008203125, + "grad_norm": 4.857771873474121, + "learning_rate": 0.00028000000000000003, + "loss": 5.8086, + "step": 84 + }, + { + "epoch": 0.00830078125, + "grad_norm": 4.6594085693359375, + "learning_rate": 0.00028333333333333335, + "loss": 5.832, + "step": 85 + }, + { + "epoch": 0.0083984375, + "grad_norm": 4.48290491104126, + "learning_rate": 0.0002866666666666667, + "loss": 5.8164, + "step": 86 + }, + { + "epoch": 0.00849609375, + "grad_norm": 6.183064937591553, + "learning_rate": 0.00029, + "loss": 5.8398, + "step": 87 + }, + { + "epoch": 0.00859375, + "grad_norm": 4.494201183319092, + "learning_rate": 0.0002933333333333333, + "loss": 5.7969, + "step": 88 + }, + { + "epoch": 0.00869140625, + "grad_norm": 5.650720596313477, + "learning_rate": 0.0002966666666666667, + "loss": 5.8438, + "step": 89 + }, + { + "epoch": 0.0087890625, + "grad_norm": 5.558387756347656, + "learning_rate": 0.0003, + "loss": 5.8555, + "step": 90 + }, + { + "epoch": 0.00888671875, + "grad_norm": 5.16208553314209, + "learning_rate": 0.00030333333333333335, + "loss": 5.8086, + "step": 91 + }, + { + "epoch": 0.008984375, + "grad_norm": 6.598042964935303, + "learning_rate": 0.0003066666666666667, + "loss": 5.8516, + "step": 92 + }, + { + "epoch": 0.00908203125, + "grad_norm": 4.591591835021973, + "learning_rate": 0.00031, + "loss": 5.8555, + "step": 93 + }, + { + "epoch": 0.0091796875, + "grad_norm": 4.879761695861816, + "learning_rate": 0.0003133333333333334, + "loss": 5.8203, + "step": 94 + }, + { + "epoch": 0.00927734375, + "grad_norm": 5.21987771987915, + "learning_rate": 0.00031666666666666665, + "loss": 5.8203, + "step": 95 + }, + { + "epoch": 0.009375, + "grad_norm": 4.257232666015625, + "learning_rate": 0.00032, + "loss": 5.8047, + "step": 96 + }, + { + "epoch": 0.00947265625, + "grad_norm": 6.07011604309082, + "learning_rate": 0.0003233333333333333, + "loss": 5.8359, + "step": 97 + }, + { + "epoch": 0.0095703125, + "grad_norm": 5.662813186645508, + "learning_rate": 0.0003266666666666667, + "loss": 5.8633, + "step": 98 + }, + { + "epoch": 0.00966796875, + "grad_norm": 4.59433650970459, + "learning_rate": 0.00033, + "loss": 5.8086, + "step": 99 + }, + { + "epoch": 0.009765625, + "grad_norm": 4.295780181884766, + "learning_rate": 0.0003333333333333333, + "loss": 5.8008, + "step": 100 + }, + { + "epoch": 0.00986328125, + "grad_norm": 4.587396144866943, + "learning_rate": 0.0003366666666666667, + "loss": 5.8008, + "step": 101 + }, + { + "epoch": 0.0099609375, + "grad_norm": 4.299502849578857, + "learning_rate": 0.00034, + "loss": 5.7773, + "step": 102 + }, + { + "epoch": 0.01005859375, + "grad_norm": 4.284260272979736, + "learning_rate": 0.00034333333333333335, + "loss": 5.793, + "step": 103 + }, + { + "epoch": 0.01015625, + "grad_norm": 6.048828601837158, + "learning_rate": 0.00034666666666666667, + "loss": 5.8008, + "step": 104 + }, + { + "epoch": 0.01025390625, + "grad_norm": 4.235161304473877, + "learning_rate": 0.00035, + "loss": 5.7656, + "step": 105 + }, + { + "epoch": 0.0103515625, + "grad_norm": 5.187899589538574, + "learning_rate": 0.0003533333333333333, + "loss": 5.8125, + "step": 106 + }, + { + "epoch": 0.01044921875, + "grad_norm": 4.133037567138672, + "learning_rate": 0.0003566666666666667, + "loss": 5.8047, + "step": 107 + }, + { + "epoch": 0.010546875, + "grad_norm": 5.822926044464111, + "learning_rate": 0.00035999999999999997, + "loss": 5.8281, + "step": 108 + }, + { + "epoch": 0.01064453125, + "grad_norm": 4.3357672691345215, + "learning_rate": 0.00036333333333333335, + "loss": 5.7734, + "step": 109 + }, + { + "epoch": 0.0107421875, + "grad_norm": 4.930606842041016, + "learning_rate": 0.00036666666666666667, + "loss": 5.7812, + "step": 110 + }, + { + "epoch": 0.01083984375, + "grad_norm": 4.797028064727783, + "learning_rate": 0.00037, + "loss": 5.7695, + "step": 111 + }, + { + "epoch": 0.0109375, + "grad_norm": 3.1973586082458496, + "learning_rate": 0.0003733333333333334, + "loss": 5.7539, + "step": 112 + }, + { + "epoch": 0.01103515625, + "grad_norm": 4.5399980545043945, + "learning_rate": 0.00037666666666666664, + "loss": 5.7852, + "step": 113 + }, + { + "epoch": 0.0111328125, + "grad_norm": 4.550619602203369, + "learning_rate": 0.00038, + "loss": 5.7773, + "step": 114 + }, + { + "epoch": 0.01123046875, + "grad_norm": 5.377904415130615, + "learning_rate": 0.00038333333333333334, + "loss": 5.7891, + "step": 115 + }, + { + "epoch": 0.011328125, + "grad_norm": 4.06483268737793, + "learning_rate": 0.00038666666666666667, + "loss": 5.6992, + "step": 116 + }, + { + "epoch": 0.01142578125, + "grad_norm": 3.8046791553497314, + "learning_rate": 0.00039000000000000005, + "loss": 5.7344, + "step": 117 + }, + { + "epoch": 0.0115234375, + "grad_norm": 4.709420204162598, + "learning_rate": 0.0003933333333333333, + "loss": 5.7461, + "step": 118 + }, + { + "epoch": 0.01162109375, + "grad_norm": 4.36158561706543, + "learning_rate": 0.0003966666666666667, + "loss": 5.75, + "step": 119 + }, + { + "epoch": 0.01171875, + "grad_norm": 4.972657680511475, + "learning_rate": 0.0004, + "loss": 5.7188, + "step": 120 + }, + { + "epoch": 0.01181640625, + "grad_norm": 3.862230062484741, + "learning_rate": 0.00040333333333333334, + "loss": 5.7617, + "step": 121 + }, + { + "epoch": 0.0119140625, + "grad_norm": 6.694273948669434, + "learning_rate": 0.00040666666666666667, + "loss": 5.7734, + "step": 122 + }, + { + "epoch": 0.01201171875, + "grad_norm": 4.289857864379883, + "learning_rate": 0.00041, + "loss": 5.7539, + "step": 123 + }, + { + "epoch": 0.012109375, + "grad_norm": 4.241764545440674, + "learning_rate": 0.0004133333333333333, + "loss": 5.7422, + "step": 124 + }, + { + "epoch": 0.01220703125, + "grad_norm": 5.350276947021484, + "learning_rate": 0.0004166666666666667, + "loss": 5.7578, + "step": 125 + }, + { + "epoch": 0.0123046875, + "grad_norm": 4.058553695678711, + "learning_rate": 0.00042, + "loss": 5.6992, + "step": 126 + }, + { + "epoch": 0.01240234375, + "grad_norm": 4.70885705947876, + "learning_rate": 0.00042333333333333334, + "loss": 5.7383, + "step": 127 + }, + { + "epoch": 0.0125, + "grad_norm": 4.190490245819092, + "learning_rate": 0.0004266666666666667, + "loss": 5.7305, + "step": 128 + }, + { + "epoch": 0.01259765625, + "grad_norm": 6.176610469818115, + "learning_rate": 0.00043, + "loss": 5.7344, + "step": 129 + }, + { + "epoch": 0.0126953125, + "grad_norm": 4.583580017089844, + "learning_rate": 0.00043333333333333337, + "loss": 5.7266, + "step": 130 + }, + { + "epoch": 0.01279296875, + "grad_norm": 3.8871593475341797, + "learning_rate": 0.00043666666666666664, + "loss": 5.7422, + "step": 131 + }, + { + "epoch": 0.012890625, + "grad_norm": 6.952763557434082, + "learning_rate": 0.00044, + "loss": 5.7812, + "step": 132 + }, + { + "epoch": 0.01298828125, + "grad_norm": 4.236309051513672, + "learning_rate": 0.00044333333333333334, + "loss": 5.75, + "step": 133 + }, + { + "epoch": 0.0130859375, + "grad_norm": 7.308345794677734, + "learning_rate": 0.00044666666666666666, + "loss": 5.8125, + "step": 134 + }, + { + "epoch": 0.01318359375, + "grad_norm": 8.599268913269043, + "learning_rate": 0.00045000000000000004, + "loss": 5.918, + "step": 135 + }, + { + "epoch": 0.01328125, + "grad_norm": 4.937841415405273, + "learning_rate": 0.0004533333333333333, + "loss": 5.8242, + "step": 136 + }, + { + "epoch": 0.01337890625, + "grad_norm": 4.781856536865234, + "learning_rate": 0.0004566666666666667, + "loss": 5.7891, + "step": 137 + }, + { + "epoch": 0.0134765625, + "grad_norm": 7.261330604553223, + "learning_rate": 0.00046, + "loss": 5.8203, + "step": 138 + }, + { + "epoch": 0.01357421875, + "grad_norm": 5.891605854034424, + "learning_rate": 0.00046333333333333334, + "loss": 5.875, + "step": 139 + }, + { + "epoch": 0.013671875, + "grad_norm": 7.775816440582275, + "learning_rate": 0.00046666666666666666, + "loss": 5.8672, + "step": 140 + }, + { + "epoch": 0.01376953125, + "grad_norm": 3.8871777057647705, + "learning_rate": 0.00047, + "loss": 5.8359, + "step": 141 + }, + { + "epoch": 0.0138671875, + "grad_norm": 5.772259712219238, + "learning_rate": 0.00047333333333333336, + "loss": 5.8477, + "step": 142 + }, + { + "epoch": 0.01396484375, + "grad_norm": 4.485783100128174, + "learning_rate": 0.0004766666666666667, + "loss": 5.793, + "step": 143 + }, + { + "epoch": 0.0140625, + "grad_norm": 4.418201446533203, + "learning_rate": 0.00048, + "loss": 5.7812, + "step": 144 + }, + { + "epoch": 0.01416015625, + "grad_norm": 7.3408074378967285, + "learning_rate": 0.00048333333333333334, + "loss": 5.8516, + "step": 145 + }, + { + "epoch": 0.0142578125, + "grad_norm": 5.989964962005615, + "learning_rate": 0.0004866666666666667, + "loss": 5.8945, + "step": 146 + }, + { + "epoch": 0.01435546875, + "grad_norm": 4.477835655212402, + "learning_rate": 0.00049, + "loss": 5.8477, + "step": 147 + }, + { + "epoch": 0.014453125, + "grad_norm": 6.493783950805664, + "learning_rate": 0.0004933333333333334, + "loss": 5.8594, + "step": 148 + }, + { + "epoch": 0.01455078125, + "grad_norm": 6.190314292907715, + "learning_rate": 0.0004966666666666666, + "loss": 5.8906, + "step": 149 + }, + { + "epoch": 0.0146484375, + "grad_norm": 6.708803653717041, + "learning_rate": 0.0005, + "loss": 5.8711, + "step": 150 + }, + { + "epoch": 0.01474609375, + "grad_norm": 5.7710113525390625, + "learning_rate": 0.0004999999890938886, + "loss": 5.8633, + "step": 151 + }, + { + "epoch": 0.01484375, + "grad_norm": 5.553577899932861, + "learning_rate": 0.0004999999563755552, + "loss": 5.8594, + "step": 152 + }, + { + "epoch": 0.01494140625, + "grad_norm": 4.852464199066162, + "learning_rate": 0.0004999999018450032, + "loss": 5.8555, + "step": 153 + }, + { + "epoch": 0.0150390625, + "grad_norm": 4.127274990081787, + "learning_rate": 0.0004999998255022377, + "loss": 5.793, + "step": 154 + }, + { + "epoch": 0.01513671875, + "grad_norm": 5.139339923858643, + "learning_rate": 0.0004999997273472664, + "loss": 5.8398, + "step": 155 + }, + { + "epoch": 0.015234375, + "grad_norm": 3.9165873527526855, + "learning_rate": 0.0004999996073800985, + "loss": 5.7852, + "step": 156 + }, + { + "epoch": 0.01533203125, + "grad_norm": 4.544178485870361, + "learning_rate": 0.0004999994656007457, + "loss": 5.8125, + "step": 157 + }, + { + "epoch": 0.0154296875, + "grad_norm": 4.998808860778809, + "learning_rate": 0.0004999993020092219, + "loss": 5.8359, + "step": 158 + }, + { + "epoch": 0.01552734375, + "grad_norm": 5.184920787811279, + "learning_rate": 0.0004999991166055426, + "loss": 5.8281, + "step": 159 + }, + { + "epoch": 0.015625, + "grad_norm": 3.716250419616699, + "learning_rate": 0.0004999989093897262, + "loss": 5.7969, + "step": 160 + }, + { + "epoch": 0.01572265625, + "grad_norm": 3.8317952156066895, + "learning_rate": 0.0004999986803617926, + "loss": 5.7617, + "step": 161 + }, + { + "epoch": 0.0158203125, + "grad_norm": 4.461795330047607, + "learning_rate": 0.0004999984295217641, + "loss": 5.8047, + "step": 162 + }, + { + "epoch": 0.01591796875, + "grad_norm": 3.0825703144073486, + "learning_rate": 0.0004999981568696648, + "loss": 5.7656, + "step": 163 + }, + { + "epoch": 0.016015625, + "grad_norm": 4.151459217071533, + "learning_rate": 0.0004999978624055212, + "loss": 5.7617, + "step": 164 + }, + { + "epoch": 0.01611328125, + "grad_norm": 3.66987681388855, + "learning_rate": 0.0004999975461293621, + "loss": 5.7461, + "step": 165 + }, + { + "epoch": 0.0162109375, + "grad_norm": 4.092290878295898, + "learning_rate": 0.0004999972080412177, + "loss": 5.75, + "step": 166 + }, + { + "epoch": 0.01630859375, + "grad_norm": 4.104146480560303, + "learning_rate": 0.0004999968481411212, + "loss": 5.7656, + "step": 167 + }, + { + "epoch": 0.01640625, + "grad_norm": 4.437919616699219, + "learning_rate": 0.0004999964664291073, + "loss": 5.7812, + "step": 168 + }, + { + "epoch": 0.01650390625, + "grad_norm": 7.200653553009033, + "learning_rate": 0.0004999960629052131, + "loss": 5.7578, + "step": 169 + }, + { + "epoch": 0.0166015625, + "grad_norm": 2.731268882751465, + "learning_rate": 0.0004999956375694776, + "loss": 5.707, + "step": 170 + }, + { + "epoch": 0.01669921875, + "grad_norm": 5.332357406616211, + "learning_rate": 0.0004999951904219421, + "loss": 5.7461, + "step": 171 + }, + { + "epoch": 0.016796875, + "grad_norm": 3.130514144897461, + "learning_rate": 0.0004999947214626501, + "loss": 5.7109, + "step": 172 + }, + { + "epoch": 0.01689453125, + "grad_norm": 3.6535258293151855, + "learning_rate": 0.0004999942306916466, + "loss": 5.7344, + "step": 173 + }, + { + "epoch": 0.0169921875, + "grad_norm": 3.5438027381896973, + "learning_rate": 0.0004999937181089796, + "loss": 5.6953, + "step": 174 + }, + { + "epoch": 0.01708984375, + "grad_norm": 4.228607177734375, + "learning_rate": 0.0004999931837146987, + "loss": 5.7031, + "step": 175 + }, + { + "epoch": 0.0171875, + "grad_norm": 3.217113971710205, + "learning_rate": 0.0004999926275088556, + "loss": 5.7148, + "step": 176 + }, + { + "epoch": 0.01728515625, + "grad_norm": 4.9072041511535645, + "learning_rate": 0.0004999920494915043, + "loss": 5.75, + "step": 177 + }, + { + "epoch": 0.0173828125, + "grad_norm": 2.8025128841400146, + "learning_rate": 0.0004999914496627009, + "loss": 5.7148, + "step": 178 + }, + { + "epoch": 0.01748046875, + "grad_norm": 2.481431245803833, + "learning_rate": 0.0004999908280225035, + "loss": 5.6445, + "step": 179 + }, + { + "epoch": 0.017578125, + "grad_norm": 3.0694146156311035, + "learning_rate": 0.0004999901845709722, + "loss": 5.6602, + "step": 180 + }, + { + "epoch": 0.01767578125, + "grad_norm": 2.6255242824554443, + "learning_rate": 0.0004999895193081698, + "loss": 5.6523, + "step": 181 + }, + { + "epoch": 0.0177734375, + "grad_norm": 3.0787289142608643, + "learning_rate": 0.0004999888322341602, + "loss": 5.6367, + "step": 182 + }, + { + "epoch": 0.01787109375, + "grad_norm": 3.5675113201141357, + "learning_rate": 0.0004999881233490104, + "loss": 5.6875, + "step": 183 + }, + { + "epoch": 0.01796875, + "grad_norm": 3.278024673461914, + "learning_rate": 0.0004999873926527891, + "loss": 5.6406, + "step": 184 + }, + { + "epoch": 0.01806640625, + "grad_norm": 3.4470205307006836, + "learning_rate": 0.0004999866401455671, + "loss": 5.6328, + "step": 185 + }, + { + "epoch": 0.0181640625, + "grad_norm": 2.537705659866333, + "learning_rate": 0.0004999858658274172, + "loss": 5.625, + "step": 186 + }, + { + "epoch": 0.01826171875, + "grad_norm": 3.446849822998047, + "learning_rate": 0.0004999850696984147, + "loss": 5.6367, + "step": 187 + }, + { + "epoch": 0.018359375, + "grad_norm": 2.9557571411132812, + "learning_rate": 0.0004999842517586367, + "loss": 5.6523, + "step": 188 + }, + { + "epoch": 0.01845703125, + "grad_norm": 2.224954605102539, + "learning_rate": 0.0004999834120081624, + "loss": 5.6133, + "step": 189 + }, + { + "epoch": 0.0185546875, + "grad_norm": 2.847534418106079, + "learning_rate": 0.0004999825504470732, + "loss": 5.5977, + "step": 190 + }, + { + "epoch": 0.01865234375, + "grad_norm": 3.08176851272583, + "learning_rate": 0.0004999816670754527, + "loss": 5.6094, + "step": 191 + }, + { + "epoch": 0.01875, + "grad_norm": 2.509631872177124, + "learning_rate": 0.0004999807618933866, + "loss": 5.582, + "step": 192 + }, + { + "epoch": 0.01884765625, + "grad_norm": 2.551470994949341, + "learning_rate": 0.0004999798349009626, + "loss": 5.582, + "step": 193 + }, + { + "epoch": 0.0189453125, + "grad_norm": 2.8080625534057617, + "learning_rate": 0.0004999788860982706, + "loss": 5.5898, + "step": 194 + }, + { + "epoch": 0.01904296875, + "grad_norm": 2.784714698791504, + "learning_rate": 0.0004999779154854024, + "loss": 5.6016, + "step": 195 + }, + { + "epoch": 0.019140625, + "grad_norm": 2.6262359619140625, + "learning_rate": 0.0004999769230624524, + "loss": 5.5664, + "step": 196 + }, + { + "epoch": 0.01923828125, + "grad_norm": 2.2458934783935547, + "learning_rate": 0.0004999759088295165, + "loss": 5.5508, + "step": 197 + }, + { + "epoch": 0.0193359375, + "grad_norm": 2.5689525604248047, + "learning_rate": 0.0004999748727866932, + "loss": 5.5625, + "step": 198 + }, + { + "epoch": 0.01943359375, + "grad_norm": 2.44174861907959, + "learning_rate": 0.0004999738149340828, + "loss": 5.5352, + "step": 199 + }, + { + "epoch": 0.01953125, + "grad_norm": 2.5862767696380615, + "learning_rate": 0.000499972735271788, + "loss": 5.5234, + "step": 200 + }, + { + "epoch": 0.01962890625, + "grad_norm": 2.2569644451141357, + "learning_rate": 0.0004999716337999135, + "loss": 5.5078, + "step": 201 + }, + { + "epoch": 0.0197265625, + "grad_norm": 2.5576300621032715, + "learning_rate": 0.0004999705105185659, + "loss": 5.5391, + "step": 202 + }, + { + "epoch": 0.01982421875, + "grad_norm": 2.7671685218811035, + "learning_rate": 0.0004999693654278542, + "loss": 5.5234, + "step": 203 + }, + { + "epoch": 0.019921875, + "grad_norm": 1.9716945886611938, + "learning_rate": 0.0004999681985278894, + "loss": 5.4961, + "step": 204 + }, + { + "epoch": 0.02001953125, + "grad_norm": 2.345658302307129, + "learning_rate": 0.0004999670098187846, + "loss": 5.4961, + "step": 205 + }, + { + "epoch": 0.0201171875, + "grad_norm": 2.1516318321228027, + "learning_rate": 0.0004999657993006551, + "loss": 5.5078, + "step": 206 + }, + { + "epoch": 0.02021484375, + "grad_norm": 2.983196258544922, + "learning_rate": 0.0004999645669736181, + "loss": 5.5273, + "step": 207 + }, + { + "epoch": 0.0203125, + "grad_norm": 2.5061614513397217, + "learning_rate": 0.0004999633128377932, + "loss": 5.5195, + "step": 208 + }, + { + "epoch": 0.02041015625, + "grad_norm": 1.7015206813812256, + "learning_rate": 0.000499962036893302, + "loss": 5.4961, + "step": 209 + }, + { + "epoch": 0.0205078125, + "grad_norm": 2.2488410472869873, + "learning_rate": 0.0004999607391402681, + "loss": 5.4258, + "step": 210 + }, + { + "epoch": 0.02060546875, + "grad_norm": 2.7714498043060303, + "learning_rate": 0.0004999594195788175, + "loss": 5.5039, + "step": 211 + }, + { + "epoch": 0.020703125, + "grad_norm": 2.1639745235443115, + "learning_rate": 0.0004999580782090778, + "loss": 5.4922, + "step": 212 + }, + { + "epoch": 0.02080078125, + "grad_norm": 1.9490896463394165, + "learning_rate": 0.0004999567150311793, + "loss": 5.4492, + "step": 213 + }, + { + "epoch": 0.0208984375, + "grad_norm": 2.291276454925537, + "learning_rate": 0.0004999553300452541, + "loss": 5.4648, + "step": 214 + }, + { + "epoch": 0.02099609375, + "grad_norm": 2.4977705478668213, + "learning_rate": 0.0004999539232514363, + "loss": 5.4688, + "step": 215 + }, + { + "epoch": 0.02109375, + "grad_norm": 2.013157367706299, + "learning_rate": 0.0004999524946498624, + "loss": 5.4727, + "step": 216 + }, + { + "epoch": 0.02119140625, + "grad_norm": 2.35416579246521, + "learning_rate": 0.000499951044240671, + "loss": 5.457, + "step": 217 + }, + { + "epoch": 0.0212890625, + "grad_norm": 1.7172154188156128, + "learning_rate": 0.0004999495720240027, + "loss": 5.4531, + "step": 218 + }, + { + "epoch": 0.02138671875, + "grad_norm": 2.345547676086426, + "learning_rate": 0.000499948078, + "loss": 5.4609, + "step": 219 + }, + { + "epoch": 0.021484375, + "grad_norm": 2.2435221672058105, + "learning_rate": 0.0004999465621688078, + "loss": 5.4453, + "step": 220 + }, + { + "epoch": 0.02158203125, + "grad_norm": 2.306879997253418, + "learning_rate": 0.0004999450245305732, + "loss": 5.4414, + "step": 221 + }, + { + "epoch": 0.0216796875, + "grad_norm": 2.2427804470062256, + "learning_rate": 0.0004999434650854452, + "loss": 5.4531, + "step": 222 + }, + { + "epoch": 0.02177734375, + "grad_norm": 1.790202260017395, + "learning_rate": 0.000499941883833575, + "loss": 5.457, + "step": 223 + }, + { + "epoch": 0.021875, + "grad_norm": 1.838149070739746, + "learning_rate": 0.0004999402807751157, + "loss": 5.4297, + "step": 224 + }, + { + "epoch": 0.02197265625, + "grad_norm": 2.016653299331665, + "learning_rate": 0.000499938655910223, + "loss": 5.4141, + "step": 225 + }, + { + "epoch": 0.0220703125, + "grad_norm": 2.0161890983581543, + "learning_rate": 0.0004999370092390541, + "loss": 5.3906, + "step": 226 + }, + { + "epoch": 0.02216796875, + "grad_norm": 2.036649465560913, + "learning_rate": 0.0004999353407617689, + "loss": 5.4297, + "step": 227 + }, + { + "epoch": 0.022265625, + "grad_norm": 1.5296927690505981, + "learning_rate": 0.000499933650478529, + "loss": 5.4023, + "step": 228 + }, + { + "epoch": 0.02236328125, + "grad_norm": 1.713457703590393, + "learning_rate": 0.0004999319383894985, + "loss": 5.4219, + "step": 229 + }, + { + "epoch": 0.0224609375, + "grad_norm": 2.3026342391967773, + "learning_rate": 0.000499930204494843, + "loss": 5.4453, + "step": 230 + }, + { + "epoch": 0.02255859375, + "grad_norm": 2.2491424083709717, + "learning_rate": 0.0004999284487947308, + "loss": 5.4219, + "step": 231 + }, + { + "epoch": 0.02265625, + "grad_norm": 2.2838447093963623, + "learning_rate": 0.0004999266712893322, + "loss": 5.4336, + "step": 232 + }, + { + "epoch": 0.02275390625, + "grad_norm": 1.7964040040969849, + "learning_rate": 0.0004999248719788193, + "loss": 5.4141, + "step": 233 + }, + { + "epoch": 0.0228515625, + "grad_norm": 2.0387890338897705, + "learning_rate": 0.0004999230508633667, + "loss": 5.4102, + "step": 234 + }, + { + "epoch": 0.02294921875, + "grad_norm": 2.3577651977539062, + "learning_rate": 0.0004999212079431507, + "loss": 5.4297, + "step": 235 + }, + { + "epoch": 0.023046875, + "grad_norm": 1.8143409490585327, + "learning_rate": 0.0004999193432183504, + "loss": 5.4023, + "step": 236 + }, + { + "epoch": 0.02314453125, + "grad_norm": 1.6521782875061035, + "learning_rate": 0.0004999174566891461, + "loss": 5.3906, + "step": 237 + }, + { + "epoch": 0.0232421875, + "grad_norm": 1.8481347560882568, + "learning_rate": 0.000499915548355721, + "loss": 5.3984, + "step": 238 + }, + { + "epoch": 0.02333984375, + "grad_norm": 2.1952035427093506, + "learning_rate": 0.0004999136182182601, + "loss": 5.3984, + "step": 239 + }, + { + "epoch": 0.0234375, + "grad_norm": 2.0808660984039307, + "learning_rate": 0.0004999116662769502, + "loss": 5.3945, + "step": 240 + }, + { + "epoch": 0.02353515625, + "grad_norm": 2.293430805206299, + "learning_rate": 0.0004999096925319808, + "loss": 5.4102, + "step": 241 + }, + { + "epoch": 0.0236328125, + "grad_norm": 1.6741044521331787, + "learning_rate": 0.0004999076969835432, + "loss": 5.4062, + "step": 242 + }, + { + "epoch": 0.02373046875, + "grad_norm": 1.95404851436615, + "learning_rate": 0.0004999056796318308, + "loss": 5.3672, + "step": 243 + }, + { + "epoch": 0.023828125, + "grad_norm": 1.8848334550857544, + "learning_rate": 0.0004999036404770391, + "loss": 5.375, + "step": 244 + }, + { + "epoch": 0.02392578125, + "grad_norm": 1.6347148418426514, + "learning_rate": 0.0004999015795193661, + "loss": 5.3711, + "step": 245 + }, + { + "epoch": 0.0240234375, + "grad_norm": 2.3380987644195557, + "learning_rate": 0.0004998994967590113, + "loss": 5.4023, + "step": 246 + }, + { + "epoch": 0.02412109375, + "grad_norm": 1.4733819961547852, + "learning_rate": 0.0004998973921961767, + "loss": 5.3711, + "step": 247 + }, + { + "epoch": 0.02421875, + "grad_norm": 2.443174362182617, + "learning_rate": 0.0004998952658310662, + "loss": 5.3633, + "step": 248 + }, + { + "epoch": 0.02431640625, + "grad_norm": 1.9105093479156494, + "learning_rate": 0.0004998931176638861, + "loss": 5.375, + "step": 249 + }, + { + "epoch": 0.0244140625, + "grad_norm": 2.14117431640625, + "learning_rate": 0.0004998909476948446, + "loss": 5.3672, + "step": 250 + }, + { + "epoch": 0.02451171875, + "grad_norm": 1.5175701379776, + "learning_rate": 0.0004998887559241521, + "loss": 5.3711, + "step": 251 + }, + { + "epoch": 0.024609375, + "grad_norm": 1.7954213619232178, + "learning_rate": 0.000499886542352021, + "loss": 5.3359, + "step": 252 + }, + { + "epoch": 0.02470703125, + "grad_norm": 2.0219264030456543, + "learning_rate": 0.0004998843069786659, + "loss": 5.3516, + "step": 253 + }, + { + "epoch": 0.0248046875, + "grad_norm": 2.42618465423584, + "learning_rate": 0.0004998820498043036, + "loss": 5.3594, + "step": 254 + }, + { + "epoch": 0.02490234375, + "grad_norm": 2.4111266136169434, + "learning_rate": 0.0004998797708291528, + "loss": 5.3477, + "step": 255 + }, + { + "epoch": 0.025, + "grad_norm": 1.6236835718154907, + "learning_rate": 0.0004998774700534346, + "loss": 5.3516, + "step": 256 + }, + { + "epoch": 0.02509765625, + "grad_norm": 2.249431610107422, + "learning_rate": 0.0004998751474773718, + "loss": 5.3789, + "step": 257 + }, + { + "epoch": 0.0251953125, + "grad_norm": 1.878957986831665, + "learning_rate": 0.0004998728031011897, + "loss": 5.3359, + "step": 258 + }, + { + "epoch": 0.02529296875, + "grad_norm": 1.7500239610671997, + "learning_rate": 0.0004998704369251156, + "loss": 5.3086, + "step": 259 + }, + { + "epoch": 0.025390625, + "grad_norm": 1.5108247995376587, + "learning_rate": 0.0004998680489493788, + "loss": 5.3008, + "step": 260 + }, + { + "epoch": 0.02548828125, + "grad_norm": 1.8362010717391968, + "learning_rate": 0.0004998656391742108, + "loss": 5.3398, + "step": 261 + }, + { + "epoch": 0.0255859375, + "grad_norm": 1.7218053340911865, + "learning_rate": 0.0004998632075998453, + "loss": 5.3633, + "step": 262 + }, + { + "epoch": 0.02568359375, + "grad_norm": 1.8602724075317383, + "learning_rate": 0.000499860754226518, + "loss": 5.3477, + "step": 263 + }, + { + "epoch": 0.02578125, + "grad_norm": 1.3672821521759033, + "learning_rate": 0.0004998582790544667, + "loss": 5.3047, + "step": 264 + }, + { + "epoch": 0.02587890625, + "grad_norm": 1.5733098983764648, + "learning_rate": 0.0004998557820839313, + "loss": 5.3203, + "step": 265 + }, + { + "epoch": 0.0259765625, + "grad_norm": 1.547677993774414, + "learning_rate": 0.0004998532633151539, + "loss": 5.3203, + "step": 266 + }, + { + "epoch": 0.02607421875, + "grad_norm": 1.8635348081588745, + "learning_rate": 0.0004998507227483789, + "loss": 5.332, + "step": 267 + }, + { + "epoch": 0.026171875, + "grad_norm": 1.7884019613265991, + "learning_rate": 0.0004998481603838523, + "loss": 5.3281, + "step": 268 + }, + { + "epoch": 0.02626953125, + "grad_norm": 1.6009703874588013, + "learning_rate": 0.0004998455762218225, + "loss": 5.3125, + "step": 269 + }, + { + "epoch": 0.0263671875, + "grad_norm": 1.79560387134552, + "learning_rate": 0.0004998429702625401, + "loss": 5.3125, + "step": 270 + }, + { + "epoch": 0.02646484375, + "grad_norm": 2.2909066677093506, + "learning_rate": 0.0004998403425062579, + "loss": 5.3242, + "step": 271 + }, + { + "epoch": 0.0265625, + "grad_norm": 1.209547758102417, + "learning_rate": 0.0004998376929532305, + "loss": 5.3164, + "step": 272 + }, + { + "epoch": 0.02666015625, + "grad_norm": 1.811440110206604, + "learning_rate": 0.0004998350216037146, + "loss": 5.3047, + "step": 273 + }, + { + "epoch": 0.0267578125, + "grad_norm": 1.956322193145752, + "learning_rate": 0.0004998323284579694, + "loss": 5.3086, + "step": 274 + }, + { + "epoch": 0.02685546875, + "grad_norm": 2.4985668659210205, + "learning_rate": 0.000499829613516256, + "loss": 5.3047, + "step": 275 + }, + { + "epoch": 0.026953125, + "grad_norm": 2.2196872234344482, + "learning_rate": 0.0004998268767788373, + "loss": 5.3125, + "step": 276 + }, + { + "epoch": 0.02705078125, + "grad_norm": 2.2135283946990967, + "learning_rate": 0.0004998241182459789, + "loss": 5.3242, + "step": 277 + }, + { + "epoch": 0.0271484375, + "grad_norm": 1.4025847911834717, + "learning_rate": 0.0004998213379179481, + "loss": 5.3398, + "step": 278 + }, + { + "epoch": 0.02724609375, + "grad_norm": 2.832472801208496, + "learning_rate": 0.0004998185357950144, + "loss": 5.3047, + "step": 279 + }, + { + "epoch": 0.02734375, + "grad_norm": 1.4372154474258423, + "learning_rate": 0.0004998157118774496, + "loss": 5.3047, + "step": 280 + }, + { + "epoch": 0.02744140625, + "grad_norm": 2.209719657897949, + "learning_rate": 0.0004998128661655274, + "loss": 5.3008, + "step": 281 + }, + { + "epoch": 0.0275390625, + "grad_norm": 2.680424928665161, + "learning_rate": 0.0004998099986595235, + "loss": 5.3047, + "step": 282 + }, + { + "epoch": 0.02763671875, + "grad_norm": 2.1611225605010986, + "learning_rate": 0.0004998071093597162, + "loss": 5.3008, + "step": 283 + }, + { + "epoch": 0.027734375, + "grad_norm": 1.9571424722671509, + "learning_rate": 0.0004998041982663851, + "loss": 5.2891, + "step": 284 + }, + { + "epoch": 0.02783203125, + "grad_norm": 1.3447707891464233, + "learning_rate": 0.000499801265379813, + "loss": 5.2852, + "step": 285 + }, + { + "epoch": 0.0279296875, + "grad_norm": 2.0131936073303223, + "learning_rate": 0.0004997983107002838, + "loss": 5.3086, + "step": 286 + }, + { + "epoch": 0.02802734375, + "grad_norm": 1.1873258352279663, + "learning_rate": 0.0004997953342280843, + "loss": 5.2656, + "step": 287 + }, + { + "epoch": 0.028125, + "grad_norm": 2.239952564239502, + "learning_rate": 0.0004997923359635027, + "loss": 5.2773, + "step": 288 + }, + { + "epoch": 0.02822265625, + "grad_norm": 1.3604589700698853, + "learning_rate": 0.0004997893159068297, + "loss": 5.2773, + "step": 289 + }, + { + "epoch": 0.0283203125, + "grad_norm": 1.888407826423645, + "learning_rate": 0.0004997862740583584, + "loss": 5.2852, + "step": 290 + }, + { + "epoch": 0.02841796875, + "grad_norm": 1.8944205045700073, + "learning_rate": 0.0004997832104183833, + "loss": 5.2812, + "step": 291 + }, + { + "epoch": 0.028515625, + "grad_norm": 1.1297378540039062, + "learning_rate": 0.0004997801249872016, + "loss": 5.25, + "step": 292 + }, + { + "epoch": 0.02861328125, + "grad_norm": 1.6378464698791504, + "learning_rate": 0.0004997770177651123, + "loss": 5.2461, + "step": 293 + }, + { + "epoch": 0.0287109375, + "grad_norm": 1.3006982803344727, + "learning_rate": 0.0004997738887524168, + "loss": 5.2578, + "step": 294 + }, + { + "epoch": 0.02880859375, + "grad_norm": 1.8940247297286987, + "learning_rate": 0.0004997707379494183, + "loss": 5.2773, + "step": 295 + }, + { + "epoch": 0.02890625, + "grad_norm": 1.5350399017333984, + "learning_rate": 0.0004997675653564223, + "loss": 5.2656, + "step": 296 + }, + { + "epoch": 0.02900390625, + "grad_norm": 1.58504056930542, + "learning_rate": 0.0004997643709737363, + "loss": 5.2695, + "step": 297 + }, + { + "epoch": 0.0291015625, + "grad_norm": 1.5148996114730835, + "learning_rate": 0.0004997611548016699, + "loss": 5.2734, + "step": 298 + }, + { + "epoch": 0.02919921875, + "grad_norm": 1.2484545707702637, + "learning_rate": 0.0004997579168405352, + "loss": 5.25, + "step": 299 + }, + { + "epoch": 0.029296875, + "grad_norm": 1.4745291471481323, + "learning_rate": 0.0004997546570906458, + "loss": 5.2773, + "step": 300 + }, + { + "epoch": 0.02939453125, + "grad_norm": 1.0975582599639893, + "learning_rate": 0.0004997513755523179, + "loss": 5.2344, + "step": 301 + }, + { + "epoch": 0.0294921875, + "grad_norm": 1.4826176166534424, + "learning_rate": 0.0004997480722258694, + "loss": 5.2617, + "step": 302 + }, + { + "epoch": 0.02958984375, + "grad_norm": 1.3512847423553467, + "learning_rate": 0.0004997447471116207, + "loss": 5.2422, + "step": 303 + }, + { + "epoch": 0.0296875, + "grad_norm": 1.2088016271591187, + "learning_rate": 0.0004997414002098941, + "loss": 5.2539, + "step": 304 + }, + { + "epoch": 0.02978515625, + "grad_norm": 1.3594310283660889, + "learning_rate": 0.0004997380315210142, + "loss": 5.2695, + "step": 305 + }, + { + "epoch": 0.0298828125, + "grad_norm": 1.6119142770767212, + "learning_rate": 0.0004997346410453074, + "loss": 5.2383, + "step": 306 + }, + { + "epoch": 0.02998046875, + "grad_norm": 1.2976305484771729, + "learning_rate": 0.0004997312287831024, + "loss": 5.2773, + "step": 307 + }, + { + "epoch": 0.030078125, + "grad_norm": 1.1194605827331543, + "learning_rate": 0.00049972779473473, + "loss": 5.2656, + "step": 308 + }, + { + "epoch": 0.03017578125, + "grad_norm": 1.244971752166748, + "learning_rate": 0.0004997243389005232, + "loss": 5.2383, + "step": 309 + }, + { + "epoch": 0.0302734375, + "grad_norm": 1.1144415140151978, + "learning_rate": 0.0004997208612808168, + "loss": 5.2188, + "step": 310 + }, + { + "epoch": 0.03037109375, + "grad_norm": 1.1849431991577148, + "learning_rate": 0.0004997173618759482, + "loss": 5.2422, + "step": 311 + }, + { + "epoch": 0.03046875, + "grad_norm": 1.3395856618881226, + "learning_rate": 0.0004997138406862565, + "loss": 5.2422, + "step": 312 + }, + { + "epoch": 0.03056640625, + "grad_norm": 1.1774636507034302, + "learning_rate": 0.0004997102977120832, + "loss": 5.2188, + "step": 313 + }, + { + "epoch": 0.0306640625, + "grad_norm": 1.4251941442489624, + "learning_rate": 0.0004997067329537714, + "loss": 5.2578, + "step": 314 + }, + { + "epoch": 0.03076171875, + "grad_norm": 1.4117344617843628, + "learning_rate": 0.0004997031464116672, + "loss": 5.207, + "step": 315 + }, + { + "epoch": 0.030859375, + "grad_norm": 1.3700896501541138, + "learning_rate": 0.0004996995380861178, + "loss": 5.2305, + "step": 316 + }, + { + "epoch": 0.03095703125, + "grad_norm": 1.682862401008606, + "learning_rate": 0.0004996959079774734, + "loss": 5.25, + "step": 317 + }, + { + "epoch": 0.0310546875, + "grad_norm": 1.1571077108383179, + "learning_rate": 0.0004996922560860856, + "loss": 5.2109, + "step": 318 + }, + { + "epoch": 0.03115234375, + "grad_norm": 1.1706202030181885, + "learning_rate": 0.0004996885824123087, + "loss": 5.2344, + "step": 319 + }, + { + "epoch": 0.03125, + "grad_norm": 1.116774082183838, + "learning_rate": 0.0004996848869564986, + "loss": 5.2227, + "step": 320 + }, + { + "epoch": 0.03134765625, + "grad_norm": 1.3174951076507568, + "learning_rate": 0.0004996811697190137, + "loss": 5.207, + "step": 321 + }, + { + "epoch": 0.0314453125, + "grad_norm": 1.7159594297409058, + "learning_rate": 0.0004996774307002142, + "loss": 5.2305, + "step": 322 + }, + { + "epoch": 0.03154296875, + "grad_norm": 1.1668405532836914, + "learning_rate": 0.0004996736699004628, + "loss": 5.2344, + "step": 323 + }, + { + "epoch": 0.031640625, + "grad_norm": 1.4254777431488037, + "learning_rate": 0.000499669887320124, + "loss": 5.2188, + "step": 324 + }, + { + "epoch": 0.03173828125, + "grad_norm": 1.4150874614715576, + "learning_rate": 0.0004996660829595643, + "loss": 5.2461, + "step": 325 + }, + { + "epoch": 0.0318359375, + "grad_norm": 1.4464291334152222, + "learning_rate": 0.0004996622568191529, + "loss": 5.207, + "step": 326 + }, + { + "epoch": 0.03193359375, + "grad_norm": 1.3123974800109863, + "learning_rate": 0.0004996584088992603, + "loss": 5.2227, + "step": 327 + }, + { + "epoch": 0.03203125, + "grad_norm": 1.2821930646896362, + "learning_rate": 0.0004996545392002597, + "loss": 5.2344, + "step": 328 + }, + { + "epoch": 0.03212890625, + "grad_norm": 1.6003687381744385, + "learning_rate": 0.0004996506477225264, + "loss": 5.1836, + "step": 329 + }, + { + "epoch": 0.0322265625, + "grad_norm": 1.305853009223938, + "learning_rate": 0.0004996467344664374, + "loss": 5.2266, + "step": 330 + }, + { + "epoch": 0.03232421875, + "grad_norm": 1.6081973314285278, + "learning_rate": 0.0004996427994323723, + "loss": 5.2031, + "step": 331 + }, + { + "epoch": 0.032421875, + "grad_norm": 1.0995270013809204, + "learning_rate": 0.0004996388426207123, + "loss": 5.1797, + "step": 332 + }, + { + "epoch": 0.03251953125, + "grad_norm": 1.1569955348968506, + "learning_rate": 0.0004996348640318412, + "loss": 5.2305, + "step": 333 + }, + { + "epoch": 0.0326171875, + "grad_norm": 1.3775110244750977, + "learning_rate": 0.0004996308636661447, + "loss": 5.2266, + "step": 334 + }, + { + "epoch": 0.03271484375, + "grad_norm": 1.2185978889465332, + "learning_rate": 0.0004996268415240104, + "loss": 5.1641, + "step": 335 + }, + { + "epoch": 0.0328125, + "grad_norm": 1.349692702293396, + "learning_rate": 0.0004996227976058285, + "loss": 5.1914, + "step": 336 + }, + { + "epoch": 0.03291015625, + "grad_norm": 0.9755382537841797, + "learning_rate": 0.0004996187319119908, + "loss": 5.1914, + "step": 337 + }, + { + "epoch": 0.0330078125, + "grad_norm": 1.187360405921936, + "learning_rate": 0.0004996146444428916, + "loss": 5.1875, + "step": 338 + }, + { + "epoch": 0.03310546875, + "grad_norm": 1.2067431211471558, + "learning_rate": 0.000499610535198927, + "loss": 5.2109, + "step": 339 + }, + { + "epoch": 0.033203125, + "grad_norm": 1.2557358741760254, + "learning_rate": 0.0004996064041804956, + "loss": 5.1953, + "step": 340 + }, + { + "epoch": 0.03330078125, + "grad_norm": 1.191425085067749, + "learning_rate": 0.0004996022513879977, + "loss": 5.1719, + "step": 341 + }, + { + "epoch": 0.0333984375, + "grad_norm": 0.985287606716156, + "learning_rate": 0.0004995980768218358, + "loss": 5.1719, + "step": 342 + }, + { + "epoch": 0.03349609375, + "grad_norm": 2.132200002670288, + "learning_rate": 0.0004995938804824146, + "loss": 5.2109, + "step": 343 + }, + { + "epoch": 0.03359375, + "grad_norm": 0.8363533616065979, + "learning_rate": 0.0004995896623701412, + "loss": 5.1914, + "step": 344 + }, + { + "epoch": 0.03369140625, + "grad_norm": 1.090147614479065, + "learning_rate": 0.0004995854224854243, + "loss": 5.2031, + "step": 345 + }, + { + "epoch": 0.0337890625, + "grad_norm": 1.313947319984436, + "learning_rate": 0.000499581160828675, + "loss": 5.1523, + "step": 346 + }, + { + "epoch": 0.03388671875, + "grad_norm": 1.3441343307495117, + "learning_rate": 0.0004995768774003062, + "loss": 5.1797, + "step": 347 + }, + { + "epoch": 0.033984375, + "grad_norm": 1.4852529764175415, + "learning_rate": 0.0004995725722007335, + "loss": 5.2031, + "step": 348 + }, + { + "epoch": 0.03408203125, + "grad_norm": 1.9018231630325317, + "learning_rate": 0.0004995682452303741, + "loss": 5.1797, + "step": 349 + }, + { + "epoch": 0.0341796875, + "grad_norm": 0.8266469836235046, + "learning_rate": 0.0004995638964896475, + "loss": 5.1758, + "step": 350 + }, + { + "epoch": 0.03427734375, + "grad_norm": 0.971101701259613, + "learning_rate": 0.0004995595259789752, + "loss": 5.168, + "step": 351 + }, + { + "epoch": 0.034375, + "grad_norm": 1.2720019817352295, + "learning_rate": 0.0004995551336987809, + "loss": 5.1914, + "step": 352 + }, + { + "epoch": 0.03447265625, + "grad_norm": 1.524877667427063, + "learning_rate": 0.0004995507196494905, + "loss": 5.1836, + "step": 353 + }, + { + "epoch": 0.0345703125, + "grad_norm": 1.155196189880371, + "learning_rate": 0.0004995462838315319, + "loss": 5.1484, + "step": 354 + }, + { + "epoch": 0.03466796875, + "grad_norm": 1.2725322246551514, + "learning_rate": 0.0004995418262453351, + "loss": 5.1562, + "step": 355 + }, + { + "epoch": 0.034765625, + "grad_norm": 1.064315915107727, + "learning_rate": 0.000499537346891332, + "loss": 5.1641, + "step": 356 + }, + { + "epoch": 0.03486328125, + "grad_norm": 1.1174203157424927, + "learning_rate": 0.0004995328457699573, + "loss": 5.1289, + "step": 357 + }, + { + "epoch": 0.0349609375, + "grad_norm": 1.5893704891204834, + "learning_rate": 0.000499528322881647, + "loss": 5.1523, + "step": 358 + }, + { + "epoch": 0.03505859375, + "grad_norm": 1.3740280866622925, + "learning_rate": 0.0004995237782268397, + "loss": 5.1289, + "step": 359 + }, + { + "epoch": 0.03515625, + "grad_norm": 1.560889720916748, + "learning_rate": 0.000499519211805976, + "loss": 5.1875, + "step": 360 + }, + { + "epoch": 0.03525390625, + "grad_norm": 1.1352694034576416, + "learning_rate": 0.0004995146236194984, + "loss": 5.1758, + "step": 361 + }, + { + "epoch": 0.0353515625, + "grad_norm": 1.4046801328659058, + "learning_rate": 0.0004995100136678519, + "loss": 5.1758, + "step": 362 + }, + { + "epoch": 0.03544921875, + "grad_norm": 1.10200035572052, + "learning_rate": 0.0004995053819514834, + "loss": 5.1523, + "step": 363 + }, + { + "epoch": 0.035546875, + "grad_norm": 1.179033875465393, + "learning_rate": 0.0004995007284708418, + "loss": 5.1367, + "step": 364 + }, + { + "epoch": 0.03564453125, + "grad_norm": 1.1701732873916626, + "learning_rate": 0.0004994960532263782, + "loss": 5.1406, + "step": 365 + }, + { + "epoch": 0.0357421875, + "grad_norm": 1.096832513809204, + "learning_rate": 0.000499491356218546, + "loss": 5.1484, + "step": 366 + }, + { + "epoch": 0.03583984375, + "grad_norm": 1.4660563468933105, + "learning_rate": 0.0004994866374478003, + "loss": 5.1562, + "step": 367 + }, + { + "epoch": 0.0359375, + "grad_norm": 1.2725329399108887, + "learning_rate": 0.0004994818969145989, + "loss": 5.1719, + "step": 368 + }, + { + "epoch": 0.03603515625, + "grad_norm": 1.5378856658935547, + "learning_rate": 0.000499477134619401, + "loss": 5.1758, + "step": 369 + }, + { + "epoch": 0.0361328125, + "grad_norm": 1.0054930448532104, + "learning_rate": 0.0004994723505626685, + "loss": 5.2031, + "step": 370 + }, + { + "epoch": 0.03623046875, + "grad_norm": 1.3905240297317505, + "learning_rate": 0.0004994675447448651, + "loss": 5.1562, + "step": 371 + }, + { + "epoch": 0.036328125, + "grad_norm": 1.2946956157684326, + "learning_rate": 0.0004994627171664565, + "loss": 5.1641, + "step": 372 + }, + { + "epoch": 0.03642578125, + "grad_norm": 0.8594855070114136, + "learning_rate": 0.0004994578678279112, + "loss": 5.1133, + "step": 373 + }, + { + "epoch": 0.0365234375, + "grad_norm": 1.0982156991958618, + "learning_rate": 0.0004994529967296989, + "loss": 5.1445, + "step": 374 + }, + { + "epoch": 0.03662109375, + "grad_norm": 1.197832703590393, + "learning_rate": 0.0004994481038722919, + "loss": 5.1875, + "step": 375 + }, + { + "epoch": 0.03671875, + "grad_norm": 0.903097927570343, + "learning_rate": 0.0004994431892561646, + "loss": 5.1523, + "step": 376 + }, + { + "epoch": 0.03681640625, + "grad_norm": 1.064278483390808, + "learning_rate": 0.0004994382528817935, + "loss": 5.1602, + "step": 377 + }, + { + "epoch": 0.0369140625, + "grad_norm": 1.0776413679122925, + "learning_rate": 0.0004994332947496568, + "loss": 5.1406, + "step": 378 + }, + { + "epoch": 0.03701171875, + "grad_norm": 0.99156254529953, + "learning_rate": 0.0004994283148602357, + "loss": 5.1484, + "step": 379 + }, + { + "epoch": 0.037109375, + "grad_norm": 1.25286066532135, + "learning_rate": 0.0004994233132140125, + "loss": 5.1367, + "step": 380 + }, + { + "epoch": 0.03720703125, + "grad_norm": 0.9926852583885193, + "learning_rate": 0.0004994182898114722, + "loss": 5.1328, + "step": 381 + }, + { + "epoch": 0.0373046875, + "grad_norm": 1.0031254291534424, + "learning_rate": 0.0004994132446531019, + "loss": 5.125, + "step": 382 + }, + { + "epoch": 0.03740234375, + "grad_norm": 1.1352587938308716, + "learning_rate": 0.0004994081777393907, + "loss": 5.1719, + "step": 383 + }, + { + "epoch": 0.0375, + "grad_norm": 1.145836591720581, + "learning_rate": 0.0004994030890708297, + "loss": 5.1211, + "step": 384 + }, + { + "epoch": 0.03759765625, + "grad_norm": 1.125534176826477, + "learning_rate": 0.0004993979786479121, + "loss": 5.125, + "step": 385 + }, + { + "epoch": 0.0376953125, + "grad_norm": 0.9138064384460449, + "learning_rate": 0.0004993928464711337, + "loss": 5.1289, + "step": 386 + }, + { + "epoch": 0.03779296875, + "grad_norm": 1.0376123189926147, + "learning_rate": 0.0004993876925409917, + "loss": 5.1328, + "step": 387 + }, + { + "epoch": 0.037890625, + "grad_norm": 1.0181726217269897, + "learning_rate": 0.0004993825168579857, + "loss": 5.1289, + "step": 388 + }, + { + "epoch": 0.03798828125, + "grad_norm": 1.191735029220581, + "learning_rate": 0.0004993773194226178, + "loss": 5.1172, + "step": 389 + }, + { + "epoch": 0.0380859375, + "grad_norm": 1.0628931522369385, + "learning_rate": 0.0004993721002353915, + "loss": 5.1133, + "step": 390 + }, + { + "epoch": 0.03818359375, + "grad_norm": 1.2050843238830566, + "learning_rate": 0.0004993668592968129, + "loss": 5.1172, + "step": 391 + }, + { + "epoch": 0.03828125, + "grad_norm": 0.8847583532333374, + "learning_rate": 0.0004993615966073902, + "loss": 5.1328, + "step": 392 + }, + { + "epoch": 0.03837890625, + "grad_norm": 0.9037215709686279, + "learning_rate": 0.0004993563121676332, + "loss": 5.1406, + "step": 393 + }, + { + "epoch": 0.0384765625, + "grad_norm": 0.8788461685180664, + "learning_rate": 0.0004993510059780546, + "loss": 5.1523, + "step": 394 + }, + { + "epoch": 0.03857421875, + "grad_norm": 0.788327693939209, + "learning_rate": 0.0004993456780391686, + "loss": 5.1016, + "step": 395 + }, + { + "epoch": 0.038671875, + "grad_norm": 0.874692440032959, + "learning_rate": 0.0004993403283514919, + "loss": 5.0938, + "step": 396 + }, + { + "epoch": 0.03876953125, + "grad_norm": 1.0633156299591064, + "learning_rate": 0.0004993349569155428, + "loss": 5.1289, + "step": 397 + }, + { + "epoch": 0.0388671875, + "grad_norm": 1.076167106628418, + "learning_rate": 0.0004993295637318423, + "loss": 5.1172, + "step": 398 + }, + { + "epoch": 0.03896484375, + "grad_norm": 0.9785029292106628, + "learning_rate": 0.0004993241488009131, + "loss": 5.0977, + "step": 399 + }, + { + "epoch": 0.0390625, + "grad_norm": 1.0597959756851196, + "learning_rate": 0.0004993187121232801, + "loss": 5.1094, + "step": 400 + }, + { + "epoch": 0.03916015625, + "grad_norm": 1.2436349391937256, + "learning_rate": 0.0004993132536994705, + "loss": 5.1289, + "step": 401 + }, + { + "epoch": 0.0392578125, + "grad_norm": 1.0709871053695679, + "learning_rate": 0.0004993077735300133, + "loss": 5.1328, + "step": 402 + }, + { + "epoch": 0.03935546875, + "grad_norm": 1.018754005432129, + "learning_rate": 0.0004993022716154399, + "loss": 5.1094, + "step": 403 + }, + { + "epoch": 0.039453125, + "grad_norm": 0.8745573163032532, + "learning_rate": 0.0004992967479562836, + "loss": 5.1172, + "step": 404 + }, + { + "epoch": 0.03955078125, + "grad_norm": 0.8379338979721069, + "learning_rate": 0.0004992912025530799, + "loss": 5.1406, + "step": 405 + }, + { + "epoch": 0.0396484375, + "grad_norm": 0.8244563341140747, + "learning_rate": 0.0004992856354063663, + "loss": 5.1328, + "step": 406 + }, + { + "epoch": 0.03974609375, + "grad_norm": 1.0282996892929077, + "learning_rate": 0.0004992800465166826, + "loss": 5.1016, + "step": 407 + }, + { + "epoch": 0.03984375, + "grad_norm": 0.8590074777603149, + "learning_rate": 0.0004992744358845707, + "loss": 5.1094, + "step": 408 + }, + { + "epoch": 0.03994140625, + "grad_norm": 0.7324565052986145, + "learning_rate": 0.0004992688035105744, + "loss": 5.1094, + "step": 409 + }, + { + "epoch": 0.0400390625, + "grad_norm": 0.8005876541137695, + "learning_rate": 0.0004992631493952395, + "loss": 5.1094, + "step": 410 + }, + { + "epoch": 0.04013671875, + "grad_norm": 1.217882513999939, + "learning_rate": 0.0004992574735391144, + "loss": 5.0938, + "step": 411 + }, + { + "epoch": 0.040234375, + "grad_norm": 1.186151385307312, + "learning_rate": 0.0004992517759427494, + "loss": 5.1094, + "step": 412 + }, + { + "epoch": 0.04033203125, + "grad_norm": 0.851876974105835, + "learning_rate": 0.0004992460566066967, + "loss": 5.0781, + "step": 413 + }, + { + "epoch": 0.0404296875, + "grad_norm": 0.8220931887626648, + "learning_rate": 0.0004992403155315107, + "loss": 5.0977, + "step": 414 + }, + { + "epoch": 0.04052734375, + "grad_norm": 0.848565399646759, + "learning_rate": 0.0004992345527177482, + "loss": 5.0898, + "step": 415 + }, + { + "epoch": 0.040625, + "grad_norm": 1.1592659950256348, + "learning_rate": 0.0004992287681659676, + "loss": 5.0781, + "step": 416 + }, + { + "epoch": 0.04072265625, + "grad_norm": 1.0059559345245361, + "learning_rate": 0.0004992229618767298, + "loss": 5.0781, + "step": 417 + }, + { + "epoch": 0.0408203125, + "grad_norm": 0.9418927431106567, + "learning_rate": 0.0004992171338505975, + "loss": 5.0898, + "step": 418 + }, + { + "epoch": 0.04091796875, + "grad_norm": 0.8450960516929626, + "learning_rate": 0.0004992112840881359, + "loss": 5.0898, + "step": 419 + }, + { + "epoch": 0.041015625, + "grad_norm": 0.739745557308197, + "learning_rate": 0.0004992054125899122, + "loss": 5.0625, + "step": 420 + }, + { + "epoch": 0.04111328125, + "grad_norm": 0.6499938368797302, + "learning_rate": 0.0004991995193564953, + "loss": 5.0742, + "step": 421 + }, + { + "epoch": 0.0412109375, + "grad_norm": 0.5470448136329651, + "learning_rate": 0.0004991936043884566, + "loss": 5.0859, + "step": 422 + }, + { + "epoch": 0.04130859375, + "grad_norm": 0.582874059677124, + "learning_rate": 0.0004991876676863697, + "loss": 5.0664, + "step": 423 + }, + { + "epoch": 0.04140625, + "grad_norm": 0.61722731590271, + "learning_rate": 0.0004991817092508099, + "loss": 5.0898, + "step": 424 + }, + { + "epoch": 0.04150390625, + "grad_norm": 0.6884714365005493, + "learning_rate": 0.0004991757290823548, + "loss": 5.0703, + "step": 425 + }, + { + "epoch": 0.0416015625, + "grad_norm": 0.8316398859024048, + "learning_rate": 0.0004991697271815845, + "loss": 5.1367, + "step": 426 + }, + { + "epoch": 0.04169921875, + "grad_norm": 0.9839829802513123, + "learning_rate": 0.0004991637035490805, + "loss": 5.0703, + "step": 427 + }, + { + "epoch": 0.041796875, + "grad_norm": 1.0404824018478394, + "learning_rate": 0.0004991576581854267, + "loss": 5.0586, + "step": 428 + }, + { + "epoch": 0.04189453125, + "grad_norm": 0.8250628709793091, + "learning_rate": 0.0004991515910912095, + "loss": 5.0625, + "step": 429 + }, + { + "epoch": 0.0419921875, + "grad_norm": 0.7997605204582214, + "learning_rate": 0.0004991455022670168, + "loss": 5.0938, + "step": 430 + }, + { + "epoch": 0.04208984375, + "grad_norm": 0.6865594983100891, + "learning_rate": 0.0004991393917134388, + "loss": 5.0781, + "step": 431 + }, + { + "epoch": 0.0421875, + "grad_norm": 0.6902148723602295, + "learning_rate": 0.0004991332594310682, + "loss": 5.0586, + "step": 432 + }, + { + "epoch": 0.04228515625, + "grad_norm": 0.6424334645271301, + "learning_rate": 0.0004991271054204992, + "loss": 5.0742, + "step": 433 + }, + { + "epoch": 0.0423828125, + "grad_norm": 0.6057596802711487, + "learning_rate": 0.0004991209296823285, + "loss": 5.0625, + "step": 434 + }, + { + "epoch": 0.04248046875, + "grad_norm": 0.6817944645881653, + "learning_rate": 0.0004991147322171548, + "loss": 5.082, + "step": 435 + }, + { + "epoch": 0.042578125, + "grad_norm": 0.7707593441009521, + "learning_rate": 0.0004991085130255789, + "loss": 5.0586, + "step": 436 + }, + { + "epoch": 0.04267578125, + "grad_norm": 0.9107893109321594, + "learning_rate": 0.0004991022721082037, + "loss": 5.0898, + "step": 437 + }, + { + "epoch": 0.0427734375, + "grad_norm": 1.1807111501693726, + "learning_rate": 0.0004990960094656342, + "loss": 5.0195, + "step": 438 + }, + { + "epoch": 0.04287109375, + "grad_norm": 1.1950851678848267, + "learning_rate": 0.0004990897250984774, + "loss": 5.0469, + "step": 439 + }, + { + "epoch": 0.04296875, + "grad_norm": 0.7908074855804443, + "learning_rate": 0.0004990834190073428, + "loss": 5.1016, + "step": 440 + }, + { + "epoch": 0.04306640625, + "grad_norm": 0.7892453670501709, + "learning_rate": 0.0004990770911928416, + "loss": 5.0938, + "step": 441 + }, + { + "epoch": 0.0431640625, + "grad_norm": 0.7331576347351074, + "learning_rate": 0.0004990707416555871, + "loss": 5.0508, + "step": 442 + }, + { + "epoch": 0.04326171875, + "grad_norm": 0.7229627966880798, + "learning_rate": 0.000499064370396195, + "loss": 5.0234, + "step": 443 + }, + { + "epoch": 0.043359375, + "grad_norm": 0.8491083979606628, + "learning_rate": 0.000499057977415283, + "loss": 5.0859, + "step": 444 + }, + { + "epoch": 0.04345703125, + "grad_norm": 1.1333506107330322, + "learning_rate": 0.0004990515627134706, + "loss": 5.0859, + "step": 445 + }, + { + "epoch": 0.0435546875, + "grad_norm": 1.0400718450546265, + "learning_rate": 0.00049904512629138, + "loss": 5.0703, + "step": 446 + }, + { + "epoch": 0.04365234375, + "grad_norm": 0.8677104115486145, + "learning_rate": 0.000499038668149635, + "loss": 5.0273, + "step": 447 + }, + { + "epoch": 0.04375, + "grad_norm": 1.0702966451644897, + "learning_rate": 0.0004990321882888617, + "loss": 5.0664, + "step": 448 + }, + { + "epoch": 0.04384765625, + "grad_norm": 0.9019585847854614, + "learning_rate": 0.0004990256867096881, + "loss": 5.0352, + "step": 449 + }, + { + "epoch": 0.0439453125, + "grad_norm": 0.7547096014022827, + "learning_rate": 0.0004990191634127448, + "loss": 5.0664, + "step": 450 + }, + { + "epoch": 0.04404296875, + "grad_norm": 0.6731748580932617, + "learning_rate": 0.0004990126183986639, + "loss": 5.0625, + "step": 451 + }, + { + "epoch": 0.044140625, + "grad_norm": 0.738862931728363, + "learning_rate": 0.0004990060516680802, + "loss": 5.082, + "step": 452 + }, + { + "epoch": 0.04423828125, + "grad_norm": 0.7783480286598206, + "learning_rate": 0.00049899946322163, + "loss": 5.0703, + "step": 453 + }, + { + "epoch": 0.0443359375, + "grad_norm": 0.7385779023170471, + "learning_rate": 0.0004989928530599521, + "loss": 5.0547, + "step": 454 + }, + { + "epoch": 0.04443359375, + "grad_norm": 1.394942283630371, + "learning_rate": 0.0004989862211836873, + "loss": 5.0547, + "step": 455 + }, + { + "epoch": 0.04453125, + "grad_norm": 0.5369975566864014, + "learning_rate": 0.0004989795675934787, + "loss": 5.0781, + "step": 456 + }, + { + "epoch": 0.04462890625, + "grad_norm": 0.6641459465026855, + "learning_rate": 0.0004989728922899712, + "loss": 5.0117, + "step": 457 + }, + { + "epoch": 0.0447265625, + "grad_norm": 0.7093449831008911, + "learning_rate": 0.0004989661952738118, + "loss": 5.0586, + "step": 458 + }, + { + "epoch": 0.04482421875, + "grad_norm": 0.8085068464279175, + "learning_rate": 0.0004989594765456498, + "loss": 5.0508, + "step": 459 + }, + { + "epoch": 0.044921875, + "grad_norm": 1.188184142112732, + "learning_rate": 0.0004989527361061366, + "loss": 5.0391, + "step": 460 + }, + { + "epoch": 0.04501953125, + "grad_norm": 0.9588766098022461, + "learning_rate": 0.0004989459739559257, + "loss": 5.0195, + "step": 461 + }, + { + "epoch": 0.0451171875, + "grad_norm": 0.8630503416061401, + "learning_rate": 0.0004989391900956725, + "loss": 5.0195, + "step": 462 + }, + { + "epoch": 0.04521484375, + "grad_norm": 0.8916110396385193, + "learning_rate": 0.0004989323845260346, + "loss": 5.0586, + "step": 463 + }, + { + "epoch": 0.0453125, + "grad_norm": 0.8974289298057556, + "learning_rate": 0.0004989255572476719, + "loss": 5.0547, + "step": 464 + }, + { + "epoch": 0.04541015625, + "grad_norm": 0.9970776438713074, + "learning_rate": 0.0004989187082612462, + "loss": 5.0469, + "step": 465 + }, + { + "epoch": 0.0455078125, + "grad_norm": 0.885099470615387, + "learning_rate": 0.0004989118375674216, + "loss": 5.043, + "step": 466 + }, + { + "epoch": 0.04560546875, + "grad_norm": 0.659532368183136, + "learning_rate": 0.0004989049451668639, + "loss": 5.0547, + "step": 467 + }, + { + "epoch": 0.045703125, + "grad_norm": 0.8999412655830383, + "learning_rate": 0.0004988980310602415, + "loss": 5.0117, + "step": 468 + }, + { + "epoch": 0.04580078125, + "grad_norm": 0.7855193018913269, + "learning_rate": 0.0004988910952482246, + "loss": 5.0859, + "step": 469 + }, + { + "epoch": 0.0458984375, + "grad_norm": 0.9827584028244019, + "learning_rate": 0.0004988841377314855, + "loss": 5.0586, + "step": 470 + }, + { + "epoch": 0.04599609375, + "grad_norm": 1.13511323928833, + "learning_rate": 0.0004988771585106988, + "loss": 5.0547, + "step": 471 + }, + { + "epoch": 0.04609375, + "grad_norm": 0.8385241627693176, + "learning_rate": 0.000498870157586541, + "loss": 5.0586, + "step": 472 + }, + { + "epoch": 0.04619140625, + "grad_norm": 0.7410266995429993, + "learning_rate": 0.0004988631349596909, + "loss": 5.0664, + "step": 473 + }, + { + "epoch": 0.0462890625, + "grad_norm": 0.6939350962638855, + "learning_rate": 0.0004988560906308291, + "loss": 5.043, + "step": 474 + }, + { + "epoch": 0.04638671875, + "grad_norm": 0.7858873605728149, + "learning_rate": 0.0004988490246006388, + "loss": 5.0469, + "step": 475 + }, + { + "epoch": 0.046484375, + "grad_norm": 1.0619468688964844, + "learning_rate": 0.0004988419368698047, + "loss": 5.0234, + "step": 476 + }, + { + "epoch": 0.04658203125, + "grad_norm": 1.3020342588424683, + "learning_rate": 0.0004988348274390141, + "loss": 5.0469, + "step": 477 + }, + { + "epoch": 0.0466796875, + "grad_norm": 0.7949604988098145, + "learning_rate": 0.0004988276963089561, + "loss": 5.0391, + "step": 478 + }, + { + "epoch": 0.04677734375, + "grad_norm": 0.8782525658607483, + "learning_rate": 0.0004988205434803222, + "loss": 5.043, + "step": 479 + }, + { + "epoch": 0.046875, + "grad_norm": 0.9638606309890747, + "learning_rate": 0.0004988133689538055, + "loss": 5.0703, + "step": 480 + }, + { + "epoch": 0.04697265625, + "grad_norm": 0.8633697032928467, + "learning_rate": 0.0004988061727301019, + "loss": 5.0, + "step": 481 + }, + { + "epoch": 0.0470703125, + "grad_norm": 0.719272255897522, + "learning_rate": 0.0004987989548099087, + "loss": 5.043, + "step": 482 + }, + { + "epoch": 0.04716796875, + "grad_norm": 0.7947590947151184, + "learning_rate": 0.0004987917151939257, + "loss": 5.0273, + "step": 483 + }, + { + "epoch": 0.047265625, + "grad_norm": 0.8721004128456116, + "learning_rate": 0.000498784453882855, + "loss": 5.0469, + "step": 484 + }, + { + "epoch": 0.04736328125, + "grad_norm": 0.840782880783081, + "learning_rate": 0.0004987771708774001, + "loss": 5.0273, + "step": 485 + }, + { + "epoch": 0.0474609375, + "grad_norm": 0.6539798974990845, + "learning_rate": 0.0004987698661782674, + "loss": 5.0273, + "step": 486 + }, + { + "epoch": 0.04755859375, + "grad_norm": 0.6629400253295898, + "learning_rate": 0.0004987625397861649, + "loss": 5.0273, + "step": 487 + }, + { + "epoch": 0.04765625, + "grad_norm": 0.69793301820755, + "learning_rate": 0.0004987551917018028, + "loss": 4.9961, + "step": 488 + }, + { + "epoch": 0.04775390625, + "grad_norm": 0.8525344133377075, + "learning_rate": 0.0004987478219258936, + "loss": 5.0508, + "step": 489 + }, + { + "epoch": 0.0478515625, + "grad_norm": 0.9601410627365112, + "learning_rate": 0.0004987404304591514, + "loss": 5.0391, + "step": 490 + }, + { + "epoch": 0.04794921875, + "grad_norm": 0.8520921468734741, + "learning_rate": 0.0004987330173022932, + "loss": 5.0273, + "step": 491 + }, + { + "epoch": 0.048046875, + "grad_norm": 0.8635746240615845, + "learning_rate": 0.0004987255824560374, + "loss": 5.0234, + "step": 492 + }, + { + "epoch": 0.04814453125, + "grad_norm": 0.998202383518219, + "learning_rate": 0.0004987181259211048, + "loss": 5.0273, + "step": 493 + }, + { + "epoch": 0.0482421875, + "grad_norm": 1.2011066675186157, + "learning_rate": 0.0004987106476982184, + "loss": 5.0312, + "step": 494 + }, + { + "epoch": 0.04833984375, + "grad_norm": 0.7638375163078308, + "learning_rate": 0.0004987031477881029, + "loss": 5.0312, + "step": 495 + }, + { + "epoch": 0.0484375, + "grad_norm": 0.70158451795578, + "learning_rate": 0.0004986956261914856, + "loss": 5.0156, + "step": 496 + }, + { + "epoch": 0.04853515625, + "grad_norm": 0.7986351847648621, + "learning_rate": 0.0004986880829090954, + "loss": 5.0234, + "step": 497 + }, + { + "epoch": 0.0486328125, + "grad_norm": 0.8812294006347656, + "learning_rate": 0.0004986805179416638, + "loss": 5.0156, + "step": 498 + }, + { + "epoch": 0.04873046875, + "grad_norm": 0.8842159509658813, + "learning_rate": 0.0004986729312899242, + "loss": 5.0156, + "step": 499 + }, + { + "epoch": 0.048828125, + "grad_norm": 0.8196664452552795, + "learning_rate": 0.000498665322954612, + "loss": 5.0078, + "step": 500 + }, + { + "epoch": 0.04892578125, + "grad_norm": 0.774849534034729, + "learning_rate": 0.0004986576929364646, + "loss": 5.0273, + "step": 501 + }, + { + "epoch": 0.0490234375, + "grad_norm": 0.6376418471336365, + "learning_rate": 0.0004986500412362219, + "loss": 5.0234, + "step": 502 + }, + { + "epoch": 0.04912109375, + "grad_norm": 0.49704745411872864, + "learning_rate": 0.0004986423678546257, + "loss": 4.9922, + "step": 503 + }, + { + "epoch": 0.04921875, + "grad_norm": 0.49575215578079224, + "learning_rate": 0.0004986346727924197, + "loss": 5.0234, + "step": 504 + }, + { + "epoch": 0.04931640625, + "grad_norm": 0.5091367363929749, + "learning_rate": 0.0004986269560503501, + "loss": 4.9961, + "step": 505 + }, + { + "epoch": 0.0494140625, + "grad_norm": 0.7224910855293274, + "learning_rate": 0.0004986192176291647, + "loss": 4.9922, + "step": 506 + }, + { + "epoch": 0.04951171875, + "grad_norm": 0.8590059280395508, + "learning_rate": 0.0004986114575296141, + "loss": 5.0234, + "step": 507 + }, + { + "epoch": 0.049609375, + "grad_norm": 0.7906845808029175, + "learning_rate": 0.0004986036757524501, + "loss": 5.0195, + "step": 508 + }, + { + "epoch": 0.04970703125, + "grad_norm": 0.6573899388313293, + "learning_rate": 0.0004985958722984275, + "loss": 5.0117, + "step": 509 + }, + { + "epoch": 0.0498046875, + "grad_norm": 0.5708054900169373, + "learning_rate": 0.0004985880471683026, + "loss": 5.0, + "step": 510 + }, + { + "epoch": 0.04990234375, + "grad_norm": 0.6391012668609619, + "learning_rate": 0.000498580200362834, + "loss": 5.0156, + "step": 511 + }, + { + "epoch": 0.05, + "grad_norm": 0.6893157362937927, + "learning_rate": 0.0004985723318827824, + "loss": 5.0078, + "step": 512 + }, + { + "epoch": 0.05009765625, + "grad_norm": 0.7473337054252625, + "learning_rate": 0.0004985644417289107, + "loss": 5.0352, + "step": 513 + }, + { + "epoch": 0.0501953125, + "grad_norm": 0.6992818713188171, + "learning_rate": 0.0004985565299019836, + "loss": 4.9766, + "step": 514 + }, + { + "epoch": 0.05029296875, + "grad_norm": 0.5503939986228943, + "learning_rate": 0.0004985485964027682, + "loss": 5.0156, + "step": 515 + }, + { + "epoch": 0.050390625, + "grad_norm": 0.5531403422355652, + "learning_rate": 0.0004985406412320337, + "loss": 5.0234, + "step": 516 + }, + { + "epoch": 0.05048828125, + "grad_norm": 0.5232397317886353, + "learning_rate": 0.0004985326643905511, + "loss": 4.9922, + "step": 517 + }, + { + "epoch": 0.0505859375, + "grad_norm": 0.5153787732124329, + "learning_rate": 0.0004985246658790939, + "loss": 4.9922, + "step": 518 + }, + { + "epoch": 0.05068359375, + "grad_norm": 0.7395737767219543, + "learning_rate": 0.0004985166456984375, + "loss": 4.9961, + "step": 519 + }, + { + "epoch": 0.05078125, + "grad_norm": 0.7297258973121643, + "learning_rate": 0.0004985086038493591, + "loss": 5.0195, + "step": 520 + }, + { + "epoch": 0.05087890625, + "grad_norm": 0.8792772889137268, + "learning_rate": 0.0004985005403326387, + "loss": 5.0195, + "step": 521 + }, + { + "epoch": 0.0509765625, + "grad_norm": 0.8485317230224609, + "learning_rate": 0.0004984924551490577, + "loss": 5.0039, + "step": 522 + }, + { + "epoch": 0.05107421875, + "grad_norm": 0.7504170536994934, + "learning_rate": 0.0004984843482994001, + "loss": 5.0391, + "step": 523 + }, + { + "epoch": 0.051171875, + "grad_norm": 0.6931923031806946, + "learning_rate": 0.0004984762197844516, + "loss": 4.9883, + "step": 524 + }, + { + "epoch": 0.05126953125, + "grad_norm": 0.7201303839683533, + "learning_rate": 0.0004984680696050004, + "loss": 4.9844, + "step": 525 + }, + { + "epoch": 0.0513671875, + "grad_norm": 0.8321727514266968, + "learning_rate": 0.0004984598977618366, + "loss": 5.0234, + "step": 526 + }, + { + "epoch": 0.05146484375, + "grad_norm": 1.0283933877944946, + "learning_rate": 0.0004984517042557523, + "loss": 5.0156, + "step": 527 + }, + { + "epoch": 0.0515625, + "grad_norm": 1.3810787200927734, + "learning_rate": 0.0004984434890875417, + "loss": 5.0273, + "step": 528 + }, + { + "epoch": 0.05166015625, + "grad_norm": 0.75733482837677, + "learning_rate": 0.0004984352522580015, + "loss": 4.9844, + "step": 529 + }, + { + "epoch": 0.0517578125, + "grad_norm": 0.7140693068504333, + "learning_rate": 0.0004984269937679301, + "loss": 4.9961, + "step": 530 + }, + { + "epoch": 0.05185546875, + "grad_norm": 0.9317624568939209, + "learning_rate": 0.000498418713618128, + "loss": 4.9805, + "step": 531 + }, + { + "epoch": 0.051953125, + "grad_norm": 0.8527510762214661, + "learning_rate": 0.0004984104118093979, + "loss": 4.9961, + "step": 532 + }, + { + "epoch": 0.05205078125, + "grad_norm": 0.8085970878601074, + "learning_rate": 0.0004984020883425447, + "loss": 4.9883, + "step": 533 + }, + { + "epoch": 0.0521484375, + "grad_norm": 0.8414132595062256, + "learning_rate": 0.0004983937432183754, + "loss": 4.9727, + "step": 534 + }, + { + "epoch": 0.05224609375, + "grad_norm": 0.7883987426757812, + "learning_rate": 0.0004983853764376987, + "loss": 4.9766, + "step": 535 + }, + { + "epoch": 0.05234375, + "grad_norm": 0.8301931619644165, + "learning_rate": 0.000498376988001326, + "loss": 4.9688, + "step": 536 + }, + { + "epoch": 0.05244140625, + "grad_norm": 0.815565288066864, + "learning_rate": 0.0004983685779100702, + "loss": 5.0273, + "step": 537 + }, + { + "epoch": 0.0525390625, + "grad_norm": 0.7249122262001038, + "learning_rate": 0.0004983601461647469, + "loss": 5.0234, + "step": 538 + }, + { + "epoch": 0.05263671875, + "grad_norm": 0.6938775181770325, + "learning_rate": 0.0004983516927661733, + "loss": 4.9805, + "step": 539 + }, + { + "epoch": 0.052734375, + "grad_norm": 0.6271894574165344, + "learning_rate": 0.0004983432177151691, + "loss": 4.9922, + "step": 540 + }, + { + "epoch": 0.05283203125, + "grad_norm": 0.5879946351051331, + "learning_rate": 0.0004983347210125558, + "loss": 4.9883, + "step": 541 + }, + { + "epoch": 0.0529296875, + "grad_norm": 0.6193601489067078, + "learning_rate": 0.0004983262026591569, + "loss": 4.9688, + "step": 542 + }, + { + "epoch": 0.05302734375, + "grad_norm": 0.7117279767990112, + "learning_rate": 0.0004983176626557986, + "loss": 5.0195, + "step": 543 + }, + { + "epoch": 0.053125, + "grad_norm": 0.6014550924301147, + "learning_rate": 0.0004983091010033083, + "loss": 5.0, + "step": 544 + }, + { + "epoch": 0.05322265625, + "grad_norm": 0.5970187783241272, + "learning_rate": 0.0004983005177025164, + "loss": 4.9688, + "step": 545 + }, + { + "epoch": 0.0533203125, + "grad_norm": 0.602258563041687, + "learning_rate": 0.0004982919127542549, + "loss": 5.0156, + "step": 546 + }, + { + "epoch": 0.05341796875, + "grad_norm": 0.5443257689476013, + "learning_rate": 0.0004982832861593579, + "loss": 4.9805, + "step": 547 + }, + { + "epoch": 0.053515625, + "grad_norm": 0.5458142161369324, + "learning_rate": 0.0004982746379186616, + "loss": 4.9531, + "step": 548 + }, + { + "epoch": 0.05361328125, + "grad_norm": 0.5225309729576111, + "learning_rate": 0.0004982659680330047, + "loss": 5.0039, + "step": 549 + }, + { + "epoch": 0.0537109375, + "grad_norm": 0.5642147660255432, + "learning_rate": 0.0004982572765032275, + "loss": 4.9961, + "step": 550 + }, + { + "epoch": 0.05380859375, + "grad_norm": 0.6520267724990845, + "learning_rate": 0.0004982485633301725, + "loss": 5.0078, + "step": 551 + }, + { + "epoch": 0.05390625, + "grad_norm": 0.7051550149917603, + "learning_rate": 0.0004982398285146846, + "loss": 5.0, + "step": 552 + }, + { + "epoch": 0.05400390625, + "grad_norm": 0.6687464714050293, + "learning_rate": 0.0004982310720576103, + "loss": 4.9766, + "step": 553 + }, + { + "epoch": 0.0541015625, + "grad_norm": 0.60595703125, + "learning_rate": 0.0004982222939597989, + "loss": 4.957, + "step": 554 + }, + { + "epoch": 0.05419921875, + "grad_norm": 0.5594586133956909, + "learning_rate": 0.0004982134942221009, + "loss": 4.9883, + "step": 555 + }, + { + "epoch": 0.054296875, + "grad_norm": 0.5148446559906006, + "learning_rate": 0.0004982046728453696, + "loss": 4.9844, + "step": 556 + }, + { + "epoch": 0.05439453125, + "grad_norm": 0.5946763753890991, + "learning_rate": 0.0004981958298304602, + "loss": 5.0156, + "step": 557 + }, + { + "epoch": 0.0544921875, + "grad_norm": 0.716316282749176, + "learning_rate": 0.0004981869651782299, + "loss": 4.9805, + "step": 558 + }, + { + "epoch": 0.05458984375, + "grad_norm": 0.794476330280304, + "learning_rate": 0.0004981780788895382, + "loss": 5.0117, + "step": 559 + }, + { + "epoch": 0.0546875, + "grad_norm": 0.7105295658111572, + "learning_rate": 0.0004981691709652464, + "loss": 4.9766, + "step": 560 + }, + { + "epoch": 0.05478515625, + "grad_norm": 0.5912784337997437, + "learning_rate": 0.0004981602414062181, + "loss": 4.9414, + "step": 561 + }, + { + "epoch": 0.0548828125, + "grad_norm": 0.5801540613174438, + "learning_rate": 0.000498151290213319, + "loss": 4.9883, + "step": 562 + }, + { + "epoch": 0.05498046875, + "grad_norm": 0.6369305849075317, + "learning_rate": 0.0004981423173874169, + "loss": 5.0039, + "step": 563 + }, + { + "epoch": 0.055078125, + "grad_norm": 0.6735507845878601, + "learning_rate": 0.0004981333229293816, + "loss": 4.9648, + "step": 564 + }, + { + "epoch": 0.05517578125, + "grad_norm": 0.6315239667892456, + "learning_rate": 0.000498124306840085, + "loss": 5.0078, + "step": 565 + }, + { + "epoch": 0.0552734375, + "grad_norm": 0.710253119468689, + "learning_rate": 0.0004981152691204011, + "loss": 4.9844, + "step": 566 + }, + { + "epoch": 0.05537109375, + "grad_norm": 0.6913606524467468, + "learning_rate": 0.0004981062097712064, + "loss": 4.9766, + "step": 567 + }, + { + "epoch": 0.05546875, + "grad_norm": 0.5899546146392822, + "learning_rate": 0.0004980971287933787, + "loss": 4.9688, + "step": 568 + }, + { + "epoch": 0.05556640625, + "grad_norm": 0.5400623083114624, + "learning_rate": 0.0004980880261877986, + "loss": 4.9805, + "step": 569 + }, + { + "epoch": 0.0556640625, + "grad_norm": 0.5193256139755249, + "learning_rate": 0.0004980789019553484, + "loss": 4.9453, + "step": 570 + }, + { + "epoch": 0.05576171875, + "grad_norm": 0.5622848272323608, + "learning_rate": 0.0004980697560969127, + "loss": 4.957, + "step": 571 + }, + { + "epoch": 0.055859375, + "grad_norm": 0.6163820028305054, + "learning_rate": 0.0004980605886133782, + "loss": 4.9766, + "step": 572 + }, + { + "epoch": 0.05595703125, + "grad_norm": 0.6021759510040283, + "learning_rate": 0.0004980513995056334, + "loss": 4.957, + "step": 573 + }, + { + "epoch": 0.0560546875, + "grad_norm": 0.5321788191795349, + "learning_rate": 0.0004980421887745695, + "loss": 4.9805, + "step": 574 + }, + { + "epoch": 0.05615234375, + "grad_norm": 0.4939315617084503, + "learning_rate": 0.0004980329564210792, + "loss": 5.0195, + "step": 575 + }, + { + "epoch": 0.05625, + "grad_norm": 0.4567241370677948, + "learning_rate": 0.0004980237024460573, + "loss": 4.9648, + "step": 576 + }, + { + "epoch": 0.05634765625, + "grad_norm": 0.44762229919433594, + "learning_rate": 0.0004980144268504012, + "loss": 4.9961, + "step": 577 + }, + { + "epoch": 0.0564453125, + "grad_norm": 0.5556433200836182, + "learning_rate": 0.00049800512963501, + "loss": 4.9844, + "step": 578 + }, + { + "epoch": 0.05654296875, + "grad_norm": 0.6940642595291138, + "learning_rate": 0.0004979958108007851, + "loss": 4.9844, + "step": 579 + }, + { + "epoch": 0.056640625, + "grad_norm": 0.7792614102363586, + "learning_rate": 0.0004979864703486297, + "loss": 4.9648, + "step": 580 + }, + { + "epoch": 0.05673828125, + "grad_norm": 0.8071824908256531, + "learning_rate": 0.0004979771082794495, + "loss": 4.9727, + "step": 581 + }, + { + "epoch": 0.0568359375, + "grad_norm": 0.9120080471038818, + "learning_rate": 0.0004979677245941519, + "loss": 4.9883, + "step": 582 + }, + { + "epoch": 0.05693359375, + "grad_norm": 0.9239040017127991, + "learning_rate": 0.0004979583192936468, + "loss": 4.9609, + "step": 583 + }, + { + "epoch": 0.05703125, + "grad_norm": 1.002406120300293, + "learning_rate": 0.0004979488923788459, + "loss": 4.9766, + "step": 584 + }, + { + "epoch": 0.05712890625, + "grad_norm": 0.8661404848098755, + "learning_rate": 0.0004979394438506629, + "loss": 5.0, + "step": 585 + }, + { + "epoch": 0.0572265625, + "grad_norm": 0.6225184798240662, + "learning_rate": 0.000497929973710014, + "loss": 4.9727, + "step": 586 + }, + { + "epoch": 0.05732421875, + "grad_norm": 0.5944101810455322, + "learning_rate": 0.0004979204819578172, + "loss": 4.9414, + "step": 587 + }, + { + "epoch": 0.057421875, + "grad_norm": 0.5297685265541077, + "learning_rate": 0.0004979109685949926, + "loss": 4.9844, + "step": 588 + }, + { + "epoch": 0.05751953125, + "grad_norm": 0.43607059121131897, + "learning_rate": 0.0004979014336224625, + "loss": 4.9961, + "step": 589 + }, + { + "epoch": 0.0576171875, + "grad_norm": 0.48334574699401855, + "learning_rate": 0.0004978918770411513, + "loss": 4.9648, + "step": 590 + }, + { + "epoch": 0.05771484375, + "grad_norm": 0.5257386565208435, + "learning_rate": 0.0004978822988519853, + "loss": 4.9766, + "step": 591 + }, + { + "epoch": 0.0578125, + "grad_norm": 0.6425731182098389, + "learning_rate": 0.0004978726990558931, + "loss": 4.9492, + "step": 592 + }, + { + "epoch": 0.05791015625, + "grad_norm": 1.0080450773239136, + "learning_rate": 0.0004978630776538056, + "loss": 4.9727, + "step": 593 + }, + { + "epoch": 0.0580078125, + "grad_norm": 1.2617547512054443, + "learning_rate": 0.000497853434646655, + "loss": 5.0156, + "step": 594 + }, + { + "epoch": 0.05810546875, + "grad_norm": 0.8185104727745056, + "learning_rate": 0.0004978437700353766, + "loss": 4.9648, + "step": 595 + }, + { + "epoch": 0.058203125, + "grad_norm": 0.996583104133606, + "learning_rate": 0.0004978340838209071, + "loss": 4.9336, + "step": 596 + }, + { + "epoch": 0.05830078125, + "grad_norm": 0.8999793529510498, + "learning_rate": 0.0004978243760041855, + "loss": 4.9805, + "step": 597 + }, + { + "epoch": 0.0583984375, + "grad_norm": 0.7666017413139343, + "learning_rate": 0.0004978146465861531, + "loss": 4.918, + "step": 598 + }, + { + "epoch": 0.05849609375, + "grad_norm": 0.7835460901260376, + "learning_rate": 0.0004978048955677529, + "loss": 4.9766, + "step": 599 + }, + { + "epoch": 0.05859375, + "grad_norm": 0.7298296689987183, + "learning_rate": 0.0004977951229499302, + "loss": 4.9648, + "step": 600 + }, + { + "epoch": 0.05869140625, + "grad_norm": 0.7160419225692749, + "learning_rate": 0.0004977853287336325, + "loss": 4.9531, + "step": 601 + }, + { + "epoch": 0.0587890625, + "grad_norm": 0.9255022406578064, + "learning_rate": 0.0004977755129198092, + "loss": 4.9727, + "step": 602 + }, + { + "epoch": 0.05888671875, + "grad_norm": 1.1262624263763428, + "learning_rate": 0.0004977656755094119, + "loss": 4.9766, + "step": 603 + }, + { + "epoch": 0.058984375, + "grad_norm": 0.8687927722930908, + "learning_rate": 0.0004977558165033942, + "loss": 4.9727, + "step": 604 + }, + { + "epoch": 0.05908203125, + "grad_norm": 0.827407717704773, + "learning_rate": 0.0004977459359027121, + "loss": 4.957, + "step": 605 + }, + { + "epoch": 0.0591796875, + "grad_norm": 0.7043539881706238, + "learning_rate": 0.0004977360337083232, + "loss": 4.9961, + "step": 606 + }, + { + "epoch": 0.05927734375, + "grad_norm": 0.6146707534790039, + "learning_rate": 0.0004977261099211876, + "loss": 4.957, + "step": 607 + }, + { + "epoch": 0.059375, + "grad_norm": 0.5867034196853638, + "learning_rate": 0.0004977161645422672, + "loss": 4.9648, + "step": 608 + }, + { + "epoch": 0.05947265625, + "grad_norm": 0.456230491399765, + "learning_rate": 0.0004977061975725264, + "loss": 4.9805, + "step": 609 + }, + { + "epoch": 0.0595703125, + "grad_norm": 0.4468291997909546, + "learning_rate": 0.000497696209012931, + "loss": 4.9844, + "step": 610 + }, + { + "epoch": 0.05966796875, + "grad_norm": 0.4772776961326599, + "learning_rate": 0.0004976861988644498, + "loss": 4.9453, + "step": 611 + }, + { + "epoch": 0.059765625, + "grad_norm": 0.510353147983551, + "learning_rate": 0.0004976761671280529, + "loss": 4.9688, + "step": 612 + }, + { + "epoch": 0.05986328125, + "grad_norm": 0.4383707046508789, + "learning_rate": 0.0004976661138047128, + "loss": 4.9531, + "step": 613 + }, + { + "epoch": 0.0599609375, + "grad_norm": 0.42035961151123047, + "learning_rate": 0.0004976560388954044, + "loss": 4.9453, + "step": 614 + }, + { + "epoch": 0.06005859375, + "grad_norm": 0.45699962973594666, + "learning_rate": 0.0004976459424011041, + "loss": 4.9531, + "step": 615 + }, + { + "epoch": 0.06015625, + "grad_norm": 0.47027596831321716, + "learning_rate": 0.0004976358243227908, + "loss": 4.9219, + "step": 616 + }, + { + "epoch": 0.06025390625, + "grad_norm": 0.4678795337677002, + "learning_rate": 0.0004976256846614454, + "loss": 4.9609, + "step": 617 + }, + { + "epoch": 0.0603515625, + "grad_norm": 0.4823755919933319, + "learning_rate": 0.0004976155234180507, + "loss": 4.957, + "step": 618 + }, + { + "epoch": 0.06044921875, + "grad_norm": 0.4279949367046356, + "learning_rate": 0.0004976053405935921, + "loss": 4.918, + "step": 619 + }, + { + "epoch": 0.060546875, + "grad_norm": 0.44116416573524475, + "learning_rate": 0.0004975951361890565, + "loss": 4.9297, + "step": 620 + }, + { + "epoch": 0.06064453125, + "grad_norm": 0.5708990097045898, + "learning_rate": 0.0004975849102054332, + "loss": 4.9492, + "step": 621 + }, + { + "epoch": 0.0607421875, + "grad_norm": 0.6033137440681458, + "learning_rate": 0.0004975746626437135, + "loss": 4.9336, + "step": 622 + }, + { + "epoch": 0.06083984375, + "grad_norm": 0.7036421895027161, + "learning_rate": 0.0004975643935048908, + "loss": 4.9297, + "step": 623 + }, + { + "epoch": 0.0609375, + "grad_norm": 0.6817371249198914, + "learning_rate": 0.0004975541027899609, + "loss": 4.9805, + "step": 624 + }, + { + "epoch": 0.06103515625, + "grad_norm": 0.6419718265533447, + "learning_rate": 0.0004975437904999211, + "loss": 4.9375, + "step": 625 + }, + { + "epoch": 0.0611328125, + "grad_norm": 0.6562728881835938, + "learning_rate": 0.0004975334566357712, + "loss": 4.957, + "step": 626 + }, + { + "epoch": 0.06123046875, + "grad_norm": 0.7153109312057495, + "learning_rate": 0.000497523101198513, + "loss": 4.9336, + "step": 627 + }, + { + "epoch": 0.061328125, + "grad_norm": 0.7886594533920288, + "learning_rate": 0.0004975127241891505, + "loss": 4.9492, + "step": 628 + }, + { + "epoch": 0.06142578125, + "grad_norm": 0.8648955225944519, + "learning_rate": 0.0004975023256086896, + "loss": 4.9453, + "step": 629 + }, + { + "epoch": 0.0615234375, + "grad_norm": 0.7228974103927612, + "learning_rate": 0.0004974919054581382, + "loss": 4.9531, + "step": 630 + }, + { + "epoch": 0.06162109375, + "grad_norm": 0.563917875289917, + "learning_rate": 0.0004974814637385067, + "loss": 4.9219, + "step": 631 + }, + { + "epoch": 0.06171875, + "grad_norm": 0.6323292851448059, + "learning_rate": 0.0004974710004508073, + "loss": 4.957, + "step": 632 + }, + { + "epoch": 0.06181640625, + "grad_norm": 0.6532883048057556, + "learning_rate": 0.0004974605155960545, + "loss": 4.957, + "step": 633 + }, + { + "epoch": 0.0619140625, + "grad_norm": 0.5852863788604736, + "learning_rate": 0.0004974500091752643, + "loss": 4.9492, + "step": 634 + }, + { + "epoch": 0.06201171875, + "grad_norm": 0.5243951082229614, + "learning_rate": 0.0004974394811894555, + "loss": 4.9102, + "step": 635 + }, + { + "epoch": 0.062109375, + "grad_norm": 0.4946766495704651, + "learning_rate": 0.0004974289316396487, + "loss": 4.957, + "step": 636 + }, + { + "epoch": 0.06220703125, + "grad_norm": 0.5417998433113098, + "learning_rate": 0.0004974183605268667, + "loss": 4.9531, + "step": 637 + }, + { + "epoch": 0.0623046875, + "grad_norm": 0.5387380719184875, + "learning_rate": 0.0004974077678521343, + "loss": 4.9805, + "step": 638 + }, + { + "epoch": 0.06240234375, + "grad_norm": 0.4830368161201477, + "learning_rate": 0.000497397153616478, + "loss": 4.9375, + "step": 639 + }, + { + "epoch": 0.0625, + "grad_norm": 0.5016372203826904, + "learning_rate": 0.0004973865178209274, + "loss": 4.9297, + "step": 640 + }, + { + "epoch": 0.06259765625, + "grad_norm": 0.5527303814888, + "learning_rate": 0.000497375860466513, + "loss": 4.9297, + "step": 641 + }, + { + "epoch": 0.0626953125, + "grad_norm": 0.5242739319801331, + "learning_rate": 0.0004973651815542682, + "loss": 4.9805, + "step": 642 + }, + { + "epoch": 0.06279296875, + "grad_norm": 0.48130086064338684, + "learning_rate": 0.0004973544810852284, + "loss": 4.957, + "step": 643 + }, + { + "epoch": 0.062890625, + "grad_norm": 0.5349105596542358, + "learning_rate": 0.0004973437590604307, + "loss": 4.9648, + "step": 644 + }, + { + "epoch": 0.06298828125, + "grad_norm": 0.5169801712036133, + "learning_rate": 0.0004973330154809146, + "loss": 4.9375, + "step": 645 + }, + { + "epoch": 0.0630859375, + "grad_norm": 0.480214387178421, + "learning_rate": 0.0004973222503477216, + "loss": 4.9531, + "step": 646 + }, + { + "epoch": 0.06318359375, + "grad_norm": 0.4224573075771332, + "learning_rate": 0.0004973114636618954, + "loss": 4.9219, + "step": 647 + }, + { + "epoch": 0.06328125, + "grad_norm": 0.41642358899116516, + "learning_rate": 0.0004973006554244816, + "loss": 4.918, + "step": 648 + }, + { + "epoch": 0.06337890625, + "grad_norm": 0.4622966945171356, + "learning_rate": 0.000497289825636528, + "loss": 4.9297, + "step": 649 + }, + { + "epoch": 0.0634765625, + "grad_norm": 0.5849423408508301, + "learning_rate": 0.0004972789742990846, + "loss": 4.9258, + "step": 650 + }, + { + "epoch": 0.06357421875, + "grad_norm": 0.5855206847190857, + "learning_rate": 0.0004972681014132031, + "loss": 4.9414, + "step": 651 + }, + { + "epoch": 0.063671875, + "grad_norm": 0.6138719320297241, + "learning_rate": 0.0004972572069799378, + "loss": 4.9141, + "step": 652 + }, + { + "epoch": 0.06376953125, + "grad_norm": 0.5948016047477722, + "learning_rate": 0.0004972462910003447, + "loss": 4.9414, + "step": 653 + }, + { + "epoch": 0.0638671875, + "grad_norm": 0.558039665222168, + "learning_rate": 0.0004972353534754821, + "loss": 4.918, + "step": 654 + }, + { + "epoch": 0.06396484375, + "grad_norm": 0.6696071624755859, + "learning_rate": 0.0004972243944064103, + "loss": 4.9258, + "step": 655 + }, + { + "epoch": 0.0640625, + "grad_norm": 0.7494735717773438, + "learning_rate": 0.0004972134137941918, + "loss": 4.8945, + "step": 656 + }, + { + "epoch": 0.06416015625, + "grad_norm": 0.8260038495063782, + "learning_rate": 0.0004972024116398908, + "loss": 4.9336, + "step": 657 + }, + { + "epoch": 0.0642578125, + "grad_norm": 0.9088923931121826, + "learning_rate": 0.0004971913879445742, + "loss": 4.9531, + "step": 658 + }, + { + "epoch": 0.06435546875, + "grad_norm": 0.7802959084510803, + "learning_rate": 0.0004971803427093105, + "loss": 4.9219, + "step": 659 + }, + { + "epoch": 0.064453125, + "grad_norm": 0.6456225514411926, + "learning_rate": 0.0004971692759351705, + "loss": 4.9414, + "step": 660 + }, + { + "epoch": 0.06455078125, + "grad_norm": 0.6266503930091858, + "learning_rate": 0.0004971581876232272, + "loss": 4.9219, + "step": 661 + }, + { + "epoch": 0.0646484375, + "grad_norm": 0.5530596971511841, + "learning_rate": 0.0004971470777745553, + "loss": 4.9336, + "step": 662 + }, + { + "epoch": 0.06474609375, + "grad_norm": 0.5629117488861084, + "learning_rate": 0.0004971359463902319, + "loss": 4.9727, + "step": 663 + }, + { + "epoch": 0.06484375, + "grad_norm": 0.5490591526031494, + "learning_rate": 0.0004971247934713362, + "loss": 4.9453, + "step": 664 + }, + { + "epoch": 0.06494140625, + "grad_norm": 0.5852833986282349, + "learning_rate": 0.0004971136190189494, + "loss": 4.9414, + "step": 665 + }, + { + "epoch": 0.0650390625, + "grad_norm": 0.6156993508338928, + "learning_rate": 0.0004971024230341546, + "loss": 4.957, + "step": 666 + }, + { + "epoch": 0.06513671875, + "grad_norm": 0.6128381490707397, + "learning_rate": 0.0004970912055180373, + "loss": 4.8867, + "step": 667 + }, + { + "epoch": 0.065234375, + "grad_norm": 0.4521043598651886, + "learning_rate": 0.000497079966471685, + "loss": 4.9141, + "step": 668 + }, + { + "epoch": 0.06533203125, + "grad_norm": 0.42642197012901306, + "learning_rate": 0.0004970687058961873, + "loss": 4.9336, + "step": 669 + }, + { + "epoch": 0.0654296875, + "grad_norm": 0.41460588574409485, + "learning_rate": 0.0004970574237926356, + "loss": 4.9062, + "step": 670 + }, + { + "epoch": 0.06552734375, + "grad_norm": 0.4534105956554413, + "learning_rate": 0.000497046120162124, + "loss": 4.8945, + "step": 671 + }, + { + "epoch": 0.065625, + "grad_norm": 0.45969158411026, + "learning_rate": 0.0004970347950057478, + "loss": 4.918, + "step": 672 + }, + { + "epoch": 0.06572265625, + "grad_norm": 0.45954084396362305, + "learning_rate": 0.0004970234483246053, + "loss": 4.9375, + "step": 673 + }, + { + "epoch": 0.0658203125, + "grad_norm": 0.5340402126312256, + "learning_rate": 0.0004970120801197964, + "loss": 4.9219, + "step": 674 + }, + { + "epoch": 0.06591796875, + "grad_norm": 0.5017228722572327, + "learning_rate": 0.0004970006903924231, + "loss": 4.9141, + "step": 675 + }, + { + "epoch": 0.066015625, + "grad_norm": 0.4929012060165405, + "learning_rate": 0.0004969892791435896, + "loss": 4.9062, + "step": 676 + }, + { + "epoch": 0.06611328125, + "grad_norm": 0.5938189625740051, + "learning_rate": 0.0004969778463744021, + "loss": 4.9414, + "step": 677 + }, + { + "epoch": 0.0662109375, + "grad_norm": 0.6704724431037903, + "learning_rate": 0.000496966392085969, + "loss": 4.9297, + "step": 678 + }, + { + "epoch": 0.06630859375, + "grad_norm": 0.6303629875183105, + "learning_rate": 0.0004969549162794007, + "loss": 4.9258, + "step": 679 + }, + { + "epoch": 0.06640625, + "grad_norm": 0.5393560528755188, + "learning_rate": 0.0004969434189558096, + "loss": 4.9258, + "step": 680 + }, + { + "epoch": 0.06650390625, + "grad_norm": 0.4611497223377228, + "learning_rate": 0.0004969319001163104, + "loss": 4.9102, + "step": 681 + }, + { + "epoch": 0.0666015625, + "grad_norm": 0.47802746295928955, + "learning_rate": 0.0004969203597620197, + "loss": 4.9219, + "step": 682 + }, + { + "epoch": 0.06669921875, + "grad_norm": 0.5258490443229675, + "learning_rate": 0.0004969087978940564, + "loss": 4.9492, + "step": 683 + }, + { + "epoch": 0.066796875, + "grad_norm": 0.5325090289115906, + "learning_rate": 0.0004968972145135412, + "loss": 4.9727, + "step": 684 + }, + { + "epoch": 0.06689453125, + "grad_norm": 0.5076428651809692, + "learning_rate": 0.0004968856096215971, + "loss": 4.8984, + "step": 685 + }, + { + "epoch": 0.0669921875, + "grad_norm": 0.5641031861305237, + "learning_rate": 0.000496873983219349, + "loss": 4.9414, + "step": 686 + }, + { + "epoch": 0.06708984375, + "grad_norm": 0.7779815793037415, + "learning_rate": 0.0004968623353079242, + "loss": 4.9102, + "step": 687 + }, + { + "epoch": 0.0671875, + "grad_norm": 0.618226170539856, + "learning_rate": 0.0004968506658884517, + "loss": 4.9336, + "step": 688 + }, + { + "epoch": 0.06728515625, + "grad_norm": 0.6069994568824768, + "learning_rate": 0.0004968389749620629, + "loss": 4.9453, + "step": 689 + }, + { + "epoch": 0.0673828125, + "grad_norm": 0.602695643901825, + "learning_rate": 0.000496827262529891, + "loss": 4.9141, + "step": 690 + }, + { + "epoch": 0.06748046875, + "grad_norm": 0.8353970050811768, + "learning_rate": 0.0004968155285930717, + "loss": 4.957, + "step": 691 + }, + { + "epoch": 0.067578125, + "grad_norm": 0.7601311206817627, + "learning_rate": 0.0004968037731527422, + "loss": 4.8867, + "step": 692 + }, + { + "epoch": 0.06767578125, + "grad_norm": 0.8417662978172302, + "learning_rate": 0.0004967919962100424, + "loss": 4.918, + "step": 693 + }, + { + "epoch": 0.0677734375, + "grad_norm": 0.864201009273529, + "learning_rate": 0.0004967801977661138, + "loss": 4.9336, + "step": 694 + }, + { + "epoch": 0.06787109375, + "grad_norm": 0.6699989438056946, + "learning_rate": 0.0004967683778221003, + "loss": 4.9531, + "step": 695 + }, + { + "epoch": 0.06796875, + "grad_norm": 0.5989636778831482, + "learning_rate": 0.0004967565363791478, + "loss": 4.9023, + "step": 696 + }, + { + "epoch": 0.06806640625, + "grad_norm": 0.5926920771598816, + "learning_rate": 0.000496744673438404, + "loss": 4.9141, + "step": 697 + }, + { + "epoch": 0.0681640625, + "grad_norm": 0.6211676597595215, + "learning_rate": 0.0004967327890010192, + "loss": 4.8984, + "step": 698 + }, + { + "epoch": 0.06826171875, + "grad_norm": 0.985095739364624, + "learning_rate": 0.0004967208830681454, + "loss": 4.9102, + "step": 699 + }, + { + "epoch": 0.068359375, + "grad_norm": 0.8903228044509888, + "learning_rate": 0.0004967089556409367, + "loss": 4.918, + "step": 700 + }, + { + "epoch": 0.06845703125, + "grad_norm": 0.7326058745384216, + "learning_rate": 0.0004966970067205496, + "loss": 4.9219, + "step": 701 + }, + { + "epoch": 0.0685546875, + "grad_norm": 0.5591720938682556, + "learning_rate": 0.0004966850363081423, + "loss": 4.9062, + "step": 702 + }, + { + "epoch": 0.06865234375, + "grad_norm": 0.5330259799957275, + "learning_rate": 0.0004966730444048754, + "loss": 4.9258, + "step": 703 + }, + { + "epoch": 0.06875, + "grad_norm": 0.5547099113464355, + "learning_rate": 0.0004966610310119113, + "loss": 4.9141, + "step": 704 + }, + { + "epoch": 0.06884765625, + "grad_norm": 0.5156508088111877, + "learning_rate": 0.0004966489961304147, + "loss": 4.918, + "step": 705 + }, + { + "epoch": 0.0689453125, + "grad_norm": 0.5710839629173279, + "learning_rate": 0.0004966369397615522, + "loss": 4.875, + "step": 706 + }, + { + "epoch": 0.06904296875, + "grad_norm": 0.6018747091293335, + "learning_rate": 0.0004966248619064927, + "loss": 4.9102, + "step": 707 + }, + { + "epoch": 0.069140625, + "grad_norm": 0.7630068063735962, + "learning_rate": 0.000496612762566407, + "loss": 4.9375, + "step": 708 + }, + { + "epoch": 0.06923828125, + "grad_norm": 0.9202441573143005, + "learning_rate": 0.000496600641742468, + "loss": 4.9375, + "step": 709 + }, + { + "epoch": 0.0693359375, + "grad_norm": 0.801213800907135, + "learning_rate": 0.0004965884994358508, + "loss": 4.9102, + "step": 710 + }, + { + "epoch": 0.06943359375, + "grad_norm": 0.6991446614265442, + "learning_rate": 0.0004965763356477326, + "loss": 4.9219, + "step": 711 + }, + { + "epoch": 0.06953125, + "grad_norm": 0.6677018404006958, + "learning_rate": 0.0004965641503792924, + "loss": 4.8984, + "step": 712 + }, + { + "epoch": 0.06962890625, + "grad_norm": 0.5393925905227661, + "learning_rate": 0.0004965519436317115, + "loss": 4.9141, + "step": 713 + }, + { + "epoch": 0.0697265625, + "grad_norm": 0.4885493814945221, + "learning_rate": 0.0004965397154061736, + "loss": 4.9102, + "step": 714 + }, + { + "epoch": 0.06982421875, + "grad_norm": 0.42598259449005127, + "learning_rate": 0.0004965274657038637, + "loss": 4.8867, + "step": 715 + }, + { + "epoch": 0.069921875, + "grad_norm": 0.3984520733356476, + "learning_rate": 0.0004965151945259696, + "loss": 4.918, + "step": 716 + }, + { + "epoch": 0.07001953125, + "grad_norm": 0.4602973163127899, + "learning_rate": 0.0004965029018736807, + "loss": 4.9336, + "step": 717 + }, + { + "epoch": 0.0701171875, + "grad_norm": 0.4962187111377716, + "learning_rate": 0.0004964905877481889, + "loss": 4.9141, + "step": 718 + }, + { + "epoch": 0.07021484375, + "grad_norm": 0.572699785232544, + "learning_rate": 0.0004964782521506879, + "loss": 4.9219, + "step": 719 + }, + { + "epoch": 0.0703125, + "grad_norm": 0.6187204718589783, + "learning_rate": 0.0004964658950823734, + "loss": 4.9219, + "step": 720 + }, + { + "epoch": 0.07041015625, + "grad_norm": 0.5086959600448608, + "learning_rate": 0.0004964535165444436, + "loss": 4.8945, + "step": 721 + }, + { + "epoch": 0.0705078125, + "grad_norm": 0.4606127142906189, + "learning_rate": 0.0004964411165380983, + "loss": 4.9102, + "step": 722 + }, + { + "epoch": 0.07060546875, + "grad_norm": 0.4611072242259979, + "learning_rate": 0.0004964286950645397, + "loss": 4.8945, + "step": 723 + }, + { + "epoch": 0.070703125, + "grad_norm": 0.4993979036808014, + "learning_rate": 0.000496416252124972, + "loss": 4.9023, + "step": 724 + }, + { + "epoch": 0.07080078125, + "grad_norm": 0.5374334454536438, + "learning_rate": 0.0004964037877206014, + "loss": 4.9023, + "step": 725 + }, + { + "epoch": 0.0708984375, + "grad_norm": 0.5764836668968201, + "learning_rate": 0.0004963913018526363, + "loss": 4.8789, + "step": 726 + }, + { + "epoch": 0.07099609375, + "grad_norm": 0.8342440724372864, + "learning_rate": 0.000496378794522287, + "loss": 4.8945, + "step": 727 + }, + { + "epoch": 0.07109375, + "grad_norm": 0.8987082839012146, + "learning_rate": 0.0004963662657307661, + "loss": 4.9375, + "step": 728 + }, + { + "epoch": 0.07119140625, + "grad_norm": 0.7202743291854858, + "learning_rate": 0.0004963537154792881, + "loss": 4.918, + "step": 729 + }, + { + "epoch": 0.0712890625, + "grad_norm": 0.8998580574989319, + "learning_rate": 0.0004963411437690696, + "loss": 4.8867, + "step": 730 + }, + { + "epoch": 0.07138671875, + "grad_norm": 0.808499813079834, + "learning_rate": 0.0004963285506013297, + "loss": 4.9414, + "step": 731 + }, + { + "epoch": 0.071484375, + "grad_norm": 0.5454872250556946, + "learning_rate": 0.0004963159359772889, + "loss": 4.9219, + "step": 732 + }, + { + "epoch": 0.07158203125, + "grad_norm": 0.4970705807209015, + "learning_rate": 0.0004963032998981702, + "loss": 4.918, + "step": 733 + }, + { + "epoch": 0.0716796875, + "grad_norm": 0.4672599732875824, + "learning_rate": 0.0004962906423651985, + "loss": 4.9102, + "step": 734 + }, + { + "epoch": 0.07177734375, + "grad_norm": 0.4656890630722046, + "learning_rate": 0.000496277963379601, + "loss": 4.9062, + "step": 735 + }, + { + "epoch": 0.071875, + "grad_norm": 0.4242008626461029, + "learning_rate": 0.0004962652629426068, + "loss": 4.9023, + "step": 736 + }, + { + "epoch": 0.07197265625, + "grad_norm": 0.38737159967422485, + "learning_rate": 0.000496252541055447, + "loss": 4.9102, + "step": 737 + }, + { + "epoch": 0.0720703125, + "grad_norm": 0.4093025326728821, + "learning_rate": 0.000496239797719355, + "loss": 4.8984, + "step": 738 + }, + { + "epoch": 0.07216796875, + "grad_norm": 0.4339083135128021, + "learning_rate": 0.0004962270329355662, + "loss": 4.9414, + "step": 739 + }, + { + "epoch": 0.072265625, + "grad_norm": 0.5110611915588379, + "learning_rate": 0.000496214246705318, + "loss": 4.9141, + "step": 740 + }, + { + "epoch": 0.07236328125, + "grad_norm": 0.5828584432601929, + "learning_rate": 0.00049620143902985, + "loss": 4.918, + "step": 741 + }, + { + "epoch": 0.0724609375, + "grad_norm": 0.5776299834251404, + "learning_rate": 0.0004961886099104038, + "loss": 4.8945, + "step": 742 + }, + { + "epoch": 0.07255859375, + "grad_norm": 0.48184171319007874, + "learning_rate": 0.0004961757593482229, + "loss": 4.8945, + "step": 743 + }, + { + "epoch": 0.07265625, + "grad_norm": 0.4601769745349884, + "learning_rate": 0.0004961628873445535, + "loss": 4.9023, + "step": 744 + }, + { + "epoch": 0.07275390625, + "grad_norm": 0.44548672437667847, + "learning_rate": 0.0004961499939006431, + "loss": 4.9141, + "step": 745 + }, + { + "epoch": 0.0728515625, + "grad_norm": 0.5392785668373108, + "learning_rate": 0.0004961370790177418, + "loss": 4.8945, + "step": 746 + }, + { + "epoch": 0.07294921875, + "grad_norm": 0.5730900764465332, + "learning_rate": 0.0004961241426971014, + "loss": 4.9258, + "step": 747 + }, + { + "epoch": 0.073046875, + "grad_norm": 0.6566439270973206, + "learning_rate": 0.0004961111849399763, + "loss": 4.9102, + "step": 748 + }, + { + "epoch": 0.07314453125, + "grad_norm": 0.6477753520011902, + "learning_rate": 0.0004960982057476224, + "loss": 4.918, + "step": 749 + }, + { + "epoch": 0.0732421875, + "grad_norm": 0.5533462166786194, + "learning_rate": 0.0004960852051212982, + "loss": 4.8711, + "step": 750 + }, + { + "epoch": 0.07333984375, + "grad_norm": 0.5516164302825928, + "learning_rate": 0.0004960721830622637, + "loss": 4.8828, + "step": 751 + }, + { + "epoch": 0.0734375, + "grad_norm": 0.5473514795303345, + "learning_rate": 0.0004960591395717816, + "loss": 4.9297, + "step": 752 + }, + { + "epoch": 0.07353515625, + "grad_norm": 0.5541347861289978, + "learning_rate": 0.0004960460746511162, + "loss": 4.8906, + "step": 753 + }, + { + "epoch": 0.0736328125, + "grad_norm": 0.5999252200126648, + "learning_rate": 0.0004960329883015341, + "loss": 4.9023, + "step": 754 + }, + { + "epoch": 0.07373046875, + "grad_norm": 0.5422800779342651, + "learning_rate": 0.0004960198805243039, + "loss": 4.9023, + "step": 755 + }, + { + "epoch": 0.073828125, + "grad_norm": 0.4828493595123291, + "learning_rate": 0.0004960067513206964, + "loss": 4.8555, + "step": 756 + }, + { + "epoch": 0.07392578125, + "grad_norm": 0.44730255007743835, + "learning_rate": 0.0004959936006919843, + "loss": 4.9141, + "step": 757 + }, + { + "epoch": 0.0740234375, + "grad_norm": 0.5136485695838928, + "learning_rate": 0.0004959804286394425, + "loss": 4.9062, + "step": 758 + }, + { + "epoch": 0.07412109375, + "grad_norm": 0.558743417263031, + "learning_rate": 0.000495967235164348, + "loss": 4.9297, + "step": 759 + }, + { + "epoch": 0.07421875, + "grad_norm": 0.503806471824646, + "learning_rate": 0.0004959540202679797, + "loss": 4.9102, + "step": 760 + }, + { + "epoch": 0.07431640625, + "grad_norm": 0.4152722954750061, + "learning_rate": 0.0004959407839516188, + "loss": 4.8711, + "step": 761 + }, + { + "epoch": 0.0744140625, + "grad_norm": 0.38589704036712646, + "learning_rate": 0.0004959275262165485, + "loss": 4.8906, + "step": 762 + }, + { + "epoch": 0.07451171875, + "grad_norm": 0.43141141533851624, + "learning_rate": 0.0004959142470640539, + "loss": 4.8516, + "step": 763 + }, + { + "epoch": 0.074609375, + "grad_norm": 0.42636358737945557, + "learning_rate": 0.0004959009464954224, + "loss": 4.8789, + "step": 764 + }, + { + "epoch": 0.07470703125, + "grad_norm": 0.4754694700241089, + "learning_rate": 0.0004958876245119433, + "loss": 4.8828, + "step": 765 + }, + { + "epoch": 0.0748046875, + "grad_norm": 0.4849173128604889, + "learning_rate": 0.0004958742811149083, + "loss": 4.9062, + "step": 766 + }, + { + "epoch": 0.07490234375, + "grad_norm": 0.5047995448112488, + "learning_rate": 0.0004958609163056108, + "loss": 4.8828, + "step": 767 + }, + { + "epoch": 0.075, + "grad_norm": 0.5260794758796692, + "learning_rate": 0.0004958475300853464, + "loss": 4.8828, + "step": 768 + }, + { + "epoch": 0.07509765625, + "grad_norm": 0.4804491102695465, + "learning_rate": 0.0004958341224554129, + "loss": 4.8906, + "step": 769 + }, + { + "epoch": 0.0751953125, + "grad_norm": 0.4267667233943939, + "learning_rate": 0.00049582069341711, + "loss": 4.875, + "step": 770 + }, + { + "epoch": 0.07529296875, + "grad_norm": 0.460480660200119, + "learning_rate": 0.0004958072429717395, + "loss": 4.8789, + "step": 771 + }, + { + "epoch": 0.075390625, + "grad_norm": 0.4774636924266815, + "learning_rate": 0.0004957937711206055, + "loss": 4.8945, + "step": 772 + }, + { + "epoch": 0.07548828125, + "grad_norm": 0.44169384241104126, + "learning_rate": 0.000495780277865014, + "loss": 4.875, + "step": 773 + }, + { + "epoch": 0.0755859375, + "grad_norm": 0.4531261622905731, + "learning_rate": 0.000495766763206273, + "loss": 4.8945, + "step": 774 + }, + { + "epoch": 0.07568359375, + "grad_norm": 0.49369335174560547, + "learning_rate": 0.0004957532271456926, + "loss": 4.9023, + "step": 775 + }, + { + "epoch": 0.07578125, + "grad_norm": 0.5407163500785828, + "learning_rate": 0.000495739669684585, + "loss": 4.9023, + "step": 776 + }, + { + "epoch": 0.07587890625, + "grad_norm": 0.6588068008422852, + "learning_rate": 0.0004957260908242647, + "loss": 4.8906, + "step": 777 + }, + { + "epoch": 0.0759765625, + "grad_norm": 0.708760142326355, + "learning_rate": 0.000495712490566048, + "loss": 4.8828, + "step": 778 + }, + { + "epoch": 0.07607421875, + "grad_norm": 0.6345651745796204, + "learning_rate": 0.0004956988689112533, + "loss": 4.8984, + "step": 779 + }, + { + "epoch": 0.076171875, + "grad_norm": 0.5177327990531921, + "learning_rate": 0.0004956852258612011, + "loss": 4.8906, + "step": 780 + }, + { + "epoch": 0.07626953125, + "grad_norm": 0.5784163475036621, + "learning_rate": 0.0004956715614172141, + "loss": 4.8906, + "step": 781 + }, + { + "epoch": 0.0763671875, + "grad_norm": 0.4898063540458679, + "learning_rate": 0.0004956578755806168, + "loss": 4.8945, + "step": 782 + }, + { + "epoch": 0.07646484375, + "grad_norm": 0.43215012550354004, + "learning_rate": 0.0004956441683527361, + "loss": 4.9219, + "step": 783 + }, + { + "epoch": 0.0765625, + "grad_norm": 0.42248815298080444, + "learning_rate": 0.0004956304397349009, + "loss": 4.8711, + "step": 784 + }, + { + "epoch": 0.07666015625, + "grad_norm": 0.4588373303413391, + "learning_rate": 0.0004956166897284419, + "loss": 4.8828, + "step": 785 + }, + { + "epoch": 0.0767578125, + "grad_norm": 0.5150899887084961, + "learning_rate": 0.0004956029183346922, + "loss": 4.875, + "step": 786 + }, + { + "epoch": 0.07685546875, + "grad_norm": 0.49109452962875366, + "learning_rate": 0.0004955891255549868, + "loss": 4.8828, + "step": 787 + }, + { + "epoch": 0.076953125, + "grad_norm": 0.5134442448616028, + "learning_rate": 0.0004955753113906629, + "loss": 4.8789, + "step": 788 + }, + { + "epoch": 0.07705078125, + "grad_norm": 0.5372751951217651, + "learning_rate": 0.0004955614758430594, + "loss": 4.8711, + "step": 789 + }, + { + "epoch": 0.0771484375, + "grad_norm": 0.5520480871200562, + "learning_rate": 0.0004955476189135179, + "loss": 4.875, + "step": 790 + }, + { + "epoch": 0.07724609375, + "grad_norm": 0.5645185112953186, + "learning_rate": 0.0004955337406033817, + "loss": 4.8555, + "step": 791 + }, + { + "epoch": 0.07734375, + "grad_norm": 0.5430117249488831, + "learning_rate": 0.000495519840913996, + "loss": 4.8906, + "step": 792 + }, + { + "epoch": 0.07744140625, + "grad_norm": 0.5487677454948425, + "learning_rate": 0.0004955059198467085, + "loss": 4.8906, + "step": 793 + }, + { + "epoch": 0.0775390625, + "grad_norm": 0.5324546098709106, + "learning_rate": 0.0004954919774028685, + "loss": 4.8672, + "step": 794 + }, + { + "epoch": 0.07763671875, + "grad_norm": 0.48905646800994873, + "learning_rate": 0.0004954780135838278, + "loss": 4.8906, + "step": 795 + }, + { + "epoch": 0.077734375, + "grad_norm": 0.49760356545448303, + "learning_rate": 0.0004954640283909401, + "loss": 4.8984, + "step": 796 + }, + { + "epoch": 0.07783203125, + "grad_norm": 0.5531620979309082, + "learning_rate": 0.0004954500218255613, + "loss": 4.9062, + "step": 797 + }, + { + "epoch": 0.0779296875, + "grad_norm": 0.5735150575637817, + "learning_rate": 0.0004954359938890489, + "loss": 4.8594, + "step": 798 + }, + { + "epoch": 0.07802734375, + "grad_norm": 0.5884649753570557, + "learning_rate": 0.000495421944582763, + "loss": 4.918, + "step": 799 + }, + { + "epoch": 0.078125, + "grad_norm": 0.48785004019737244, + "learning_rate": 0.0004954078739080656, + "loss": 4.9102, + "step": 800 + }, + { + "epoch": 0.07822265625, + "grad_norm": 0.47542282938957214, + "learning_rate": 0.0004953937818663208, + "loss": 4.8789, + "step": 801 + }, + { + "epoch": 0.0783203125, + "grad_norm": 0.5039560794830322, + "learning_rate": 0.0004953796684588946, + "loss": 4.8906, + "step": 802 + }, + { + "epoch": 0.07841796875, + "grad_norm": 0.47064536809921265, + "learning_rate": 0.0004953655336871553, + "loss": 4.8594, + "step": 803 + }, + { + "epoch": 0.078515625, + "grad_norm": 0.44826674461364746, + "learning_rate": 0.0004953513775524731, + "loss": 4.8633, + "step": 804 + }, + { + "epoch": 0.07861328125, + "grad_norm": 0.403845876455307, + "learning_rate": 0.0004953372000562204, + "loss": 4.8984, + "step": 805 + }, + { + "epoch": 0.0787109375, + "grad_norm": 0.4083589017391205, + "learning_rate": 0.0004953230011997716, + "loss": 4.9141, + "step": 806 + }, + { + "epoch": 0.07880859375, + "grad_norm": 0.3878018260002136, + "learning_rate": 0.0004953087809845031, + "loss": 4.875, + "step": 807 + }, + { + "epoch": 0.07890625, + "grad_norm": 0.39088737964630127, + "learning_rate": 0.0004952945394117936, + "loss": 4.9336, + "step": 808 + }, + { + "epoch": 0.07900390625, + "grad_norm": 0.46204832196235657, + "learning_rate": 0.0004952802764830236, + "loss": 4.8477, + "step": 809 + }, + { + "epoch": 0.0791015625, + "grad_norm": 0.5762801766395569, + "learning_rate": 0.0004952659921995758, + "loss": 4.8945, + "step": 810 + }, + { + "epoch": 0.07919921875, + "grad_norm": 0.6087040305137634, + "learning_rate": 0.0004952516865628352, + "loss": 4.8945, + "step": 811 + }, + { + "epoch": 0.079296875, + "grad_norm": 0.6062765717506409, + "learning_rate": 0.0004952373595741883, + "loss": 4.8711, + "step": 812 + }, + { + "epoch": 0.07939453125, + "grad_norm": 0.6468778252601624, + "learning_rate": 0.0004952230112350241, + "loss": 4.8633, + "step": 813 + }, + { + "epoch": 0.0794921875, + "grad_norm": 0.6803391575813293, + "learning_rate": 0.0004952086415467337, + "loss": 4.8945, + "step": 814 + }, + { + "epoch": 0.07958984375, + "grad_norm": 0.6036785244941711, + "learning_rate": 0.0004951942505107101, + "loss": 4.8633, + "step": 815 + }, + { + "epoch": 0.0796875, + "grad_norm": 0.5079777240753174, + "learning_rate": 0.0004951798381283482, + "loss": 4.8555, + "step": 816 + }, + { + "epoch": 0.07978515625, + "grad_norm": 0.47106999158859253, + "learning_rate": 0.0004951654044010455, + "loss": 4.832, + "step": 817 + }, + { + "epoch": 0.0798828125, + "grad_norm": 0.418499618768692, + "learning_rate": 0.0004951509493302011, + "loss": 4.8711, + "step": 818 + }, + { + "epoch": 0.07998046875, + "grad_norm": 0.3655601143836975, + "learning_rate": 0.0004951364729172163, + "loss": 4.8516, + "step": 819 + }, + { + "epoch": 0.080078125, + "grad_norm": 0.42946258187294006, + "learning_rate": 0.0004951219751634945, + "loss": 4.8789, + "step": 820 + }, + { + "epoch": 0.08017578125, + "grad_norm": 0.412626177072525, + "learning_rate": 0.0004951074560704412, + "loss": 4.8828, + "step": 821 + }, + { + "epoch": 0.0802734375, + "grad_norm": 0.44532856345176697, + "learning_rate": 0.0004950929156394639, + "loss": 4.8828, + "step": 822 + }, + { + "epoch": 0.08037109375, + "grad_norm": 0.4855706989765167, + "learning_rate": 0.0004950783538719723, + "loss": 4.9062, + "step": 823 + }, + { + "epoch": 0.08046875, + "grad_norm": 0.504410982131958, + "learning_rate": 0.0004950637707693779, + "loss": 4.8945, + "step": 824 + }, + { + "epoch": 0.08056640625, + "grad_norm": 0.5246967077255249, + "learning_rate": 0.0004950491663330946, + "loss": 4.875, + "step": 825 + }, + { + "epoch": 0.0806640625, + "grad_norm": 0.5594045519828796, + "learning_rate": 0.000495034540564538, + "loss": 4.8867, + "step": 826 + }, + { + "epoch": 0.08076171875, + "grad_norm": 0.5870455503463745, + "learning_rate": 0.000495019893465126, + "loss": 4.8789, + "step": 827 + }, + { + "epoch": 0.080859375, + "grad_norm": 0.5400936007499695, + "learning_rate": 0.0004950052250362786, + "loss": 4.8906, + "step": 828 + }, + { + "epoch": 0.08095703125, + "grad_norm": 0.4675130546092987, + "learning_rate": 0.000494990535279418, + "loss": 4.8789, + "step": 829 + }, + { + "epoch": 0.0810546875, + "grad_norm": 0.45326483249664307, + "learning_rate": 0.0004949758241959679, + "loss": 4.8906, + "step": 830 + }, + { + "epoch": 0.08115234375, + "grad_norm": 0.43965944647789, + "learning_rate": 0.0004949610917873547, + "loss": 4.8555, + "step": 831 + }, + { + "epoch": 0.08125, + "grad_norm": 0.5271297693252563, + "learning_rate": 0.0004949463380550065, + "loss": 4.8398, + "step": 832 + }, + { + "epoch": 0.08134765625, + "grad_norm": 0.6631085872650146, + "learning_rate": 0.0004949315630003537, + "loss": 4.8711, + "step": 833 + }, + { + "epoch": 0.0814453125, + "grad_norm": 0.5798559188842773, + "learning_rate": 0.0004949167666248285, + "loss": 4.8672, + "step": 834 + }, + { + "epoch": 0.08154296875, + "grad_norm": 0.45223212242126465, + "learning_rate": 0.0004949019489298653, + "loss": 4.8555, + "step": 835 + }, + { + "epoch": 0.081640625, + "grad_norm": 0.4504897892475128, + "learning_rate": 0.0004948871099169006, + "loss": 4.8555, + "step": 836 + }, + { + "epoch": 0.08173828125, + "grad_norm": 0.4103120267391205, + "learning_rate": 0.0004948722495873732, + "loss": 4.8906, + "step": 837 + }, + { + "epoch": 0.0818359375, + "grad_norm": 0.4391990602016449, + "learning_rate": 0.0004948573679427233, + "loss": 4.8633, + "step": 838 + }, + { + "epoch": 0.08193359375, + "grad_norm": 0.5236042141914368, + "learning_rate": 0.0004948424649843938, + "loss": 4.8633, + "step": 839 + }, + { + "epoch": 0.08203125, + "grad_norm": 0.6059775948524475, + "learning_rate": 0.0004948275407138293, + "loss": 4.8477, + "step": 840 + }, + { + "epoch": 0.08212890625, + "grad_norm": 0.6779847145080566, + "learning_rate": 0.0004948125951324768, + "loss": 4.8828, + "step": 841 + }, + { + "epoch": 0.0822265625, + "grad_norm": 0.6596094369888306, + "learning_rate": 0.000494797628241785, + "loss": 4.8633, + "step": 842 + }, + { + "epoch": 0.08232421875, + "grad_norm": 0.550703763961792, + "learning_rate": 0.000494782640043205, + "loss": 4.8672, + "step": 843 + }, + { + "epoch": 0.082421875, + "grad_norm": 0.4469245970249176, + "learning_rate": 0.0004947676305381897, + "loss": 4.8984, + "step": 844 + }, + { + "epoch": 0.08251953125, + "grad_norm": 0.4176386892795563, + "learning_rate": 0.0004947525997281941, + "loss": 4.8516, + "step": 845 + }, + { + "epoch": 0.0826171875, + "grad_norm": 0.362447589635849, + "learning_rate": 0.0004947375476146755, + "loss": 4.8633, + "step": 846 + }, + { + "epoch": 0.08271484375, + "grad_norm": 0.39331042766571045, + "learning_rate": 0.000494722474199093, + "loss": 4.8711, + "step": 847 + }, + { + "epoch": 0.0828125, + "grad_norm": 0.37120670080184937, + "learning_rate": 0.0004947073794829079, + "loss": 4.8828, + "step": 848 + }, + { + "epoch": 0.08291015625, + "grad_norm": 0.40028196573257446, + "learning_rate": 0.0004946922634675836, + "loss": 4.8789, + "step": 849 + }, + { + "epoch": 0.0830078125, + "grad_norm": 0.47256696224212646, + "learning_rate": 0.0004946771261545853, + "loss": 4.8828, + "step": 850 + }, + { + "epoch": 0.08310546875, + "grad_norm": 0.5021369457244873, + "learning_rate": 0.0004946619675453806, + "loss": 4.8555, + "step": 851 + }, + { + "epoch": 0.083203125, + "grad_norm": 0.4181872010231018, + "learning_rate": 0.000494646787641439, + "loss": 4.8672, + "step": 852 + }, + { + "epoch": 0.08330078125, + "grad_norm": 0.40219172835350037, + "learning_rate": 0.000494631586444232, + "loss": 4.8672, + "step": 853 + }, + { + "epoch": 0.0833984375, + "grad_norm": 0.42699527740478516, + "learning_rate": 0.0004946163639552335, + "loss": 4.8984, + "step": 854 + }, + { + "epoch": 0.08349609375, + "grad_norm": 0.46114829182624817, + "learning_rate": 0.0004946011201759189, + "loss": 4.832, + "step": 855 + }, + { + "epoch": 0.08359375, + "grad_norm": 0.5074323415756226, + "learning_rate": 0.0004945858551077662, + "loss": 4.8789, + "step": 856 + }, + { + "epoch": 0.08369140625, + "grad_norm": 0.5219331979751587, + "learning_rate": 0.0004945705687522552, + "loss": 4.8438, + "step": 857 + }, + { + "epoch": 0.0837890625, + "grad_norm": 0.6114364266395569, + "learning_rate": 0.0004945552611108679, + "loss": 4.8672, + "step": 858 + }, + { + "epoch": 0.08388671875, + "grad_norm": 0.7798656821250916, + "learning_rate": 0.0004945399321850879, + "loss": 4.8789, + "step": 859 + }, + { + "epoch": 0.083984375, + "grad_norm": 0.7879101634025574, + "learning_rate": 0.0004945245819764016, + "loss": 4.8828, + "step": 860 + }, + { + "epoch": 0.08408203125, + "grad_norm": 0.7016037106513977, + "learning_rate": 0.0004945092104862971, + "loss": 4.875, + "step": 861 + }, + { + "epoch": 0.0841796875, + "grad_norm": 0.49669718742370605, + "learning_rate": 0.0004944938177162644, + "loss": 4.8359, + "step": 862 + }, + { + "epoch": 0.08427734375, + "grad_norm": 0.4770626425743103, + "learning_rate": 0.0004944784036677958, + "loss": 4.8594, + "step": 863 + }, + { + "epoch": 0.084375, + "grad_norm": 0.4248046278953552, + "learning_rate": 0.0004944629683423855, + "loss": 4.8867, + "step": 864 + }, + { + "epoch": 0.08447265625, + "grad_norm": 0.4632871747016907, + "learning_rate": 0.00049444751174153, + "loss": 4.8594, + "step": 865 + }, + { + "epoch": 0.0845703125, + "grad_norm": 0.48169225454330444, + "learning_rate": 0.0004944320338667276, + "loss": 4.8633, + "step": 866 + }, + { + "epoch": 0.08466796875, + "grad_norm": 0.42320355772972107, + "learning_rate": 0.0004944165347194788, + "loss": 4.8867, + "step": 867 + }, + { + "epoch": 0.084765625, + "grad_norm": 0.39030560851097107, + "learning_rate": 0.0004944010143012861, + "loss": 4.8828, + "step": 868 + }, + { + "epoch": 0.08486328125, + "grad_norm": 0.4671940207481384, + "learning_rate": 0.0004943854726136542, + "loss": 4.8203, + "step": 869 + }, + { + "epoch": 0.0849609375, + "grad_norm": 0.4762789309024811, + "learning_rate": 0.0004943699096580897, + "loss": 4.8906, + "step": 870 + }, + { + "epoch": 0.08505859375, + "grad_norm": 0.5251975655555725, + "learning_rate": 0.0004943543254361013, + "loss": 4.8672, + "step": 871 + }, + { + "epoch": 0.08515625, + "grad_norm": 0.5693690180778503, + "learning_rate": 0.0004943387199491998, + "loss": 4.9102, + "step": 872 + }, + { + "epoch": 0.08525390625, + "grad_norm": 0.5571799278259277, + "learning_rate": 0.0004943230931988981, + "loss": 4.8672, + "step": 873 + }, + { + "epoch": 0.0853515625, + "grad_norm": 0.4701318144798279, + "learning_rate": 0.0004943074451867111, + "loss": 4.8789, + "step": 874 + }, + { + "epoch": 0.08544921875, + "grad_norm": 0.43155744671821594, + "learning_rate": 0.0004942917759141556, + "loss": 4.8555, + "step": 875 + }, + { + "epoch": 0.085546875, + "grad_norm": 0.43980658054351807, + "learning_rate": 0.0004942760853827509, + "loss": 4.8789, + "step": 876 + }, + { + "epoch": 0.08564453125, + "grad_norm": 0.46005746722221375, + "learning_rate": 0.0004942603735940179, + "loss": 4.8867, + "step": 877 + }, + { + "epoch": 0.0857421875, + "grad_norm": 0.4600110948085785, + "learning_rate": 0.0004942446405494798, + "loss": 4.8633, + "step": 878 + }, + { + "epoch": 0.08583984375, + "grad_norm": 0.40931013226509094, + "learning_rate": 0.0004942288862506618, + "loss": 4.8516, + "step": 879 + }, + { + "epoch": 0.0859375, + "grad_norm": 0.4222780466079712, + "learning_rate": 0.0004942131106990911, + "loss": 4.8789, + "step": 880 + }, + { + "epoch": 0.08603515625, + "grad_norm": 0.44859063625335693, + "learning_rate": 0.0004941973138962973, + "loss": 4.8672, + "step": 881 + }, + { + "epoch": 0.0861328125, + "grad_norm": 0.48440831899642944, + "learning_rate": 0.0004941814958438115, + "loss": 4.8555, + "step": 882 + }, + { + "epoch": 0.08623046875, + "grad_norm": 0.525878369808197, + "learning_rate": 0.0004941656565431673, + "loss": 4.8281, + "step": 883 + }, + { + "epoch": 0.086328125, + "grad_norm": 0.6066216230392456, + "learning_rate": 0.0004941497959959003, + "loss": 4.8477, + "step": 884 + }, + { + "epoch": 0.08642578125, + "grad_norm": 0.6154394149780273, + "learning_rate": 0.0004941339142035478, + "loss": 4.875, + "step": 885 + }, + { + "epoch": 0.0865234375, + "grad_norm": 0.5415538549423218, + "learning_rate": 0.0004941180111676497, + "loss": 4.8359, + "step": 886 + }, + { + "epoch": 0.08662109375, + "grad_norm": 0.43851184844970703, + "learning_rate": 0.0004941020868897474, + "loss": 4.8477, + "step": 887 + }, + { + "epoch": 0.08671875, + "grad_norm": 0.4420389235019684, + "learning_rate": 0.0004940861413713849, + "loss": 4.8516, + "step": 888 + }, + { + "epoch": 0.08681640625, + "grad_norm": 0.49204596877098083, + "learning_rate": 0.000494070174614108, + "loss": 4.8516, + "step": 889 + }, + { + "epoch": 0.0869140625, + "grad_norm": 0.4975183308124542, + "learning_rate": 0.0004940541866194645, + "loss": 4.875, + "step": 890 + }, + { + "epoch": 0.08701171875, + "grad_norm": 0.4811098873615265, + "learning_rate": 0.0004940381773890043, + "loss": 4.8555, + "step": 891 + }, + { + "epoch": 0.087109375, + "grad_norm": 0.5079295039176941, + "learning_rate": 0.0004940221469242794, + "loss": 4.8633, + "step": 892 + }, + { + "epoch": 0.08720703125, + "grad_norm": 0.5155340433120728, + "learning_rate": 0.000494006095226844, + "loss": 4.8828, + "step": 893 + }, + { + "epoch": 0.0873046875, + "grad_norm": 0.47939589619636536, + "learning_rate": 0.0004939900222982539, + "loss": 4.8164, + "step": 894 + }, + { + "epoch": 0.08740234375, + "grad_norm": 0.5215303897857666, + "learning_rate": 0.0004939739281400674, + "loss": 4.8633, + "step": 895 + }, + { + "epoch": 0.0875, + "grad_norm": 0.48365482687950134, + "learning_rate": 0.0004939578127538449, + "loss": 4.8711, + "step": 896 + }, + { + "epoch": 0.08759765625, + "grad_norm": 0.414248526096344, + "learning_rate": 0.0004939416761411484, + "loss": 4.8594, + "step": 897 + }, + { + "epoch": 0.0876953125, + "grad_norm": 0.4160768389701843, + "learning_rate": 0.0004939255183035424, + "loss": 4.8633, + "step": 898 + }, + { + "epoch": 0.08779296875, + "grad_norm": 0.4438508152961731, + "learning_rate": 0.0004939093392425933, + "loss": 4.8398, + "step": 899 + }, + { + "epoch": 0.087890625, + "grad_norm": 0.5406934022903442, + "learning_rate": 0.0004938931389598695, + "loss": 4.832, + "step": 900 + }, + { + "epoch": 0.08798828125, + "grad_norm": 0.6314405202865601, + "learning_rate": 0.0004938769174569413, + "loss": 4.8672, + "step": 901 + }, + { + "epoch": 0.0880859375, + "grad_norm": 0.6316208839416504, + "learning_rate": 0.0004938606747353818, + "loss": 4.8711, + "step": 902 + }, + { + "epoch": 0.08818359375, + "grad_norm": 0.49028605222702026, + "learning_rate": 0.0004938444107967651, + "loss": 4.8398, + "step": 903 + }, + { + "epoch": 0.08828125, + "grad_norm": 0.40383410453796387, + "learning_rate": 0.0004938281256426681, + "loss": 4.8281, + "step": 904 + }, + { + "epoch": 0.08837890625, + "grad_norm": 0.4249477982521057, + "learning_rate": 0.0004938118192746695, + "loss": 4.8633, + "step": 905 + }, + { + "epoch": 0.0884765625, + "grad_norm": 0.4824749827384949, + "learning_rate": 0.0004937954916943502, + "loss": 4.8164, + "step": 906 + }, + { + "epoch": 0.08857421875, + "grad_norm": 0.5105504989624023, + "learning_rate": 0.0004937791429032929, + "loss": 4.8477, + "step": 907 + }, + { + "epoch": 0.088671875, + "grad_norm": 0.4530431032180786, + "learning_rate": 0.0004937627729030825, + "loss": 4.8672, + "step": 908 + }, + { + "epoch": 0.08876953125, + "grad_norm": 0.37337881326675415, + "learning_rate": 0.0004937463816953061, + "loss": 4.875, + "step": 909 + }, + { + "epoch": 0.0888671875, + "grad_norm": 0.3546503484249115, + "learning_rate": 0.0004937299692815525, + "loss": 4.8555, + "step": 910 + }, + { + "epoch": 0.08896484375, + "grad_norm": 0.3876660466194153, + "learning_rate": 0.000493713535663413, + "loss": 4.8594, + "step": 911 + }, + { + "epoch": 0.0890625, + "grad_norm": 0.4789651036262512, + "learning_rate": 0.0004936970808424807, + "loss": 4.8477, + "step": 912 + }, + { + "epoch": 0.08916015625, + "grad_norm": 0.5593496561050415, + "learning_rate": 0.0004936806048203506, + "loss": 4.8711, + "step": 913 + }, + { + "epoch": 0.0892578125, + "grad_norm": 0.6056161522865295, + "learning_rate": 0.0004936641075986201, + "loss": 4.8516, + "step": 914 + }, + { + "epoch": 0.08935546875, + "grad_norm": 0.5245595574378967, + "learning_rate": 0.0004936475891788883, + "loss": 4.8164, + "step": 915 + }, + { + "epoch": 0.089453125, + "grad_norm": 0.49999815225601196, + "learning_rate": 0.0004936310495627569, + "loss": 4.832, + "step": 916 + }, + { + "epoch": 0.08955078125, + "grad_norm": 0.6258394718170166, + "learning_rate": 0.0004936144887518291, + "loss": 4.8711, + "step": 917 + }, + { + "epoch": 0.0896484375, + "grad_norm": 0.6583293080329895, + "learning_rate": 0.0004935979067477103, + "loss": 4.8633, + "step": 918 + }, + { + "epoch": 0.08974609375, + "grad_norm": 0.5308080911636353, + "learning_rate": 0.000493581303552008, + "loss": 4.8672, + "step": 919 + }, + { + "epoch": 0.08984375, + "grad_norm": 0.42577865719795227, + "learning_rate": 0.0004935646791663318, + "loss": 4.8555, + "step": 920 + }, + { + "epoch": 0.08994140625, + "grad_norm": 0.3947119414806366, + "learning_rate": 0.0004935480335922935, + "loss": 4.8359, + "step": 921 + }, + { + "epoch": 0.0900390625, + "grad_norm": 0.400806725025177, + "learning_rate": 0.0004935313668315065, + "loss": 4.8359, + "step": 922 + }, + { + "epoch": 0.09013671875, + "grad_norm": 0.3871156573295593, + "learning_rate": 0.0004935146788855868, + "loss": 4.875, + "step": 923 + }, + { + "epoch": 0.090234375, + "grad_norm": 0.3819091022014618, + "learning_rate": 0.000493497969756152, + "loss": 4.8242, + "step": 924 + }, + { + "epoch": 0.09033203125, + "grad_norm": 0.3758518397808075, + "learning_rate": 0.000493481239444822, + "loss": 4.8477, + "step": 925 + }, + { + "epoch": 0.0904296875, + "grad_norm": 0.4341889023780823, + "learning_rate": 0.0004934644879532185, + "loss": 4.832, + "step": 926 + }, + { + "epoch": 0.09052734375, + "grad_norm": 0.44294634461402893, + "learning_rate": 0.0004934477152829658, + "loss": 4.832, + "step": 927 + }, + { + "epoch": 0.090625, + "grad_norm": 0.4348030984401703, + "learning_rate": 0.0004934309214356897, + "loss": 4.8203, + "step": 928 + }, + { + "epoch": 0.09072265625, + "grad_norm": 0.497432142496109, + "learning_rate": 0.0004934141064130181, + "loss": 4.8633, + "step": 929 + }, + { + "epoch": 0.0908203125, + "grad_norm": 0.5283389687538147, + "learning_rate": 0.0004933972702165814, + "loss": 4.8672, + "step": 930 + }, + { + "epoch": 0.09091796875, + "grad_norm": 0.5673421025276184, + "learning_rate": 0.0004933804128480117, + "loss": 4.8359, + "step": 931 + }, + { + "epoch": 0.091015625, + "grad_norm": 0.5137133002281189, + "learning_rate": 0.000493363534308943, + "loss": 4.8359, + "step": 932 + }, + { + "epoch": 0.09111328125, + "grad_norm": 0.40789496898651123, + "learning_rate": 0.0004933466346010117, + "loss": 4.8359, + "step": 933 + }, + { + "epoch": 0.0912109375, + "grad_norm": 0.4613848328590393, + "learning_rate": 0.0004933297137258561, + "loss": 4.8594, + "step": 934 + }, + { + "epoch": 0.09130859375, + "grad_norm": 0.49260833859443665, + "learning_rate": 0.0004933127716851167, + "loss": 4.8359, + "step": 935 + }, + { + "epoch": 0.09140625, + "grad_norm": 0.4962104856967926, + "learning_rate": 0.0004932958084804356, + "loss": 4.8594, + "step": 936 + }, + { + "epoch": 0.09150390625, + "grad_norm": 0.5404455661773682, + "learning_rate": 0.0004932788241134576, + "loss": 4.8438, + "step": 937 + }, + { + "epoch": 0.0916015625, + "grad_norm": 0.5084996819496155, + "learning_rate": 0.0004932618185858288, + "loss": 4.8633, + "step": 938 + }, + { + "epoch": 0.09169921875, + "grad_norm": 0.4125216603279114, + "learning_rate": 0.0004932447918991983, + "loss": 4.8438, + "step": 939 + }, + { + "epoch": 0.091796875, + "grad_norm": 0.3724025785923004, + "learning_rate": 0.0004932277440552164, + "loss": 4.8359, + "step": 940 + }, + { + "epoch": 0.09189453125, + "grad_norm": 0.4485456645488739, + "learning_rate": 0.0004932106750555358, + "loss": 4.8594, + "step": 941 + }, + { + "epoch": 0.0919921875, + "grad_norm": 0.4383455216884613, + "learning_rate": 0.0004931935849018112, + "loss": 4.8359, + "step": 942 + }, + { + "epoch": 0.09208984375, + "grad_norm": 0.4460577368736267, + "learning_rate": 0.0004931764735956995, + "loss": 4.8516, + "step": 943 + }, + { + "epoch": 0.0921875, + "grad_norm": 0.4385360777378082, + "learning_rate": 0.0004931593411388594, + "loss": 4.8633, + "step": 944 + }, + { + "epoch": 0.09228515625, + "grad_norm": 0.44357937574386597, + "learning_rate": 0.0004931421875329519, + "loss": 4.8398, + "step": 945 + }, + { + "epoch": 0.0923828125, + "grad_norm": 0.5505833625793457, + "learning_rate": 0.0004931250127796398, + "loss": 4.8398, + "step": 946 + }, + { + "epoch": 0.09248046875, + "grad_norm": 0.5369737148284912, + "learning_rate": 0.0004931078168805881, + "loss": 4.8555, + "step": 947 + }, + { + "epoch": 0.092578125, + "grad_norm": 0.5083466172218323, + "learning_rate": 0.0004930905998374639, + "loss": 4.8398, + "step": 948 + }, + { + "epoch": 0.09267578125, + "grad_norm": 0.48525020480155945, + "learning_rate": 0.0004930733616519363, + "loss": 4.8594, + "step": 949 + }, + { + "epoch": 0.0927734375, + "grad_norm": 0.42595598101615906, + "learning_rate": 0.0004930561023256762, + "loss": 4.8359, + "step": 950 + }, + { + "epoch": 0.09287109375, + "grad_norm": 0.40371179580688477, + "learning_rate": 0.000493038821860357, + "loss": 4.8711, + "step": 951 + }, + { + "epoch": 0.09296875, + "grad_norm": 0.4778882563114166, + "learning_rate": 0.0004930215202576539, + "loss": 4.8359, + "step": 952 + }, + { + "epoch": 0.09306640625, + "grad_norm": 0.5457689166069031, + "learning_rate": 0.000493004197519244, + "loss": 4.8242, + "step": 953 + }, + { + "epoch": 0.0931640625, + "grad_norm": 0.4889971613883972, + "learning_rate": 0.0004929868536468069, + "loss": 4.8477, + "step": 954 + }, + { + "epoch": 0.09326171875, + "grad_norm": 0.4244244396686554, + "learning_rate": 0.0004929694886420239, + "loss": 4.8672, + "step": 955 + }, + { + "epoch": 0.093359375, + "grad_norm": 0.44931143522262573, + "learning_rate": 0.0004929521025065782, + "loss": 4.8594, + "step": 956 + }, + { + "epoch": 0.09345703125, + "grad_norm": 0.5113864541053772, + "learning_rate": 0.0004929346952421553, + "loss": 4.8359, + "step": 957 + }, + { + "epoch": 0.0935546875, + "grad_norm": 0.47967684268951416, + "learning_rate": 0.000492917266850443, + "loss": 4.8438, + "step": 958 + }, + { + "epoch": 0.09365234375, + "grad_norm": 0.41450557112693787, + "learning_rate": 0.0004928998173331306, + "loss": 4.8594, + "step": 959 + }, + { + "epoch": 0.09375, + "grad_norm": 0.4654901325702667, + "learning_rate": 0.0004928823466919098, + "loss": 4.8672, + "step": 960 + }, + { + "epoch": 0.09384765625, + "grad_norm": 0.3816904127597809, + "learning_rate": 0.0004928648549284743, + "loss": 4.8164, + "step": 961 + }, + { + "epoch": 0.0939453125, + "grad_norm": 0.37036556005477905, + "learning_rate": 0.0004928473420445198, + "loss": 4.8359, + "step": 962 + }, + { + "epoch": 0.09404296875, + "grad_norm": 0.427219420671463, + "learning_rate": 0.0004928298080417439, + "loss": 4.8516, + "step": 963 + }, + { + "epoch": 0.094140625, + "grad_norm": 0.47933682799339294, + "learning_rate": 0.0004928122529218467, + "loss": 4.8359, + "step": 964 + }, + { + "epoch": 0.09423828125, + "grad_norm": 0.5197924375534058, + "learning_rate": 0.0004927946766865298, + "loss": 4.8281, + "step": 965 + }, + { + "epoch": 0.0943359375, + "grad_norm": 0.5642294883728027, + "learning_rate": 0.0004927770793374971, + "loss": 4.8203, + "step": 966 + }, + { + "epoch": 0.09443359375, + "grad_norm": 0.5989418029785156, + "learning_rate": 0.0004927594608764546, + "loss": 4.8086, + "step": 967 + }, + { + "epoch": 0.09453125, + "grad_norm": 0.5477548837661743, + "learning_rate": 0.0004927418213051104, + "loss": 4.8555, + "step": 968 + }, + { + "epoch": 0.09462890625, + "grad_norm": 0.43198466300964355, + "learning_rate": 0.0004927241606251745, + "loss": 4.8125, + "step": 969 + }, + { + "epoch": 0.0947265625, + "grad_norm": 0.42610111832618713, + "learning_rate": 0.0004927064788383587, + "loss": 4.8477, + "step": 970 + }, + { + "epoch": 0.09482421875, + "grad_norm": 0.4663979709148407, + "learning_rate": 0.0004926887759463776, + "loss": 4.8281, + "step": 971 + }, + { + "epoch": 0.094921875, + "grad_norm": 0.46881988644599915, + "learning_rate": 0.000492671051950947, + "loss": 4.8438, + "step": 972 + }, + { + "epoch": 0.09501953125, + "grad_norm": 0.36195266246795654, + "learning_rate": 0.0004926533068537852, + "loss": 4.8477, + "step": 973 + }, + { + "epoch": 0.0951171875, + "grad_norm": 0.3922168016433716, + "learning_rate": 0.0004926355406566127, + "loss": 4.8555, + "step": 974 + }, + { + "epoch": 0.09521484375, + "grad_norm": 0.425672709941864, + "learning_rate": 0.0004926177533611514, + "loss": 4.8164, + "step": 975 + }, + { + "epoch": 0.0953125, + "grad_norm": 0.5194612145423889, + "learning_rate": 0.0004925999449691261, + "loss": 4.8281, + "step": 976 + }, + { + "epoch": 0.09541015625, + "grad_norm": 0.5651121735572815, + "learning_rate": 0.000492582115482263, + "loss": 4.8164, + "step": 977 + }, + { + "epoch": 0.0955078125, + "grad_norm": 0.5677608847618103, + "learning_rate": 0.0004925642649022903, + "loss": 4.8711, + "step": 978 + }, + { + "epoch": 0.09560546875, + "grad_norm": 0.5311588644981384, + "learning_rate": 0.000492546393230939, + "loss": 4.8242, + "step": 979 + }, + { + "epoch": 0.095703125, + "grad_norm": 0.5084513425827026, + "learning_rate": 0.0004925285004699411, + "loss": 4.832, + "step": 980 + }, + { + "epoch": 0.09580078125, + "grad_norm": 0.6232282519340515, + "learning_rate": 0.0004925105866210316, + "loss": 4.8672, + "step": 981 + }, + { + "epoch": 0.0958984375, + "grad_norm": 0.695502519607544, + "learning_rate": 0.0004924926516859469, + "loss": 4.8281, + "step": 982 + }, + { + "epoch": 0.09599609375, + "grad_norm": 0.6766898036003113, + "learning_rate": 0.000492474695666426, + "loss": 4.8203, + "step": 983 + }, + { + "epoch": 0.09609375, + "grad_norm": 0.5815792679786682, + "learning_rate": 0.0004924567185642091, + "loss": 4.8359, + "step": 984 + }, + { + "epoch": 0.09619140625, + "grad_norm": 0.48480623960494995, + "learning_rate": 0.0004924387203810393, + "loss": 4.8633, + "step": 985 + }, + { + "epoch": 0.0962890625, + "grad_norm": 0.46070924401283264, + "learning_rate": 0.0004924207011186613, + "loss": 4.8516, + "step": 986 + }, + { + "epoch": 0.09638671875, + "grad_norm": 0.4500672221183777, + "learning_rate": 0.0004924026607788219, + "loss": 4.8438, + "step": 987 + }, + { + "epoch": 0.096484375, + "grad_norm": 0.42609307169914246, + "learning_rate": 0.0004923845993632702, + "loss": 4.8164, + "step": 988 + }, + { + "epoch": 0.09658203125, + "grad_norm": 0.4055596590042114, + "learning_rate": 0.000492366516873757, + "loss": 4.8633, + "step": 989 + }, + { + "epoch": 0.0966796875, + "grad_norm": 0.37348657846450806, + "learning_rate": 0.0004923484133120351, + "loss": 4.8203, + "step": 990 + }, + { + "epoch": 0.09677734375, + "grad_norm": 0.3880004584789276, + "learning_rate": 0.0004923302886798598, + "loss": 4.8125, + "step": 991 + }, + { + "epoch": 0.096875, + "grad_norm": 0.381511926651001, + "learning_rate": 0.000492312142978988, + "loss": 4.8359, + "step": 992 + }, + { + "epoch": 0.09697265625, + "grad_norm": 0.4143732786178589, + "learning_rate": 0.0004922939762111788, + "loss": 4.8281, + "step": 993 + }, + { + "epoch": 0.0970703125, + "grad_norm": 0.4248274266719818, + "learning_rate": 0.0004922757883781934, + "loss": 4.8555, + "step": 994 + }, + { + "epoch": 0.09716796875, + "grad_norm": 0.43763217329978943, + "learning_rate": 0.000492257579481795, + "loss": 4.8047, + "step": 995 + }, + { + "epoch": 0.097265625, + "grad_norm": 0.4777696430683136, + "learning_rate": 0.0004922393495237488, + "loss": 4.8125, + "step": 996 + }, + { + "epoch": 0.09736328125, + "grad_norm": 0.5218069553375244, + "learning_rate": 0.000492221098505822, + "loss": 4.8125, + "step": 997 + }, + { + "epoch": 0.0974609375, + "grad_norm": 0.5178037285804749, + "learning_rate": 0.000492202826429784, + "loss": 4.8242, + "step": 998 + }, + { + "epoch": 0.09755859375, + "grad_norm": 0.45934903621673584, + "learning_rate": 0.0004921845332974062, + "loss": 4.8125, + "step": 999 + }, + { + "epoch": 0.09765625, + "grad_norm": 0.4091397821903229, + "learning_rate": 0.0004921662191104619, + "loss": 4.8359, + "step": 1000 + }, + { + "epoch": 0.09775390625, + "grad_norm": 0.3937183618545532, + "learning_rate": 0.0004921478838707266, + "loss": 4.8359, + "step": 1001 + }, + { + "epoch": 0.0978515625, + "grad_norm": 0.4575563073158264, + "learning_rate": 0.0004921295275799778, + "loss": 4.8281, + "step": 1002 + }, + { + "epoch": 0.09794921875, + "grad_norm": 0.490375280380249, + "learning_rate": 0.0004921111502399949, + "loss": 4.832, + "step": 1003 + }, + { + "epoch": 0.098046875, + "grad_norm": 0.4874227046966553, + "learning_rate": 0.0004920927518525594, + "loss": 4.8203, + "step": 1004 + }, + { + "epoch": 0.09814453125, + "grad_norm": 0.42199403047561646, + "learning_rate": 0.0004920743324194552, + "loss": 4.8555, + "step": 1005 + }, + { + "epoch": 0.0982421875, + "grad_norm": 0.49443531036376953, + "learning_rate": 0.0004920558919424677, + "loss": 4.8047, + "step": 1006 + }, + { + "epoch": 0.09833984375, + "grad_norm": 0.39844444394111633, + "learning_rate": 0.0004920374304233846, + "loss": 4.8203, + "step": 1007 + }, + { + "epoch": 0.0984375, + "grad_norm": 0.462501585483551, + "learning_rate": 0.0004920189478639957, + "loss": 4.8477, + "step": 1008 + }, + { + "epoch": 0.09853515625, + "grad_norm": 0.5070082545280457, + "learning_rate": 0.0004920004442660927, + "loss": 4.8203, + "step": 1009 + }, + { + "epoch": 0.0986328125, + "grad_norm": 0.47550520300865173, + "learning_rate": 0.0004919819196314695, + "loss": 4.8008, + "step": 1010 + }, + { + "epoch": 0.09873046875, + "grad_norm": 0.4895987808704376, + "learning_rate": 0.0004919633739619218, + "loss": 4.8125, + "step": 1011 + }, + { + "epoch": 0.098828125, + "grad_norm": 0.48678672313690186, + "learning_rate": 0.0004919448072592474, + "loss": 4.8086, + "step": 1012 + }, + { + "epoch": 0.09892578125, + "grad_norm": 0.4505287706851959, + "learning_rate": 0.0004919262195252465, + "loss": 4.8281, + "step": 1013 + }, + { + "epoch": 0.0990234375, + "grad_norm": 0.8226277232170105, + "learning_rate": 0.0004919076107617209, + "loss": 4.8047, + "step": 1014 + }, + { + "epoch": 0.09912109375, + "grad_norm": 0.6768991947174072, + "learning_rate": 0.0004918889809704745, + "loss": 4.8359, + "step": 1015 + }, + { + "epoch": 0.09921875, + "grad_norm": 0.573598325252533, + "learning_rate": 0.0004918703301533135, + "loss": 4.832, + "step": 1016 + }, + { + "epoch": 0.09931640625, + "grad_norm": 0.49422210454940796, + "learning_rate": 0.0004918516583120458, + "loss": 4.8555, + "step": 1017 + }, + { + "epoch": 0.0994140625, + "grad_norm": 0.5521137714385986, + "learning_rate": 0.0004918329654484817, + "loss": 4.8203, + "step": 1018 + }, + { + "epoch": 0.09951171875, + "grad_norm": 0.6036257743835449, + "learning_rate": 0.0004918142515644332, + "loss": 4.8008, + "step": 1019 + }, + { + "epoch": 0.099609375, + "grad_norm": 0.8331734538078308, + "learning_rate": 0.0004917955166617146, + "loss": 4.8203, + "step": 1020 + }, + { + "epoch": 0.09970703125, + "grad_norm": 1.3778009414672852, + "learning_rate": 0.000491776760742142, + "loss": 4.8281, + "step": 1021 + }, + { + "epoch": 0.0998046875, + "grad_norm": 0.785008430480957, + "learning_rate": 0.0004917579838075337, + "loss": 4.8477, + "step": 1022 + }, + { + "epoch": 0.09990234375, + "grad_norm": 0.7883325815200806, + "learning_rate": 0.0004917391858597099, + "loss": 4.8398, + "step": 1023 + }, + { + "epoch": 0.1, + "grad_norm": 0.5901327729225159, + "learning_rate": 0.0004917203669004932, + "loss": 4.8555, + "step": 1024 + }, + { + "epoch": 0.10009765625, + "grad_norm": 0.4483931064605713, + "learning_rate": 0.0004917015269317079, + "loss": 4.832, + "step": 1025 + }, + { + "epoch": 0.1001953125, + "grad_norm": 0.6025552749633789, + "learning_rate": 0.0004916826659551802, + "loss": 4.8438, + "step": 1026 + }, + { + "epoch": 0.10029296875, + "grad_norm": 0.691528856754303, + "learning_rate": 0.0004916637839727387, + "loss": 4.8477, + "step": 1027 + }, + { + "epoch": 0.100390625, + "grad_norm": 0.922143816947937, + "learning_rate": 0.0004916448809862137, + "loss": 4.8398, + "step": 1028 + }, + { + "epoch": 0.10048828125, + "grad_norm": 0.7501768469810486, + "learning_rate": 0.000491625956997438, + "loss": 4.8438, + "step": 1029 + }, + { + "epoch": 0.1005859375, + "grad_norm": 0.5718377828598022, + "learning_rate": 0.000491607012008246, + "loss": 4.8047, + "step": 1030 + }, + { + "epoch": 0.10068359375, + "grad_norm": 0.5166970491409302, + "learning_rate": 0.0004915880460204743, + "loss": 4.832, + "step": 1031 + }, + { + "epoch": 0.10078125, + "grad_norm": 0.41339635848999023, + "learning_rate": 0.0004915690590359615, + "loss": 4.832, + "step": 1032 + }, + { + "epoch": 0.10087890625, + "grad_norm": 0.42061278223991394, + "learning_rate": 0.0004915500510565483, + "loss": 4.832, + "step": 1033 + }, + { + "epoch": 0.1009765625, + "grad_norm": 0.8583418130874634, + "learning_rate": 0.0004915310220840774, + "loss": 4.8047, + "step": 1034 + }, + { + "epoch": 0.10107421875, + "grad_norm": 1.503048062324524, + "learning_rate": 0.0004915119721203935, + "loss": 4.8555, + "step": 1035 + }, + { + "epoch": 0.101171875, + "grad_norm": 1.4394514560699463, + "learning_rate": 0.0004914929011673434, + "loss": 4.8281, + "step": 1036 + }, + { + "epoch": 0.10126953125, + "grad_norm": 1.117308259010315, + "learning_rate": 0.0004914738092267758, + "loss": 4.8398, + "step": 1037 + }, + { + "epoch": 0.1013671875, + "grad_norm": 0.6745375990867615, + "learning_rate": 0.0004914546963005416, + "loss": 4.8438, + "step": 1038 + }, + { + "epoch": 0.10146484375, + "grad_norm": 1.0236302614212036, + "learning_rate": 0.0004914355623904938, + "loss": 4.8398, + "step": 1039 + }, + { + "epoch": 0.1015625, + "grad_norm": 0.8115978240966797, + "learning_rate": 0.0004914164074984872, + "loss": 4.8633, + "step": 1040 + }, + { + "epoch": 0.10166015625, + "grad_norm": 0.5963826775550842, + "learning_rate": 0.0004913972316263785, + "loss": 4.8203, + "step": 1041 + }, + { + "epoch": 0.1017578125, + "grad_norm": 0.49297934770584106, + "learning_rate": 0.000491378034776027, + "loss": 4.8086, + "step": 1042 + }, + { + "epoch": 0.10185546875, + "grad_norm": 0.4009886085987091, + "learning_rate": 0.0004913588169492937, + "loss": 4.8203, + "step": 1043 + }, + { + "epoch": 0.101953125, + "grad_norm": 0.4827432930469513, + "learning_rate": 0.0004913395781480414, + "loss": 4.8398, + "step": 1044 + }, + { + "epoch": 0.10205078125, + "grad_norm": 0.402046799659729, + "learning_rate": 0.0004913203183741354, + "loss": 4.832, + "step": 1045 + }, + { + "epoch": 0.1021484375, + "grad_norm": 0.4627341330051422, + "learning_rate": 0.0004913010376294425, + "loss": 4.8125, + "step": 1046 + }, + { + "epoch": 0.10224609375, + "grad_norm": 0.4753624200820923, + "learning_rate": 0.0004912817359158322, + "loss": 4.8203, + "step": 1047 + }, + { + "epoch": 0.10234375, + "grad_norm": 0.5196328163146973, + "learning_rate": 0.0004912624132351755, + "loss": 4.7969, + "step": 1048 + }, + { + "epoch": 0.10244140625, + "grad_norm": 0.9553819894790649, + "learning_rate": 0.0004912430695893456, + "loss": 4.7969, + "step": 1049 + }, + { + "epoch": 0.1025390625, + "grad_norm": 0.7087034583091736, + "learning_rate": 0.0004912237049802178, + "loss": 4.8008, + "step": 1050 + }, + { + "epoch": 0.10263671875, + "grad_norm": 0.6159484386444092, + "learning_rate": 0.0004912043194096693, + "loss": 4.8047, + "step": 1051 + }, + { + "epoch": 0.102734375, + "grad_norm": 0.5763092637062073, + "learning_rate": 0.0004911849128795793, + "loss": 4.8359, + "step": 1052 + }, + { + "epoch": 0.10283203125, + "grad_norm": 0.5660399198532104, + "learning_rate": 0.0004911654853918293, + "loss": 4.7891, + "step": 1053 + }, + { + "epoch": 0.1029296875, + "grad_norm": 0.5883915424346924, + "learning_rate": 0.0004911460369483026, + "loss": 4.8203, + "step": 1054 + }, + { + "epoch": 0.10302734375, + "grad_norm": 0.5926139950752258, + "learning_rate": 0.0004911265675508847, + "loss": 4.8438, + "step": 1055 + }, + { + "epoch": 0.103125, + "grad_norm": 0.5742599964141846, + "learning_rate": 0.000491107077201463, + "loss": 4.8203, + "step": 1056 + }, + { + "epoch": 0.10322265625, + "grad_norm": 0.4908924102783203, + "learning_rate": 0.0004910875659019267, + "loss": 4.8164, + "step": 1057 + }, + { + "epoch": 0.1033203125, + "grad_norm": 0.42929133772850037, + "learning_rate": 0.0004910680336541676, + "loss": 4.8125, + "step": 1058 + }, + { + "epoch": 0.10341796875, + "grad_norm": 0.384483277797699, + "learning_rate": 0.000491048480460079, + "loss": 4.8008, + "step": 1059 + }, + { + "epoch": 0.103515625, + "grad_norm": 0.36012735962867737, + "learning_rate": 0.0004910289063215564, + "loss": 4.8203, + "step": 1060 + }, + { + "epoch": 0.10361328125, + "grad_norm": 0.3993228077888489, + "learning_rate": 0.0004910093112404978, + "loss": 4.8477, + "step": 1061 + }, + { + "epoch": 0.1037109375, + "grad_norm": 0.3865804970264435, + "learning_rate": 0.0004909896952188024, + "loss": 4.8125, + "step": 1062 + }, + { + "epoch": 0.10380859375, + "grad_norm": 0.4794972240924835, + "learning_rate": 0.0004909700582583721, + "loss": 4.7969, + "step": 1063 + }, + { + "epoch": 0.10390625, + "grad_norm": 0.5843653082847595, + "learning_rate": 0.0004909504003611103, + "loss": 4.8008, + "step": 1064 + }, + { + "epoch": 0.10400390625, + "grad_norm": 0.7108316421508789, + "learning_rate": 0.000490930721528923, + "loss": 4.8164, + "step": 1065 + }, + { + "epoch": 0.1041015625, + "grad_norm": 0.7192037105560303, + "learning_rate": 0.0004909110217637177, + "loss": 4.8477, + "step": 1066 + }, + { + "epoch": 0.10419921875, + "grad_norm": 0.6272011995315552, + "learning_rate": 0.0004908913010674041, + "loss": 4.8125, + "step": 1067 + }, + { + "epoch": 0.104296875, + "grad_norm": 0.5953921675682068, + "learning_rate": 0.0004908715594418942, + "loss": 4.8398, + "step": 1068 + }, + { + "epoch": 0.10439453125, + "grad_norm": 0.4558872878551483, + "learning_rate": 0.0004908517968891018, + "loss": 4.8125, + "step": 1069 + }, + { + "epoch": 0.1044921875, + "grad_norm": 0.3889329135417938, + "learning_rate": 0.0004908320134109427, + "loss": 4.8281, + "step": 1070 + }, + { + "epoch": 0.10458984375, + "grad_norm": 0.3157101571559906, + "learning_rate": 0.0004908122090093347, + "loss": 4.7969, + "step": 1071 + }, + { + "epoch": 0.1046875, + "grad_norm": 0.3802729547023773, + "learning_rate": 0.0004907923836861978, + "loss": 4.8086, + "step": 1072 + }, + { + "epoch": 0.10478515625, + "grad_norm": 0.402535080909729, + "learning_rate": 0.0004907725374434539, + "loss": 4.8086, + "step": 1073 + }, + { + "epoch": 0.1048828125, + "grad_norm": 0.4550858438014984, + "learning_rate": 0.0004907526702830268, + "loss": 4.8125, + "step": 1074 + }, + { + "epoch": 0.10498046875, + "grad_norm": 0.45571887493133545, + "learning_rate": 0.0004907327822068427, + "loss": 4.8086, + "step": 1075 + }, + { + "epoch": 0.105078125, + "grad_norm": 0.3914882242679596, + "learning_rate": 0.0004907128732168297, + "loss": 4.8242, + "step": 1076 + }, + { + "epoch": 0.10517578125, + "grad_norm": 0.34561851620674133, + "learning_rate": 0.0004906929433149175, + "loss": 4.8203, + "step": 1077 + }, + { + "epoch": 0.1052734375, + "grad_norm": 0.3450526297092438, + "learning_rate": 0.0004906729925030385, + "loss": 4.8203, + "step": 1078 + }, + { + "epoch": 0.10537109375, + "grad_norm": 0.3435506224632263, + "learning_rate": 0.0004906530207831266, + "loss": 4.7969, + "step": 1079 + }, + { + "epoch": 0.10546875, + "grad_norm": 0.3784290552139282, + "learning_rate": 0.0004906330281571179, + "loss": 4.8359, + "step": 1080 + }, + { + "epoch": 0.10556640625, + "grad_norm": 0.4976685345172882, + "learning_rate": 0.0004906130146269507, + "loss": 4.793, + "step": 1081 + }, + { + "epoch": 0.1056640625, + "grad_norm": 0.5724058747291565, + "learning_rate": 0.000490592980194565, + "loss": 4.8008, + "step": 1082 + }, + { + "epoch": 0.10576171875, + "grad_norm": 0.546415388584137, + "learning_rate": 0.0004905729248619032, + "loss": 4.7891, + "step": 1083 + }, + { + "epoch": 0.105859375, + "grad_norm": 0.5786755681037903, + "learning_rate": 0.0004905528486309095, + "loss": 4.8125, + "step": 1084 + }, + { + "epoch": 0.10595703125, + "grad_norm": 0.46551647782325745, + "learning_rate": 0.00049053275150353, + "loss": 4.793, + "step": 1085 + }, + { + "epoch": 0.1060546875, + "grad_norm": 0.3895544707775116, + "learning_rate": 0.0004905126334817131, + "loss": 4.7891, + "step": 1086 + }, + { + "epoch": 0.10615234375, + "grad_norm": 0.3977148234844208, + "learning_rate": 0.000490492494567409, + "loss": 4.8516, + "step": 1087 + }, + { + "epoch": 0.10625, + "grad_norm": 0.41639018058776855, + "learning_rate": 0.0004904723347625702, + "loss": 4.8125, + "step": 1088 + }, + { + "epoch": 0.10634765625, + "grad_norm": 0.4226835370063782, + "learning_rate": 0.0004904521540691509, + "loss": 4.8242, + "step": 1089 + }, + { + "epoch": 0.1064453125, + "grad_norm": 0.47172993421554565, + "learning_rate": 0.0004904319524891076, + "loss": 4.7734, + "step": 1090 + }, + { + "epoch": 0.10654296875, + "grad_norm": 0.438104510307312, + "learning_rate": 0.0004904117300243986, + "loss": 4.8203, + "step": 1091 + }, + { + "epoch": 0.106640625, + "grad_norm": 0.38510963320732117, + "learning_rate": 0.0004903914866769843, + "loss": 4.8164, + "step": 1092 + }, + { + "epoch": 0.10673828125, + "grad_norm": 0.404243528842926, + "learning_rate": 0.0004903712224488273, + "loss": 4.8359, + "step": 1093 + }, + { + "epoch": 0.1068359375, + "grad_norm": 0.4017542898654938, + "learning_rate": 0.0004903509373418921, + "loss": 4.8242, + "step": 1094 + }, + { + "epoch": 0.10693359375, + "grad_norm": 0.4477319121360779, + "learning_rate": 0.000490330631358145, + "loss": 4.7969, + "step": 1095 + }, + { + "epoch": 0.10703125, + "grad_norm": 0.3956342041492462, + "learning_rate": 0.0004903103044995548, + "loss": 4.8047, + "step": 1096 + }, + { + "epoch": 0.10712890625, + "grad_norm": 0.43987393379211426, + "learning_rate": 0.0004902899567680917, + "loss": 4.7734, + "step": 1097 + }, + { + "epoch": 0.1072265625, + "grad_norm": 0.4343857169151306, + "learning_rate": 0.0004902695881657286, + "loss": 4.8477, + "step": 1098 + }, + { + "epoch": 0.10732421875, + "grad_norm": 0.3674246370792389, + "learning_rate": 0.0004902491986944399, + "loss": 4.8125, + "step": 1099 + }, + { + "epoch": 0.107421875, + "grad_norm": 0.3673463761806488, + "learning_rate": 0.0004902287883562023, + "loss": 4.8125, + "step": 1100 + }, + { + "epoch": 0.10751953125, + "grad_norm": 0.4112495481967926, + "learning_rate": 0.0004902083571529944, + "loss": 4.8047, + "step": 1101 + }, + { + "epoch": 0.1076171875, + "grad_norm": 0.4617629647254944, + "learning_rate": 0.000490187905086797, + "loss": 4.7695, + "step": 1102 + }, + { + "epoch": 0.10771484375, + "grad_norm": 0.49318692088127136, + "learning_rate": 0.0004901674321595925, + "loss": 4.7812, + "step": 1103 + }, + { + "epoch": 0.1078125, + "grad_norm": 0.5360081195831299, + "learning_rate": 0.0004901469383733659, + "loss": 4.8242, + "step": 1104 + }, + { + "epoch": 0.10791015625, + "grad_norm": 0.5867539048194885, + "learning_rate": 0.0004901264237301039, + "loss": 4.7852, + "step": 1105 + }, + { + "epoch": 0.1080078125, + "grad_norm": 0.5440882444381714, + "learning_rate": 0.0004901058882317951, + "loss": 4.8203, + "step": 1106 + }, + { + "epoch": 0.10810546875, + "grad_norm": 0.4591827094554901, + "learning_rate": 0.0004900853318804304, + "loss": 4.8164, + "step": 1107 + }, + { + "epoch": 0.108203125, + "grad_norm": 0.4257602095603943, + "learning_rate": 0.0004900647546780026, + "loss": 4.8008, + "step": 1108 + }, + { + "epoch": 0.10830078125, + "grad_norm": 0.418489009141922, + "learning_rate": 0.0004900441566265064, + "loss": 4.832, + "step": 1109 + }, + { + "epoch": 0.1083984375, + "grad_norm": 0.4998464286327362, + "learning_rate": 0.0004900235377279388, + "loss": 4.7773, + "step": 1110 + }, + { + "epoch": 0.10849609375, + "grad_norm": 0.47938549518585205, + "learning_rate": 0.0004900028979842986, + "loss": 4.8125, + "step": 1111 + }, + { + "epoch": 0.10859375, + "grad_norm": 0.4215286672115326, + "learning_rate": 0.0004899822373975866, + "loss": 4.7812, + "step": 1112 + }, + { + "epoch": 0.10869140625, + "grad_norm": 0.41954976320266724, + "learning_rate": 0.0004899615559698058, + "loss": 4.8047, + "step": 1113 + }, + { + "epoch": 0.1087890625, + "grad_norm": 0.4177006781101227, + "learning_rate": 0.0004899408537029611, + "loss": 4.8047, + "step": 1114 + }, + { + "epoch": 0.10888671875, + "grad_norm": 0.5628485679626465, + "learning_rate": 0.0004899201305990594, + "loss": 4.7969, + "step": 1115 + }, + { + "epoch": 0.108984375, + "grad_norm": 0.6657236814498901, + "learning_rate": 0.0004898993866601098, + "loss": 4.8086, + "step": 1116 + }, + { + "epoch": 0.10908203125, + "grad_norm": 0.6637690663337708, + "learning_rate": 0.0004898786218881232, + "loss": 4.8203, + "step": 1117 + }, + { + "epoch": 0.1091796875, + "grad_norm": 0.5784990787506104, + "learning_rate": 0.0004898578362851124, + "loss": 4.832, + "step": 1118 + }, + { + "epoch": 0.10927734375, + "grad_norm": 0.42443713545799255, + "learning_rate": 0.0004898370298530928, + "loss": 4.8008, + "step": 1119 + }, + { + "epoch": 0.109375, + "grad_norm": 0.3830028474330902, + "learning_rate": 0.0004898162025940812, + "loss": 4.8398, + "step": 1120 + }, + { + "epoch": 0.10947265625, + "grad_norm": 0.45647111535072327, + "learning_rate": 0.0004897953545100966, + "loss": 4.8086, + "step": 1121 + }, + { + "epoch": 0.1095703125, + "grad_norm": 0.49465152621269226, + "learning_rate": 0.0004897744856031604, + "loss": 4.8477, + "step": 1122 + }, + { + "epoch": 0.10966796875, + "grad_norm": 0.5248700380325317, + "learning_rate": 0.0004897535958752954, + "loss": 4.7891, + "step": 1123 + }, + { + "epoch": 0.109765625, + "grad_norm": 0.5552516579627991, + "learning_rate": 0.0004897326853285268, + "loss": 4.8125, + "step": 1124 + }, + { + "epoch": 0.10986328125, + "grad_norm": 0.4863637387752533, + "learning_rate": 0.0004897117539648818, + "loss": 4.8203, + "step": 1125 + }, + { + "epoch": 0.1099609375, + "grad_norm": 0.4048306941986084, + "learning_rate": 0.0004896908017863895, + "loss": 4.7852, + "step": 1126 + }, + { + "epoch": 0.11005859375, + "grad_norm": 0.4323672652244568, + "learning_rate": 0.0004896698287950812, + "loss": 4.7695, + "step": 1127 + }, + { + "epoch": 0.11015625, + "grad_norm": 0.5875673890113831, + "learning_rate": 0.0004896488349929898, + "loss": 4.793, + "step": 1128 + }, + { + "epoch": 0.11025390625, + "grad_norm": 0.5504245758056641, + "learning_rate": 0.0004896278203821506, + "loss": 4.7852, + "step": 1129 + }, + { + "epoch": 0.1103515625, + "grad_norm": 0.44111916422843933, + "learning_rate": 0.0004896067849646011, + "loss": 4.7852, + "step": 1130 + }, + { + "epoch": 0.11044921875, + "grad_norm": 0.37442272901535034, + "learning_rate": 0.0004895857287423802, + "loss": 4.7578, + "step": 1131 + }, + { + "epoch": 0.110546875, + "grad_norm": 0.38579708337783813, + "learning_rate": 0.0004895646517175294, + "loss": 4.7734, + "step": 1132 + }, + { + "epoch": 0.11064453125, + "grad_norm": 0.350492388010025, + "learning_rate": 0.0004895435538920918, + "loss": 4.7969, + "step": 1133 + }, + { + "epoch": 0.1107421875, + "grad_norm": 0.3320639431476593, + "learning_rate": 0.0004895224352681127, + "loss": 4.7969, + "step": 1134 + }, + { + "epoch": 0.11083984375, + "grad_norm": 0.33712780475616455, + "learning_rate": 0.0004895012958476396, + "loss": 4.793, + "step": 1135 + }, + { + "epoch": 0.1109375, + "grad_norm": 0.3295286297798157, + "learning_rate": 0.0004894801356327217, + "loss": 4.8047, + "step": 1136 + }, + { + "epoch": 0.11103515625, + "grad_norm": 0.3195537030696869, + "learning_rate": 0.0004894589546254102, + "loss": 4.8047, + "step": 1137 + }, + { + "epoch": 0.1111328125, + "grad_norm": 0.36060598492622375, + "learning_rate": 0.0004894377528277587, + "loss": 4.8008, + "step": 1138 + }, + { + "epoch": 0.11123046875, + "grad_norm": 0.4128583073616028, + "learning_rate": 0.0004894165302418224, + "loss": 4.8125, + "step": 1139 + }, + { + "epoch": 0.111328125, + "grad_norm": 0.6390661001205444, + "learning_rate": 0.0004893952868696588, + "loss": 4.8086, + "step": 1140 + }, + { + "epoch": 0.11142578125, + "grad_norm": 0.8795298933982849, + "learning_rate": 0.0004893740227133272, + "loss": 4.793, + "step": 1141 + }, + { + "epoch": 0.1115234375, + "grad_norm": 0.5830650329589844, + "learning_rate": 0.000489352737774889, + "loss": 4.8164, + "step": 1142 + }, + { + "epoch": 0.11162109375, + "grad_norm": 0.6256621479988098, + "learning_rate": 0.0004893314320564078, + "loss": 4.8086, + "step": 1143 + }, + { + "epoch": 0.11171875, + "grad_norm": 0.6185265779495239, + "learning_rate": 0.0004893101055599488, + "loss": 4.7969, + "step": 1144 + }, + { + "epoch": 0.11181640625, + "grad_norm": 0.5876705646514893, + "learning_rate": 0.0004892887582875797, + "loss": 4.7969, + "step": 1145 + }, + { + "epoch": 0.1119140625, + "grad_norm": 0.5868121981620789, + "learning_rate": 0.0004892673902413699, + "loss": 4.8125, + "step": 1146 + }, + { + "epoch": 0.11201171875, + "grad_norm": 0.46918195486068726, + "learning_rate": 0.0004892460014233907, + "loss": 4.8086, + "step": 1147 + }, + { + "epoch": 0.112109375, + "grad_norm": 0.3593907058238983, + "learning_rate": 0.0004892245918357159, + "loss": 4.832, + "step": 1148 + }, + { + "epoch": 0.11220703125, + "grad_norm": 0.4415476620197296, + "learning_rate": 0.0004892031614804208, + "loss": 4.7852, + "step": 1149 + }, + { + "epoch": 0.1123046875, + "grad_norm": 0.7689213156700134, + "learning_rate": 0.000489181710359583, + "loss": 4.7891, + "step": 1150 + }, + { + "epoch": 0.11240234375, + "grad_norm": 1.3944097757339478, + "learning_rate": 0.000489160238475282, + "loss": 4.8086, + "step": 1151 + }, + { + "epoch": 0.1125, + "grad_norm": 0.6750385165214539, + "learning_rate": 0.0004891387458295995, + "loss": 4.8047, + "step": 1152 + }, + { + "epoch": 0.11259765625, + "grad_norm": 0.6533737778663635, + "learning_rate": 0.0004891172324246189, + "loss": 4.8164, + "step": 1153 + }, + { + "epoch": 0.1126953125, + "grad_norm": 0.698330283164978, + "learning_rate": 0.0004890956982624258, + "loss": 4.8203, + "step": 1154 + }, + { + "epoch": 0.11279296875, + "grad_norm": 0.5275533199310303, + "learning_rate": 0.0004890741433451079, + "loss": 4.7969, + "step": 1155 + }, + { + "epoch": 0.112890625, + "grad_norm": 0.39748457074165344, + "learning_rate": 0.0004890525676747547, + "loss": 4.8086, + "step": 1156 + }, + { + "epoch": 0.11298828125, + "grad_norm": 0.49489057064056396, + "learning_rate": 0.0004890309712534578, + "loss": 4.8047, + "step": 1157 + }, + { + "epoch": 0.1130859375, + "grad_norm": 0.4166368544101715, + "learning_rate": 0.0004890093540833108, + "loss": 4.832, + "step": 1158 + }, + { + "epoch": 0.11318359375, + "grad_norm": 0.3709654211997986, + "learning_rate": 0.0004889877161664096, + "loss": 4.8281, + "step": 1159 + }, + { + "epoch": 0.11328125, + "grad_norm": 0.41362860798835754, + "learning_rate": 0.0004889660575048515, + "loss": 4.7773, + "step": 1160 + }, + { + "epoch": 0.11337890625, + "grad_norm": 0.3981860280036926, + "learning_rate": 0.0004889443781007364, + "loss": 4.7969, + "step": 1161 + }, + { + "epoch": 0.1134765625, + "grad_norm": 0.4014473557472229, + "learning_rate": 0.000488922677956166, + "loss": 4.793, + "step": 1162 + }, + { + "epoch": 0.11357421875, + "grad_norm": 0.5000125765800476, + "learning_rate": 0.0004889009570732436, + "loss": 4.7891, + "step": 1163 + }, + { + "epoch": 0.113671875, + "grad_norm": 0.6138625144958496, + "learning_rate": 0.0004888792154540753, + "loss": 4.7812, + "step": 1164 + }, + { + "epoch": 0.11376953125, + "grad_norm": 0.6484094858169556, + "learning_rate": 0.0004888574531007687, + "loss": 4.8008, + "step": 1165 + }, + { + "epoch": 0.1138671875, + "grad_norm": 0.44802355766296387, + "learning_rate": 0.0004888356700154333, + "loss": 4.8047, + "step": 1166 + }, + { + "epoch": 0.11396484375, + "grad_norm": 0.42703965306282043, + "learning_rate": 0.0004888138662001813, + "loss": 4.7969, + "step": 1167 + }, + { + "epoch": 0.1140625, + "grad_norm": 0.41006794571876526, + "learning_rate": 0.0004887920416571259, + "loss": 4.793, + "step": 1168 + }, + { + "epoch": 0.11416015625, + "grad_norm": 0.39348462224006653, + "learning_rate": 0.0004887701963883831, + "loss": 4.793, + "step": 1169 + }, + { + "epoch": 0.1142578125, + "grad_norm": 0.3154531419277191, + "learning_rate": 0.0004887483303960706, + "loss": 4.793, + "step": 1170 + }, + { + "epoch": 0.11435546875, + "grad_norm": 0.3364889323711395, + "learning_rate": 0.0004887264436823083, + "loss": 4.7812, + "step": 1171 + }, + { + "epoch": 0.114453125, + "grad_norm": 0.3992083966732025, + "learning_rate": 0.0004887045362492178, + "loss": 4.793, + "step": 1172 + }, + { + "epoch": 0.11455078125, + "grad_norm": 0.42253056168556213, + "learning_rate": 0.0004886826080989229, + "loss": 4.7734, + "step": 1173 + }, + { + "epoch": 0.1146484375, + "grad_norm": 0.36724087595939636, + "learning_rate": 0.0004886606592335495, + "loss": 4.793, + "step": 1174 + }, + { + "epoch": 0.11474609375, + "grad_norm": 0.3279463052749634, + "learning_rate": 0.0004886386896552252, + "loss": 4.7891, + "step": 1175 + }, + { + "epoch": 0.11484375, + "grad_norm": 0.3356574773788452, + "learning_rate": 0.0004886166993660799, + "loss": 4.7891, + "step": 1176 + }, + { + "epoch": 0.11494140625, + "grad_norm": 0.42385661602020264, + "learning_rate": 0.0004885946883682455, + "loss": 4.7969, + "step": 1177 + }, + { + "epoch": 0.1150390625, + "grad_norm": 0.582700252532959, + "learning_rate": 0.0004885726566638557, + "loss": 4.7812, + "step": 1178 + }, + { + "epoch": 0.11513671875, + "grad_norm": 0.5969910025596619, + "learning_rate": 0.0004885506042550464, + "loss": 4.8359, + "step": 1179 + }, + { + "epoch": 0.115234375, + "grad_norm": 0.5986462831497192, + "learning_rate": 0.0004885285311439553, + "loss": 4.7891, + "step": 1180 + }, + { + "epoch": 0.11533203125, + "grad_norm": 0.49811697006225586, + "learning_rate": 0.0004885064373327223, + "loss": 4.7773, + "step": 1181 + }, + { + "epoch": 0.1154296875, + "grad_norm": 0.38979536294937134, + "learning_rate": 0.0004884843228234895, + "loss": 4.8047, + "step": 1182 + }, + { + "epoch": 0.11552734375, + "grad_norm": 0.40310540795326233, + "learning_rate": 0.0004884621876184004, + "loss": 4.8047, + "step": 1183 + }, + { + "epoch": 0.115625, + "grad_norm": 0.42200982570648193, + "learning_rate": 0.0004884400317196009, + "loss": 4.8203, + "step": 1184 + }, + { + "epoch": 0.11572265625, + "grad_norm": 0.4246567487716675, + "learning_rate": 0.000488417855129239, + "loss": 4.7812, + "step": 1185 + }, + { + "epoch": 0.1158203125, + "grad_norm": 0.4499056935310364, + "learning_rate": 0.0004883956578494645, + "loss": 4.7891, + "step": 1186 + }, + { + "epoch": 0.11591796875, + "grad_norm": 0.42340394854545593, + "learning_rate": 0.0004883734398824294, + "loss": 4.7734, + "step": 1187 + }, + { + "epoch": 0.116015625, + "grad_norm": 0.42388978600502014, + "learning_rate": 0.0004883512012302874, + "loss": 4.8125, + "step": 1188 + }, + { + "epoch": 0.11611328125, + "grad_norm": 0.4048875868320465, + "learning_rate": 0.0004883289418951943, + "loss": 4.7812, + "step": 1189 + }, + { + "epoch": 0.1162109375, + "grad_norm": 0.39005008339881897, + "learning_rate": 0.0004883066618793083, + "loss": 4.7891, + "step": 1190 + }, + { + "epoch": 0.11630859375, + "grad_norm": 0.38380053639411926, + "learning_rate": 0.0004882843611847892, + "loss": 4.7852, + "step": 1191 + }, + { + "epoch": 0.11640625, + "grad_norm": 0.35080215334892273, + "learning_rate": 0.0004882620398137988, + "loss": 4.8125, + "step": 1192 + }, + { + "epoch": 0.11650390625, + "grad_norm": 0.3683309555053711, + "learning_rate": 0.00048823969776850103, + "loss": 4.8086, + "step": 1193 + }, + { + "epoch": 0.1166015625, + "grad_norm": 0.39639347791671753, + "learning_rate": 0.00048821733505106186, + "loss": 4.8164, + "step": 1194 + }, + { + "epoch": 0.11669921875, + "grad_norm": 0.41327810287475586, + "learning_rate": 0.0004881949516636491, + "loss": 4.7539, + "step": 1195 + }, + { + "epoch": 0.116796875, + "grad_norm": 0.4666888117790222, + "learning_rate": 0.0004881725476084328, + "loss": 4.7539, + "step": 1196 + }, + { + "epoch": 0.11689453125, + "grad_norm": 0.6009286046028137, + "learning_rate": 0.00048815012288758484, + "loss": 4.8086, + "step": 1197 + }, + { + "epoch": 0.1169921875, + "grad_norm": 0.6095237731933594, + "learning_rate": 0.00048812767750327905, + "loss": 4.8203, + "step": 1198 + }, + { + "epoch": 0.11708984375, + "grad_norm": 0.5841214656829834, + "learning_rate": 0.0004881052114576915, + "loss": 4.7852, + "step": 1199 + }, + { + "epoch": 0.1171875, + "grad_norm": 0.5666095614433289, + "learning_rate": 0.00048808272475299994, + "loss": 4.8164, + "step": 1200 + }, + { + "epoch": 0.11728515625, + "grad_norm": 0.4737398624420166, + "learning_rate": 0.00048806021739138453, + "loss": 4.7891, + "step": 1201 + }, + { + "epoch": 0.1173828125, + "grad_norm": 0.48223602771759033, + "learning_rate": 0.000488037689375027, + "loss": 4.7812, + "step": 1202 + }, + { + "epoch": 0.11748046875, + "grad_norm": 0.5186586976051331, + "learning_rate": 0.00048801514070611143, + "loss": 4.7812, + "step": 1203 + }, + { + "epoch": 0.117578125, + "grad_norm": 0.499870628118515, + "learning_rate": 0.0004879925713868236, + "loss": 4.7812, + "step": 1204 + }, + { + "epoch": 0.11767578125, + "grad_norm": 0.42764419317245483, + "learning_rate": 0.0004879699814193517, + "loss": 4.7617, + "step": 1205 + }, + { + "epoch": 0.1177734375, + "grad_norm": 0.42484956979751587, + "learning_rate": 0.0004879473708058855, + "loss": 4.7852, + "step": 1206 + }, + { + "epoch": 0.11787109375, + "grad_norm": 0.35906827449798584, + "learning_rate": 0.00048792473954861694, + "loss": 4.7852, + "step": 1207 + }, + { + "epoch": 0.11796875, + "grad_norm": 0.32980072498321533, + "learning_rate": 0.00048790208764973997, + "loss": 4.7695, + "step": 1208 + }, + { + "epoch": 0.11806640625, + "grad_norm": 0.33791640400886536, + "learning_rate": 0.0004878794151114507, + "loss": 4.8008, + "step": 1209 + }, + { + "epoch": 0.1181640625, + "grad_norm": 0.30102401971817017, + "learning_rate": 0.0004878567219359469, + "loss": 4.7852, + "step": 1210 + }, + { + "epoch": 0.11826171875, + "grad_norm": 0.32327136397361755, + "learning_rate": 0.0004878340081254285, + "loss": 4.7852, + "step": 1211 + }, + { + "epoch": 0.118359375, + "grad_norm": 0.37602269649505615, + "learning_rate": 0.0004878112736820976, + "loss": 4.7852, + "step": 1212 + }, + { + "epoch": 0.11845703125, + "grad_norm": 0.4111282229423523, + "learning_rate": 0.000487788518608158, + "loss": 4.7734, + "step": 1213 + }, + { + "epoch": 0.1185546875, + "grad_norm": 0.4097578525543213, + "learning_rate": 0.0004877657429058158, + "loss": 4.7852, + "step": 1214 + }, + { + "epoch": 0.11865234375, + "grad_norm": 0.3875778615474701, + "learning_rate": 0.0004877429465772788, + "loss": 4.7734, + "step": 1215 + }, + { + "epoch": 0.11875, + "grad_norm": 0.372007817029953, + "learning_rate": 0.000487720129624757, + "loss": 4.7852, + "step": 1216 + }, + { + "epoch": 0.11884765625, + "grad_norm": 0.3939751088619232, + "learning_rate": 0.00048769729205046247, + "loss": 4.7773, + "step": 1217 + }, + { + "epoch": 0.1189453125, + "grad_norm": 0.4261675775051117, + "learning_rate": 0.000487674433856609, + "loss": 4.7812, + "step": 1218 + }, + { + "epoch": 0.11904296875, + "grad_norm": 0.5432133078575134, + "learning_rate": 0.00048765155504541265, + "loss": 4.8086, + "step": 1219 + }, + { + "epoch": 0.119140625, + "grad_norm": 0.5995720624923706, + "learning_rate": 0.0004876286556190912, + "loss": 4.8125, + "step": 1220 + }, + { + "epoch": 0.11923828125, + "grad_norm": 0.6350622773170471, + "learning_rate": 0.00048760573557986476, + "loss": 4.7812, + "step": 1221 + }, + { + "epoch": 0.1193359375, + "grad_norm": 0.5896903872489929, + "learning_rate": 0.00048758279492995527, + "loss": 4.7852, + "step": 1222 + }, + { + "epoch": 0.11943359375, + "grad_norm": 0.4300802946090698, + "learning_rate": 0.0004875598336715865, + "loss": 4.7773, + "step": 1223 + }, + { + "epoch": 0.11953125, + "grad_norm": 0.45555663108825684, + "learning_rate": 0.00048753685180698465, + "loss": 4.7852, + "step": 1224 + }, + { + "epoch": 0.11962890625, + "grad_norm": 0.42764827609062195, + "learning_rate": 0.00048751384933837737, + "loss": 4.7812, + "step": 1225 + }, + { + "epoch": 0.1197265625, + "grad_norm": 0.37654969096183777, + "learning_rate": 0.0004874908262679948, + "loss": 4.7891, + "step": 1226 + }, + { + "epoch": 0.11982421875, + "grad_norm": 0.32424551248550415, + "learning_rate": 0.00048746778259806876, + "loss": 4.8086, + "step": 1227 + }, + { + "epoch": 0.119921875, + "grad_norm": 0.35854217410087585, + "learning_rate": 0.0004874447183308333, + "loss": 4.8086, + "step": 1228 + }, + { + "epoch": 0.12001953125, + "grad_norm": 0.38739636540412903, + "learning_rate": 0.0004874216334685242, + "loss": 4.7812, + "step": 1229 + }, + { + "epoch": 0.1201171875, + "grad_norm": 0.38605567812919617, + "learning_rate": 0.0004873985280133795, + "loss": 4.8086, + "step": 1230 + }, + { + "epoch": 0.12021484375, + "grad_norm": 0.36838725209236145, + "learning_rate": 0.00048737540196763904, + "loss": 4.7812, + "step": 1231 + }, + { + "epoch": 0.1203125, + "grad_norm": 0.3206044137477875, + "learning_rate": 0.0004873522553335447, + "loss": 4.7969, + "step": 1232 + }, + { + "epoch": 0.12041015625, + "grad_norm": 0.3341485261917114, + "learning_rate": 0.00048732908811334046, + "loss": 4.793, + "step": 1233 + }, + { + "epoch": 0.1205078125, + "grad_norm": 0.38392874598503113, + "learning_rate": 0.00048730590030927217, + "loss": 4.7852, + "step": 1234 + }, + { + "epoch": 0.12060546875, + "grad_norm": 0.42469480633735657, + "learning_rate": 0.0004872826919235879, + "loss": 4.7578, + "step": 1235 + }, + { + "epoch": 0.120703125, + "grad_norm": 0.5113615989685059, + "learning_rate": 0.00048725946295853737, + "loss": 4.8008, + "step": 1236 + }, + { + "epoch": 0.12080078125, + "grad_norm": 0.5344850420951843, + "learning_rate": 0.0004872362134163724, + "loss": 4.7969, + "step": 1237 + }, + { + "epoch": 0.1208984375, + "grad_norm": 0.4661124050617218, + "learning_rate": 0.0004872129432993471, + "loss": 4.8008, + "step": 1238 + }, + { + "epoch": 0.12099609375, + "grad_norm": 0.45510703325271606, + "learning_rate": 0.00048718965260971726, + "loss": 4.7812, + "step": 1239 + }, + { + "epoch": 0.12109375, + "grad_norm": 0.4054417610168457, + "learning_rate": 0.0004871663413497407, + "loss": 4.8203, + "step": 1240 + }, + { + "epoch": 0.12119140625, + "grad_norm": 0.450013130903244, + "learning_rate": 0.0004871430095216773, + "loss": 4.7812, + "step": 1241 + }, + { + "epoch": 0.1212890625, + "grad_norm": 0.46598708629608154, + "learning_rate": 0.000487119657127789, + "loss": 4.832, + "step": 1242 + }, + { + "epoch": 0.12138671875, + "grad_norm": 0.4022948741912842, + "learning_rate": 0.00048709628417033956, + "loss": 4.7891, + "step": 1243 + }, + { + "epoch": 0.121484375, + "grad_norm": 0.3764280378818512, + "learning_rate": 0.00048707289065159486, + "loss": 4.7852, + "step": 1244 + }, + { + "epoch": 0.12158203125, + "grad_norm": 0.3581683039665222, + "learning_rate": 0.0004870494765738228, + "loss": 4.7578, + "step": 1245 + }, + { + "epoch": 0.1216796875, + "grad_norm": 0.3513264060020447, + "learning_rate": 0.0004870260419392931, + "loss": 4.7734, + "step": 1246 + }, + { + "epoch": 0.12177734375, + "grad_norm": 0.3424482047557831, + "learning_rate": 0.00048700258675027776, + "loss": 4.8125, + "step": 1247 + }, + { + "epoch": 0.121875, + "grad_norm": 0.3603288233280182, + "learning_rate": 0.0004869791110090504, + "loss": 4.7969, + "step": 1248 + }, + { + "epoch": 0.12197265625, + "grad_norm": 0.3708992600440979, + "learning_rate": 0.00048695561471788696, + "loss": 4.7734, + "step": 1249 + }, + { + "epoch": 0.1220703125, + "grad_norm": 0.4168321490287781, + "learning_rate": 0.0004869320978790653, + "loss": 4.8008, + "step": 1250 + }, + { + "epoch": 0.12216796875, + "grad_norm": 0.4125996530056, + "learning_rate": 0.0004869085604948651, + "loss": 4.7734, + "step": 1251 + }, + { + "epoch": 0.122265625, + "grad_norm": 0.42782822251319885, + "learning_rate": 0.0004868850025675681, + "loss": 4.7656, + "step": 1252 + }, + { + "epoch": 0.12236328125, + "grad_norm": 0.34903016686439514, + "learning_rate": 0.0004868614240994583, + "loss": 4.7812, + "step": 1253 + }, + { + "epoch": 0.1224609375, + "grad_norm": 0.37179067730903625, + "learning_rate": 0.00048683782509282127, + "loss": 4.7578, + "step": 1254 + }, + { + "epoch": 0.12255859375, + "grad_norm": 0.3796963393688202, + "learning_rate": 0.0004868142055499448, + "loss": 4.7617, + "step": 1255 + }, + { + "epoch": 0.12265625, + "grad_norm": 0.4198116362094879, + "learning_rate": 0.0004867905654731187, + "loss": 4.8047, + "step": 1256 + }, + { + "epoch": 0.12275390625, + "grad_norm": 0.4762858748435974, + "learning_rate": 0.00048676690486463474, + "loss": 4.7617, + "step": 1257 + }, + { + "epoch": 0.1228515625, + "grad_norm": 0.5248888731002808, + "learning_rate": 0.0004867432237267867, + "loss": 4.7383, + "step": 1258 + }, + { + "epoch": 0.12294921875, + "grad_norm": 0.547907829284668, + "learning_rate": 0.00048671952206187007, + "loss": 4.8047, + "step": 1259 + }, + { + "epoch": 0.123046875, + "grad_norm": 0.42803826928138733, + "learning_rate": 0.00048669579987218285, + "loss": 4.7969, + "step": 1260 + }, + { + "epoch": 0.12314453125, + "grad_norm": 0.322132408618927, + "learning_rate": 0.00048667205716002455, + "loss": 4.7852, + "step": 1261 + }, + { + "epoch": 0.1232421875, + "grad_norm": 0.38514775037765503, + "learning_rate": 0.0004866482939276969, + "loss": 4.7695, + "step": 1262 + }, + { + "epoch": 0.12333984375, + "grad_norm": 0.4669879674911499, + "learning_rate": 0.00048662451017750377, + "loss": 4.7969, + "step": 1263 + }, + { + "epoch": 0.1234375, + "grad_norm": 0.47549909353256226, + "learning_rate": 0.0004866007059117505, + "loss": 4.7773, + "step": 1264 + }, + { + "epoch": 0.12353515625, + "grad_norm": 0.3875703513622284, + "learning_rate": 0.00048657688113274507, + "loss": 4.7852, + "step": 1265 + }, + { + "epoch": 0.1236328125, + "grad_norm": 0.31729233264923096, + "learning_rate": 0.00048655303584279686, + "loss": 4.7773, + "step": 1266 + }, + { + "epoch": 0.12373046875, + "grad_norm": 0.33539122343063354, + "learning_rate": 0.0004865291700442177, + "loss": 4.7617, + "step": 1267 + }, + { + "epoch": 0.123828125, + "grad_norm": 0.3871065676212311, + "learning_rate": 0.0004865052837393212, + "loss": 4.7773, + "step": 1268 + }, + { + "epoch": 0.12392578125, + "grad_norm": 0.40682950615882874, + "learning_rate": 0.00048648137693042283, + "loss": 4.7812, + "step": 1269 + }, + { + "epoch": 0.1240234375, + "grad_norm": 0.4068271219730377, + "learning_rate": 0.0004864574496198404, + "loss": 4.8008, + "step": 1270 + }, + { + "epoch": 0.12412109375, + "grad_norm": 0.3539895713329315, + "learning_rate": 0.0004864335018098933, + "loss": 4.7891, + "step": 1271 + }, + { + "epoch": 0.12421875, + "grad_norm": 0.34347373247146606, + "learning_rate": 0.00048640953350290324, + "loss": 4.7773, + "step": 1272 + }, + { + "epoch": 0.12431640625, + "grad_norm": 0.40734362602233887, + "learning_rate": 0.0004863855447011938, + "loss": 4.7734, + "step": 1273 + }, + { + "epoch": 0.1244140625, + "grad_norm": 0.4735758900642395, + "learning_rate": 0.00048636153540709045, + "loss": 4.7656, + "step": 1274 + }, + { + "epoch": 0.12451171875, + "grad_norm": 0.5711696147918701, + "learning_rate": 0.0004863375056229208, + "loss": 4.8203, + "step": 1275 + }, + { + "epoch": 0.124609375, + "grad_norm": 0.5063640475273132, + "learning_rate": 0.00048631345535101426, + "loss": 4.7617, + "step": 1276 + }, + { + "epoch": 0.12470703125, + "grad_norm": 0.35711005330085754, + "learning_rate": 0.0004862893845937024, + "loss": 4.8125, + "step": 1277 + }, + { + "epoch": 0.1248046875, + "grad_norm": 0.3374655842781067, + "learning_rate": 0.0004862652933533188, + "loss": 4.7734, + "step": 1278 + }, + { + "epoch": 0.12490234375, + "grad_norm": 0.3303259015083313, + "learning_rate": 0.00048624118163219875, + "loss": 4.7812, + "step": 1279 + }, + { + "epoch": 0.125, + "grad_norm": 0.31290751695632935, + "learning_rate": 0.00048621704943267995, + "loss": 4.7539, + "step": 1280 + }, + { + "epoch": 0.12509765625, + "grad_norm": 0.3193962574005127, + "learning_rate": 0.00048619289675710177, + "loss": 4.7578, + "step": 1281 + }, + { + "epoch": 0.1251953125, + "grad_norm": 0.2991279065608978, + "learning_rate": 0.0004861687236078055, + "loss": 4.7734, + "step": 1282 + }, + { + "epoch": 0.12529296875, + "grad_norm": 0.3635973334312439, + "learning_rate": 0.0004861445299871348, + "loss": 4.7344, + "step": 1283 + }, + { + "epoch": 0.125390625, + "grad_norm": 0.4499048888683319, + "learning_rate": 0.0004861203158974349, + "loss": 4.7812, + "step": 1284 + }, + { + "epoch": 0.12548828125, + "grad_norm": 0.6076017022132874, + "learning_rate": 0.00048609608134105324, + "loss": 4.7773, + "step": 1285 + }, + { + "epoch": 0.1255859375, + "grad_norm": 0.6585017442703247, + "learning_rate": 0.0004860718263203393, + "loss": 4.7812, + "step": 1286 + }, + { + "epoch": 0.12568359375, + "grad_norm": 0.48385024070739746, + "learning_rate": 0.0004860475508376442, + "loss": 4.7695, + "step": 1287 + }, + { + "epoch": 0.12578125, + "grad_norm": 0.3241603374481201, + "learning_rate": 0.00048602325489532146, + "loss": 4.7617, + "step": 1288 + }, + { + "epoch": 0.12587890625, + "grad_norm": 0.4593919813632965, + "learning_rate": 0.00048599893849572646, + "loss": 4.7734, + "step": 1289 + }, + { + "epoch": 0.1259765625, + "grad_norm": 0.46703073382377625, + "learning_rate": 0.00048597460164121636, + "loss": 4.7617, + "step": 1290 + }, + { + "epoch": 0.12607421875, + "grad_norm": 0.3566134572029114, + "learning_rate": 0.00048595024433415054, + "loss": 4.7695, + "step": 1291 + }, + { + "epoch": 0.126171875, + "grad_norm": 0.3164132833480835, + "learning_rate": 0.0004859258665768903, + "loss": 4.793, + "step": 1292 + }, + { + "epoch": 0.12626953125, + "grad_norm": 0.37039509415626526, + "learning_rate": 0.00048590146837179876, + "loss": 4.7773, + "step": 1293 + }, + { + "epoch": 0.1263671875, + "grad_norm": 0.37872517108917236, + "learning_rate": 0.00048587704972124135, + "loss": 4.8164, + "step": 1294 + }, + { + "epoch": 0.12646484375, + "grad_norm": 0.35898569226264954, + "learning_rate": 0.0004858526106275851, + "loss": 4.7773, + "step": 1295 + }, + { + "epoch": 0.1265625, + "grad_norm": 0.33450818061828613, + "learning_rate": 0.00048582815109319936, + "loss": 4.7734, + "step": 1296 + }, + { + "epoch": 0.12666015625, + "grad_norm": 0.3074091970920563, + "learning_rate": 0.0004858036711204553, + "loss": 4.7656, + "step": 1297 + }, + { + "epoch": 0.1267578125, + "grad_norm": 0.32790830731391907, + "learning_rate": 0.000485779170711726, + "loss": 4.7852, + "step": 1298 + }, + { + "epoch": 0.12685546875, + "grad_norm": 0.33045750856399536, + "learning_rate": 0.00048575464986938674, + "loss": 4.7383, + "step": 1299 + }, + { + "epoch": 0.126953125, + "grad_norm": 0.3500734269618988, + "learning_rate": 0.0004857301085958145, + "loss": 4.7578, + "step": 1300 + }, + { + "epoch": 0.12705078125, + "grad_norm": 0.39776694774627686, + "learning_rate": 0.0004857055468933885, + "loss": 4.7617, + "step": 1301 + }, + { + "epoch": 0.1271484375, + "grad_norm": 0.36641398072242737, + "learning_rate": 0.0004856809647644897, + "loss": 4.7773, + "step": 1302 + }, + { + "epoch": 0.12724609375, + "grad_norm": 0.3984200656414032, + "learning_rate": 0.00048565636221150135, + "loss": 4.7852, + "step": 1303 + }, + { + "epoch": 0.12734375, + "grad_norm": 0.40479159355163574, + "learning_rate": 0.0004856317392368084, + "loss": 4.7578, + "step": 1304 + }, + { + "epoch": 0.12744140625, + "grad_norm": 0.38542646169662476, + "learning_rate": 0.0004856070958427979, + "loss": 4.7969, + "step": 1305 + }, + { + "epoch": 0.1275390625, + "grad_norm": 0.41619765758514404, + "learning_rate": 0.0004855824320318589, + "loss": 4.7578, + "step": 1306 + }, + { + "epoch": 0.12763671875, + "grad_norm": 0.5476976633071899, + "learning_rate": 0.0004855577478063822, + "loss": 4.7656, + "step": 1307 + }, + { + "epoch": 0.127734375, + "grad_norm": 0.7227765917778015, + "learning_rate": 0.000485533043168761, + "loss": 4.7734, + "step": 1308 + }, + { + "epoch": 0.12783203125, + "grad_norm": 0.6441882848739624, + "learning_rate": 0.0004855083181213902, + "loss": 4.7695, + "step": 1309 + }, + { + "epoch": 0.1279296875, + "grad_norm": 0.5060856938362122, + "learning_rate": 0.00048548357266666657, + "loss": 4.793, + "step": 1310 + }, + { + "epoch": 0.12802734375, + "grad_norm": 0.46759769320487976, + "learning_rate": 0.0004854588068069892, + "loss": 4.7773, + "step": 1311 + }, + { + "epoch": 0.128125, + "grad_norm": 0.40848907828330994, + "learning_rate": 0.0004854340205447589, + "loss": 4.7578, + "step": 1312 + }, + { + "epoch": 0.12822265625, + "grad_norm": 0.3958721160888672, + "learning_rate": 0.00048540921388237856, + "loss": 4.7773, + "step": 1313 + }, + { + "epoch": 0.1283203125, + "grad_norm": 0.4191996157169342, + "learning_rate": 0.0004853843868222529, + "loss": 4.7891, + "step": 1314 + }, + { + "epoch": 0.12841796875, + "grad_norm": 0.38634946942329407, + "learning_rate": 0.00048535953936678885, + "loss": 4.8047, + "step": 1315 + }, + { + "epoch": 0.128515625, + "grad_norm": 0.36390602588653564, + "learning_rate": 0.00048533467151839517, + "loss": 4.7656, + "step": 1316 + }, + { + "epoch": 0.12861328125, + "grad_norm": 0.32959216833114624, + "learning_rate": 0.0004853097832794827, + "loss": 4.7461, + "step": 1317 + }, + { + "epoch": 0.1287109375, + "grad_norm": 0.3805353045463562, + "learning_rate": 0.000485284874652464, + "loss": 4.7695, + "step": 1318 + }, + { + "epoch": 0.12880859375, + "grad_norm": 0.37332767248153687, + "learning_rate": 0.000485259945639754, + "loss": 4.7734, + "step": 1319 + }, + { + "epoch": 0.12890625, + "grad_norm": 0.3370415270328522, + "learning_rate": 0.00048523499624376925, + "loss": 4.7695, + "step": 1320 + }, + { + "epoch": 0.12900390625, + "grad_norm": 0.32865846157073975, + "learning_rate": 0.00048521002646692855, + "loss": 4.8008, + "step": 1321 + }, + { + "epoch": 0.1291015625, + "grad_norm": 0.35084402561187744, + "learning_rate": 0.0004851850363116524, + "loss": 4.7656, + "step": 1322 + }, + { + "epoch": 0.12919921875, + "grad_norm": 0.38480809330940247, + "learning_rate": 0.0004851600257803636, + "loss": 4.7734, + "step": 1323 + }, + { + "epoch": 0.129296875, + "grad_norm": 0.36452239751815796, + "learning_rate": 0.00048513499487548665, + "loss": 4.7461, + "step": 1324 + }, + { + "epoch": 0.12939453125, + "grad_norm": 0.36055848002433777, + "learning_rate": 0.00048510994359944804, + "loss": 4.7656, + "step": 1325 + }, + { + "epoch": 0.1294921875, + "grad_norm": 0.3260059952735901, + "learning_rate": 0.00048508487195467653, + "loss": 4.7344, + "step": 1326 + }, + { + "epoch": 0.12958984375, + "grad_norm": 0.339093953371048, + "learning_rate": 0.0004850597799436025, + "loss": 4.7617, + "step": 1327 + }, + { + "epoch": 0.1296875, + "grad_norm": 0.40155380964279175, + "learning_rate": 0.00048503466756865847, + "loss": 4.7695, + "step": 1328 + }, + { + "epoch": 0.12978515625, + "grad_norm": 0.578285813331604, + "learning_rate": 0.00048500953483227895, + "loss": 4.7344, + "step": 1329 + }, + { + "epoch": 0.1298828125, + "grad_norm": 0.7531477212905884, + "learning_rate": 0.0004849843817369003, + "loss": 4.7812, + "step": 1330 + }, + { + "epoch": 0.12998046875, + "grad_norm": 0.6945119500160217, + "learning_rate": 0.0004849592082849611, + "loss": 4.7617, + "step": 1331 + }, + { + "epoch": 0.130078125, + "grad_norm": 0.40566393733024597, + "learning_rate": 0.0004849340144789016, + "loss": 4.7969, + "step": 1332 + }, + { + "epoch": 0.13017578125, + "grad_norm": 0.4642469584941864, + "learning_rate": 0.00048490880032116425, + "loss": 4.7539, + "step": 1333 + }, + { + "epoch": 0.1302734375, + "grad_norm": 0.5508970618247986, + "learning_rate": 0.0004848835658141934, + "loss": 4.7539, + "step": 1334 + }, + { + "epoch": 0.13037109375, + "grad_norm": 0.4666309356689453, + "learning_rate": 0.00048485831096043526, + "loss": 4.8047, + "step": 1335 + }, + { + "epoch": 0.13046875, + "grad_norm": 0.3999099135398865, + "learning_rate": 0.0004848330357623382, + "loss": 4.7617, + "step": 1336 + }, + { + "epoch": 0.13056640625, + "grad_norm": 0.41524437069892883, + "learning_rate": 0.0004848077402223524, + "loss": 4.75, + "step": 1337 + }, + { + "epoch": 0.1306640625, + "grad_norm": 0.37947431206703186, + "learning_rate": 0.0004847824243429302, + "loss": 4.7734, + "step": 1338 + }, + { + "epoch": 0.13076171875, + "grad_norm": 0.39928561449050903, + "learning_rate": 0.0004847570881265259, + "loss": 4.7852, + "step": 1339 + }, + { + "epoch": 0.130859375, + "grad_norm": 0.3333352208137512, + "learning_rate": 0.0004847317315755953, + "loss": 4.7891, + "step": 1340 + }, + { + "epoch": 0.13095703125, + "grad_norm": 0.3555909991264343, + "learning_rate": 0.00048470635469259697, + "loss": 4.7344, + "step": 1341 + }, + { + "epoch": 0.1310546875, + "grad_norm": 0.3695686459541321, + "learning_rate": 0.00048468095747999067, + "loss": 4.75, + "step": 1342 + }, + { + "epoch": 0.13115234375, + "grad_norm": 0.32535481452941895, + "learning_rate": 0.00048465553994023875, + "loss": 4.7852, + "step": 1343 + }, + { + "epoch": 0.13125, + "grad_norm": 0.3166685402393341, + "learning_rate": 0.00048463010207580517, + "loss": 4.7656, + "step": 1344 + }, + { + "epoch": 0.13134765625, + "grad_norm": 0.30036693811416626, + "learning_rate": 0.000484604643889156, + "loss": 4.7773, + "step": 1345 + }, + { + "epoch": 0.1314453125, + "grad_norm": 0.3111598491668701, + "learning_rate": 0.0004845791653827591, + "loss": 4.7695, + "step": 1346 + }, + { + "epoch": 0.13154296875, + "grad_norm": 0.3160509765148163, + "learning_rate": 0.00048455366655908455, + "loss": 4.7695, + "step": 1347 + }, + { + "epoch": 0.131640625, + "grad_norm": 0.3423873782157898, + "learning_rate": 0.0004845281474206043, + "loss": 4.7422, + "step": 1348 + }, + { + "epoch": 0.13173828125, + "grad_norm": 0.34536194801330566, + "learning_rate": 0.00048450260796979223, + "loss": 4.7891, + "step": 1349 + }, + { + "epoch": 0.1318359375, + "grad_norm": 0.347699373960495, + "learning_rate": 0.0004844770482091242, + "loss": 4.7773, + "step": 1350 + }, + { + "epoch": 0.13193359375, + "grad_norm": 0.3164156675338745, + "learning_rate": 0.00048445146814107804, + "loss": 4.7656, + "step": 1351 + }, + { + "epoch": 0.13203125, + "grad_norm": 0.2843954265117645, + "learning_rate": 0.00048442586776813363, + "loss": 4.7695, + "step": 1352 + }, + { + "epoch": 0.13212890625, + "grad_norm": 0.34582918882369995, + "learning_rate": 0.00048440024709277274, + "loss": 4.7734, + "step": 1353 + }, + { + "epoch": 0.1322265625, + "grad_norm": 0.4413771331310272, + "learning_rate": 0.00048437460611747916, + "loss": 4.7461, + "step": 1354 + }, + { + "epoch": 0.13232421875, + "grad_norm": 0.4797145426273346, + "learning_rate": 0.0004843489448447385, + "loss": 4.793, + "step": 1355 + }, + { + "epoch": 0.132421875, + "grad_norm": 0.47246068716049194, + "learning_rate": 0.0004843232632770384, + "loss": 4.8008, + "step": 1356 + }, + { + "epoch": 0.13251953125, + "grad_norm": 0.39250800013542175, + "learning_rate": 0.00048429756141686863, + "loss": 4.7773, + "step": 1357 + }, + { + "epoch": 0.1326171875, + "grad_norm": 0.3332385718822479, + "learning_rate": 0.00048427183926672083, + "loss": 4.7773, + "step": 1358 + }, + { + "epoch": 0.13271484375, + "grad_norm": 0.3847937285900116, + "learning_rate": 0.00048424609682908856, + "loss": 4.7617, + "step": 1359 + }, + { + "epoch": 0.1328125, + "grad_norm": 0.3584742546081543, + "learning_rate": 0.0004842203341064673, + "loss": 4.7734, + "step": 1360 + }, + { + "epoch": 0.13291015625, + "grad_norm": 0.34857839345932007, + "learning_rate": 0.0004841945511013547, + "loss": 4.7344, + "step": 1361 + }, + { + "epoch": 0.1330078125, + "grad_norm": 0.3411064147949219, + "learning_rate": 0.00048416874781625016, + "loss": 4.793, + "step": 1362 + }, + { + "epoch": 0.13310546875, + "grad_norm": 0.3658686578273773, + "learning_rate": 0.00048414292425365507, + "loss": 4.7617, + "step": 1363 + }, + { + "epoch": 0.133203125, + "grad_norm": 0.40414950251579285, + "learning_rate": 0.00048411708041607305, + "loss": 4.7539, + "step": 1364 + }, + { + "epoch": 0.13330078125, + "grad_norm": 0.37495216727256775, + "learning_rate": 0.0004840912163060093, + "loss": 4.7773, + "step": 1365 + }, + { + "epoch": 0.1333984375, + "grad_norm": 0.40422677993774414, + "learning_rate": 0.00048406533192597124, + "loss": 4.7539, + "step": 1366 + }, + { + "epoch": 0.13349609375, + "grad_norm": 0.454258531332016, + "learning_rate": 0.0004840394272784682, + "loss": 4.7812, + "step": 1367 + }, + { + "epoch": 0.13359375, + "grad_norm": 0.39893728494644165, + "learning_rate": 0.00048401350236601146, + "loss": 4.7695, + "step": 1368 + }, + { + "epoch": 0.13369140625, + "grad_norm": 0.3277254104614258, + "learning_rate": 0.00048398755719111417, + "loss": 4.8086, + "step": 1369 + }, + { + "epoch": 0.1337890625, + "grad_norm": 0.436477392911911, + "learning_rate": 0.00048396159175629174, + "loss": 4.7461, + "step": 1370 + }, + { + "epoch": 0.13388671875, + "grad_norm": 0.33649933338165283, + "learning_rate": 0.00048393560606406114, + "loss": 4.7539, + "step": 1371 + }, + { + "epoch": 0.133984375, + "grad_norm": 0.34157708287239075, + "learning_rate": 0.0004839096001169416, + "loss": 4.7539, + "step": 1372 + }, + { + "epoch": 0.13408203125, + "grad_norm": 0.4028213322162628, + "learning_rate": 0.0004838835739174541, + "loss": 4.7773, + "step": 1373 + }, + { + "epoch": 0.1341796875, + "grad_norm": 0.46322396397590637, + "learning_rate": 0.0004838575274681219, + "loss": 4.7617, + "step": 1374 + }, + { + "epoch": 0.13427734375, + "grad_norm": 0.5127856731414795, + "learning_rate": 0.0004838314607714699, + "loss": 4.7266, + "step": 1375 + }, + { + "epoch": 0.134375, + "grad_norm": 0.5536565780639648, + "learning_rate": 0.00048380537383002517, + "loss": 4.75, + "step": 1376 + }, + { + "epoch": 0.13447265625, + "grad_norm": 0.5256456732749939, + "learning_rate": 0.0004837792666463166, + "loss": 4.7695, + "step": 1377 + }, + { + "epoch": 0.1345703125, + "grad_norm": 0.5643973350524902, + "learning_rate": 0.00048375313922287505, + "loss": 4.7656, + "step": 1378 + }, + { + "epoch": 0.13466796875, + "grad_norm": 0.8128441572189331, + "learning_rate": 0.00048372699156223355, + "loss": 4.7734, + "step": 1379 + }, + { + "epoch": 0.134765625, + "grad_norm": 1.026166558265686, + "learning_rate": 0.0004837008236669268, + "loss": 4.7227, + "step": 1380 + }, + { + "epoch": 0.13486328125, + "grad_norm": 0.4733608365058899, + "learning_rate": 0.00048367463553949166, + "loss": 4.7734, + "step": 1381 + }, + { + "epoch": 0.1349609375, + "grad_norm": 0.46688416600227356, + "learning_rate": 0.00048364842718246685, + "loss": 4.7461, + "step": 1382 + }, + { + "epoch": 0.13505859375, + "grad_norm": 0.42120811343193054, + "learning_rate": 0.00048362219859839317, + "loss": 4.7734, + "step": 1383 + }, + { + "epoch": 0.13515625, + "grad_norm": 0.4719356298446655, + "learning_rate": 0.0004835959497898133, + "loss": 4.7695, + "step": 1384 + }, + { + "epoch": 0.13525390625, + "grad_norm": 0.5680752992630005, + "learning_rate": 0.0004835696807592718, + "loss": 4.7383, + "step": 1385 + }, + { + "epoch": 0.1353515625, + "grad_norm": 0.7138311862945557, + "learning_rate": 0.0004835433915093153, + "loss": 4.7812, + "step": 1386 + }, + { + "epoch": 0.13544921875, + "grad_norm": 0.8844993114471436, + "learning_rate": 0.00048351708204249247, + "loss": 4.7383, + "step": 1387 + }, + { + "epoch": 0.135546875, + "grad_norm": 1.3054742813110352, + "learning_rate": 0.00048349075236135366, + "loss": 4.7773, + "step": 1388 + }, + { + "epoch": 0.13564453125, + "grad_norm": 0.5796347260475159, + "learning_rate": 0.0004834644024684515, + "loss": 4.7578, + "step": 1389 + }, + { + "epoch": 0.1357421875, + "grad_norm": 0.5956249833106995, + "learning_rate": 0.00048343803236634023, + "loss": 4.7461, + "step": 1390 + }, + { + "epoch": 0.13583984375, + "grad_norm": 0.53680819272995, + "learning_rate": 0.00048341164205757654, + "loss": 4.7773, + "step": 1391 + }, + { + "epoch": 0.1359375, + "grad_norm": 0.4875905513763428, + "learning_rate": 0.0004833852315447186, + "loss": 4.75, + "step": 1392 + }, + { + "epoch": 0.13603515625, + "grad_norm": 0.4100656807422638, + "learning_rate": 0.0004833588008303267, + "loss": 4.7305, + "step": 1393 + }, + { + "epoch": 0.1361328125, + "grad_norm": 0.344696044921875, + "learning_rate": 0.00048333234991696335, + "loss": 4.7891, + "step": 1394 + }, + { + "epoch": 0.13623046875, + "grad_norm": 0.37795841693878174, + "learning_rate": 0.0004833058788071925, + "loss": 4.7383, + "step": 1395 + }, + { + "epoch": 0.136328125, + "grad_norm": 0.41078394651412964, + "learning_rate": 0.0004832793875035805, + "loss": 4.7148, + "step": 1396 + }, + { + "epoch": 0.13642578125, + "grad_norm": 0.3967403471469879, + "learning_rate": 0.0004832528760086956, + "loss": 4.7734, + "step": 1397 + }, + { + "epoch": 0.1365234375, + "grad_norm": 0.3479665517807007, + "learning_rate": 0.00048322634432510766, + "loss": 4.7539, + "step": 1398 + }, + { + "epoch": 0.13662109375, + "grad_norm": 0.3690146803855896, + "learning_rate": 0.00048319979245538887, + "loss": 4.7695, + "step": 1399 + }, + { + "epoch": 0.13671875, + "grad_norm": 0.3363431394100189, + "learning_rate": 0.0004831732204021134, + "loss": 4.7734, + "step": 1400 + }, + { + "epoch": 0.13681640625, + "grad_norm": 0.32045620679855347, + "learning_rate": 0.00048314662816785687, + "loss": 4.7734, + "step": 1401 + }, + { + "epoch": 0.1369140625, + "grad_norm": 0.3061015009880066, + "learning_rate": 0.00048312001575519757, + "loss": 4.7656, + "step": 1402 + }, + { + "epoch": 0.13701171875, + "grad_norm": 0.38208630681037903, + "learning_rate": 0.0004830933831667152, + "loss": 4.7539, + "step": 1403 + }, + { + "epoch": 0.137109375, + "grad_norm": 0.3805338442325592, + "learning_rate": 0.0004830667304049918, + "loss": 4.7461, + "step": 1404 + }, + { + "epoch": 0.13720703125, + "grad_norm": 0.5099437236785889, + "learning_rate": 0.0004830400574726109, + "loss": 4.7812, + "step": 1405 + }, + { + "epoch": 0.1373046875, + "grad_norm": 0.6449255347251892, + "learning_rate": 0.00048301336437215844, + "loss": 4.7812, + "step": 1406 + }, + { + "epoch": 0.13740234375, + "grad_norm": 0.49494004249572754, + "learning_rate": 0.0004829866511062221, + "loss": 4.7734, + "step": 1407 + }, + { + "epoch": 0.1375, + "grad_norm": 0.34257587790489197, + "learning_rate": 0.0004829599176773916, + "loss": 4.7617, + "step": 1408 + }, + { + "epoch": 0.13759765625, + "grad_norm": 0.37970221042633057, + "learning_rate": 0.0004829331640882584, + "loss": 4.7539, + "step": 1409 + }, + { + "epoch": 0.1376953125, + "grad_norm": 0.48333415389060974, + "learning_rate": 0.0004829063903414162, + "loss": 4.7773, + "step": 1410 + }, + { + "epoch": 0.13779296875, + "grad_norm": 0.3654765486717224, + "learning_rate": 0.00048287959643946056, + "loss": 4.7578, + "step": 1411 + }, + { + "epoch": 0.137890625, + "grad_norm": 0.5381461381912231, + "learning_rate": 0.0004828527823849889, + "loss": 4.7695, + "step": 1412 + }, + { + "epoch": 0.13798828125, + "grad_norm": 0.5346956849098206, + "learning_rate": 0.0004828259481806007, + "loss": 4.7383, + "step": 1413 + }, + { + "epoch": 0.1380859375, + "grad_norm": 0.5548373460769653, + "learning_rate": 0.00048279909382889735, + "loss": 4.7773, + "step": 1414 + }, + { + "epoch": 0.13818359375, + "grad_norm": 0.6224201917648315, + "learning_rate": 0.00048277221933248226, + "loss": 4.7422, + "step": 1415 + }, + { + "epoch": 0.13828125, + "grad_norm": 0.5974454283714294, + "learning_rate": 0.0004827453246939606, + "loss": 4.7734, + "step": 1416 + }, + { + "epoch": 0.13837890625, + "grad_norm": 0.4552706182003021, + "learning_rate": 0.00048271840991593966, + "loss": 4.75, + "step": 1417 + }, + { + "epoch": 0.1384765625, + "grad_norm": 0.5470201373100281, + "learning_rate": 0.00048269147500102873, + "loss": 4.7695, + "step": 1418 + }, + { + "epoch": 0.13857421875, + "grad_norm": 0.48209336400032043, + "learning_rate": 0.00048266451995183885, + "loss": 4.793, + "step": 1419 + }, + { + "epoch": 0.138671875, + "grad_norm": 0.43414023518562317, + "learning_rate": 0.0004826375447709832, + "loss": 4.7695, + "step": 1420 + }, + { + "epoch": 0.13876953125, + "grad_norm": 0.3743106424808502, + "learning_rate": 0.00048261054946107686, + "loss": 4.75, + "step": 1421 + }, + { + "epoch": 0.1388671875, + "grad_norm": 0.3445884585380554, + "learning_rate": 0.0004825835340247368, + "loss": 4.7188, + "step": 1422 + }, + { + "epoch": 0.13896484375, + "grad_norm": 0.3742680847644806, + "learning_rate": 0.00048255649846458187, + "loss": 4.7617, + "step": 1423 + }, + { + "epoch": 0.1390625, + "grad_norm": 0.4068181812763214, + "learning_rate": 0.00048252944278323324, + "loss": 4.7227, + "step": 1424 + }, + { + "epoch": 0.13916015625, + "grad_norm": 0.33727332949638367, + "learning_rate": 0.0004825023669833136, + "loss": 4.7734, + "step": 1425 + }, + { + "epoch": 0.1392578125, + "grad_norm": 0.27452924847602844, + "learning_rate": 0.00048247527106744784, + "loss": 4.7539, + "step": 1426 + }, + { + "epoch": 0.13935546875, + "grad_norm": 0.323626309633255, + "learning_rate": 0.00048244815503826257, + "loss": 4.7422, + "step": 1427 + }, + { + "epoch": 0.139453125, + "grad_norm": 0.33976978063583374, + "learning_rate": 0.0004824210188983867, + "loss": 4.7617, + "step": 1428 + }, + { + "epoch": 0.13955078125, + "grad_norm": 0.3159013092517853, + "learning_rate": 0.0004823938626504508, + "loss": 4.7617, + "step": 1429 + }, + { + "epoch": 0.1396484375, + "grad_norm": 0.3308809995651245, + "learning_rate": 0.0004823666862970876, + "loss": 4.7734, + "step": 1430 + }, + { + "epoch": 0.13974609375, + "grad_norm": 0.37637704610824585, + "learning_rate": 0.00048233948984093147, + "loss": 4.7617, + "step": 1431 + }, + { + "epoch": 0.13984375, + "grad_norm": 0.3979220986366272, + "learning_rate": 0.000482312273284619, + "loss": 4.7617, + "step": 1432 + }, + { + "epoch": 0.13994140625, + "grad_norm": 0.41099435091018677, + "learning_rate": 0.0004822850366307887, + "loss": 4.7734, + "step": 1433 + }, + { + "epoch": 0.1400390625, + "grad_norm": 0.4159080684185028, + "learning_rate": 0.00048225777988208094, + "loss": 4.75, + "step": 1434 + }, + { + "epoch": 0.14013671875, + "grad_norm": 0.440417617559433, + "learning_rate": 0.0004822305030411381, + "loss": 4.7344, + "step": 1435 + }, + { + "epoch": 0.140234375, + "grad_norm": 0.35094818472862244, + "learning_rate": 0.0004822032061106045, + "loss": 4.7422, + "step": 1436 + }, + { + "epoch": 0.14033203125, + "grad_norm": 0.33545538783073425, + "learning_rate": 0.00048217588909312627, + "loss": 4.7656, + "step": 1437 + }, + { + "epoch": 0.1404296875, + "grad_norm": 0.363555908203125, + "learning_rate": 0.0004821485519913518, + "loss": 4.7305, + "step": 1438 + }, + { + "epoch": 0.14052734375, + "grad_norm": 0.39379504323005676, + "learning_rate": 0.00048212119480793103, + "loss": 4.7656, + "step": 1439 + }, + { + "epoch": 0.140625, + "grad_norm": 0.3747332692146301, + "learning_rate": 0.00048209381754551616, + "loss": 4.7266, + "step": 1440 + }, + { + "epoch": 0.14072265625, + "grad_norm": 0.38222065567970276, + "learning_rate": 0.00048206642020676133, + "loss": 4.7656, + "step": 1441 + }, + { + "epoch": 0.1408203125, + "grad_norm": 0.4481048583984375, + "learning_rate": 0.00048203900279432247, + "loss": 4.7383, + "step": 1442 + }, + { + "epoch": 0.14091796875, + "grad_norm": 0.5370760560035706, + "learning_rate": 0.0004820115653108573, + "loss": 4.75, + "step": 1443 + }, + { + "epoch": 0.141015625, + "grad_norm": 0.5712428092956543, + "learning_rate": 0.00048198410775902596, + "loss": 4.7539, + "step": 1444 + }, + { + "epoch": 0.14111328125, + "grad_norm": 0.4819755256175995, + "learning_rate": 0.0004819566301414901, + "loss": 4.75, + "step": 1445 + }, + { + "epoch": 0.1412109375, + "grad_norm": 0.39142805337905884, + "learning_rate": 0.0004819291324609137, + "loss": 4.7188, + "step": 1446 + }, + { + "epoch": 0.14130859375, + "grad_norm": 0.391284316778183, + "learning_rate": 0.0004819016147199622, + "loss": 4.7539, + "step": 1447 + }, + { + "epoch": 0.14140625, + "grad_norm": 0.37572604417800903, + "learning_rate": 0.00048187407692130345, + "loss": 4.7539, + "step": 1448 + }, + { + "epoch": 0.14150390625, + "grad_norm": 0.40346717834472656, + "learning_rate": 0.000481846519067607, + "loss": 4.7383, + "step": 1449 + }, + { + "epoch": 0.1416015625, + "grad_norm": 0.39226216077804565, + "learning_rate": 0.00048181894116154444, + "loss": 4.7656, + "step": 1450 + }, + { + "epoch": 0.14169921875, + "grad_norm": 0.3652878403663635, + "learning_rate": 0.0004817913432057892, + "loss": 4.7461, + "step": 1451 + }, + { + "epoch": 0.141796875, + "grad_norm": 0.3091000020503998, + "learning_rate": 0.0004817637252030167, + "loss": 4.7734, + "step": 1452 + }, + { + "epoch": 0.14189453125, + "grad_norm": 0.2839193344116211, + "learning_rate": 0.00048173608715590437, + "loss": 4.7539, + "step": 1453 + }, + { + "epoch": 0.1419921875, + "grad_norm": 0.3217602074146271, + "learning_rate": 0.0004817084290671315, + "loss": 4.7031, + "step": 1454 + }, + { + "epoch": 0.14208984375, + "grad_norm": 0.3004675805568695, + "learning_rate": 0.00048168075093937943, + "loss": 4.7109, + "step": 1455 + }, + { + "epoch": 0.1421875, + "grad_norm": 0.4077158570289612, + "learning_rate": 0.0004816530527753312, + "loss": 4.7539, + "step": 1456 + }, + { + "epoch": 0.14228515625, + "grad_norm": 0.3414487838745117, + "learning_rate": 0.0004816253345776722, + "loss": 4.7695, + "step": 1457 + }, + { + "epoch": 0.1423828125, + "grad_norm": 0.30606889724731445, + "learning_rate": 0.0004815975963490893, + "loss": 4.7188, + "step": 1458 + }, + { + "epoch": 0.14248046875, + "grad_norm": 0.35760390758514404, + "learning_rate": 0.0004815698380922716, + "loss": 4.7695, + "step": 1459 + }, + { + "epoch": 0.142578125, + "grad_norm": 0.4264749586582184, + "learning_rate": 0.0004815420598099101, + "loss": 4.7344, + "step": 1460 + }, + { + "epoch": 0.14267578125, + "grad_norm": 0.48403796553611755, + "learning_rate": 0.00048151426150469767, + "loss": 4.7656, + "step": 1461 + }, + { + "epoch": 0.1427734375, + "grad_norm": 0.4852380156517029, + "learning_rate": 0.00048148644317932927, + "loss": 4.75, + "step": 1462 + }, + { + "epoch": 0.14287109375, + "grad_norm": 0.46373042464256287, + "learning_rate": 0.0004814586048365016, + "loss": 4.7539, + "step": 1463 + }, + { + "epoch": 0.14296875, + "grad_norm": 0.41835281252861023, + "learning_rate": 0.00048143074647891344, + "loss": 4.7422, + "step": 1464 + }, + { + "epoch": 0.14306640625, + "grad_norm": 0.34529751539230347, + "learning_rate": 0.00048140286810926543, + "loss": 4.7578, + "step": 1465 + }, + { + "epoch": 0.1431640625, + "grad_norm": 0.32219383120536804, + "learning_rate": 0.0004813749697302603, + "loss": 4.7695, + "step": 1466 + }, + { + "epoch": 0.14326171875, + "grad_norm": 0.33539384603500366, + "learning_rate": 0.0004813470513446025, + "loss": 4.7617, + "step": 1467 + }, + { + "epoch": 0.143359375, + "grad_norm": 0.27315089106559753, + "learning_rate": 0.00048131911295499845, + "loss": 4.7578, + "step": 1468 + }, + { + "epoch": 0.14345703125, + "grad_norm": 0.29842984676361084, + "learning_rate": 0.00048129115456415684, + "loss": 4.7188, + "step": 1469 + }, + { + "epoch": 0.1435546875, + "grad_norm": 0.3583134710788727, + "learning_rate": 0.00048126317617478776, + "loss": 4.7734, + "step": 1470 + }, + { + "epoch": 0.14365234375, + "grad_norm": 0.41695231199264526, + "learning_rate": 0.00048123517778960373, + "loss": 4.7617, + "step": 1471 + }, + { + "epoch": 0.14375, + "grad_norm": 0.5007773637771606, + "learning_rate": 0.00048120715941131893, + "loss": 4.7305, + "step": 1472 + }, + { + "epoch": 0.14384765625, + "grad_norm": 0.6086747050285339, + "learning_rate": 0.00048117912104264957, + "loss": 4.7734, + "step": 1473 + }, + { + "epoch": 0.1439453125, + "grad_norm": 0.5901089310646057, + "learning_rate": 0.00048115106268631374, + "loss": 4.7344, + "step": 1474 + }, + { + "epoch": 0.14404296875, + "grad_norm": 0.5069621205329895, + "learning_rate": 0.00048112298434503154, + "loss": 4.7773, + "step": 1475 + }, + { + "epoch": 0.144140625, + "grad_norm": 0.4707411527633667, + "learning_rate": 0.00048109488602152503, + "loss": 4.7578, + "step": 1476 + }, + { + "epoch": 0.14423828125, + "grad_norm": 0.4005588889122009, + "learning_rate": 0.00048106676771851804, + "loss": 4.7422, + "step": 1477 + }, + { + "epoch": 0.1443359375, + "grad_norm": 0.3813304007053375, + "learning_rate": 0.00048103862943873647, + "loss": 4.7461, + "step": 1478 + }, + { + "epoch": 0.14443359375, + "grad_norm": 0.3576384484767914, + "learning_rate": 0.0004810104711849082, + "loss": 4.7695, + "step": 1479 + }, + { + "epoch": 0.14453125, + "grad_norm": 0.3286953270435333, + "learning_rate": 0.0004809822929597629, + "loss": 4.7461, + "step": 1480 + }, + { + "epoch": 0.14462890625, + "grad_norm": 0.32706767320632935, + "learning_rate": 0.00048095409476603233, + "loss": 4.75, + "step": 1481 + }, + { + "epoch": 0.1447265625, + "grad_norm": 0.33062857389450073, + "learning_rate": 0.0004809258766064501, + "loss": 4.7578, + "step": 1482 + }, + { + "epoch": 0.14482421875, + "grad_norm": 0.32870644330978394, + "learning_rate": 0.0004808976384837518, + "loss": 4.7461, + "step": 1483 + }, + { + "epoch": 0.144921875, + "grad_norm": 0.28449103236198425, + "learning_rate": 0.00048086938040067486, + "loss": 4.7539, + "step": 1484 + }, + { + "epoch": 0.14501953125, + "grad_norm": 0.2756710946559906, + "learning_rate": 0.00048084110235995864, + "loss": 4.7266, + "step": 1485 + }, + { + "epoch": 0.1451171875, + "grad_norm": 0.28041645884513855, + "learning_rate": 0.00048081280436434467, + "loss": 4.7188, + "step": 1486 + }, + { + "epoch": 0.14521484375, + "grad_norm": 0.3484303951263428, + "learning_rate": 0.0004807844864165761, + "loss": 4.7383, + "step": 1487 + }, + { + "epoch": 0.1453125, + "grad_norm": 0.4143260419368744, + "learning_rate": 0.0004807561485193983, + "loss": 4.7578, + "step": 1488 + }, + { + "epoch": 0.14541015625, + "grad_norm": 0.42858263850212097, + "learning_rate": 0.00048072779067555847, + "loss": 4.7188, + "step": 1489 + }, + { + "epoch": 0.1455078125, + "grad_norm": 0.482190877199173, + "learning_rate": 0.00048069941288780545, + "loss": 4.7578, + "step": 1490 + }, + { + "epoch": 0.14560546875, + "grad_norm": 0.4588344991207123, + "learning_rate": 0.0004806710151588905, + "loss": 4.7266, + "step": 1491 + }, + { + "epoch": 0.145703125, + "grad_norm": 0.4522359371185303, + "learning_rate": 0.0004806425974915665, + "loss": 4.7773, + "step": 1492 + }, + { + "epoch": 0.14580078125, + "grad_norm": 0.47912994027137756, + "learning_rate": 0.0004806141598885884, + "loss": 4.707, + "step": 1493 + }, + { + "epoch": 0.1458984375, + "grad_norm": 0.4162638187408447, + "learning_rate": 0.000480585702352713, + "loss": 4.7422, + "step": 1494 + }, + { + "epoch": 0.14599609375, + "grad_norm": 0.3735276460647583, + "learning_rate": 0.0004805572248866991, + "loss": 4.7422, + "step": 1495 + }, + { + "epoch": 0.14609375, + "grad_norm": 0.33471450209617615, + "learning_rate": 0.0004805287274933073, + "loss": 4.7344, + "step": 1496 + }, + { + "epoch": 0.14619140625, + "grad_norm": 0.3382073640823364, + "learning_rate": 0.00048050021017530037, + "loss": 4.7461, + "step": 1497 + }, + { + "epoch": 0.1462890625, + "grad_norm": 0.35437050461769104, + "learning_rate": 0.0004804716729354428, + "loss": 4.7109, + "step": 1498 + }, + { + "epoch": 0.14638671875, + "grad_norm": 0.3948093056678772, + "learning_rate": 0.000480443115776501, + "loss": 4.707, + "step": 1499 + }, + { + "epoch": 0.146484375, + "grad_norm": 0.41001394391059875, + "learning_rate": 0.0004804145387012435, + "loss": 4.7852, + "step": 1500 + }, + { + "epoch": 0.14658203125, + "grad_norm": 0.41534173488616943, + "learning_rate": 0.00048038594171244067, + "loss": 4.7383, + "step": 1501 + }, + { + "epoch": 0.1466796875, + "grad_norm": 0.3207942247390747, + "learning_rate": 0.0004803573248128647, + "loss": 4.75, + "step": 1502 + }, + { + "epoch": 0.14677734375, + "grad_norm": 0.32587486505508423, + "learning_rate": 0.0004803286880052899, + "loss": 4.7461, + "step": 1503 + }, + { + "epoch": 0.146875, + "grad_norm": 0.33651435375213623, + "learning_rate": 0.0004803000312924923, + "loss": 4.7617, + "step": 1504 + }, + { + "epoch": 0.14697265625, + "grad_norm": 0.3294571340084076, + "learning_rate": 0.00048027135467725013, + "loss": 4.7422, + "step": 1505 + }, + { + "epoch": 0.1470703125, + "grad_norm": 0.32122352719306946, + "learning_rate": 0.00048024265816234323, + "loss": 4.7344, + "step": 1506 + }, + { + "epoch": 0.14716796875, + "grad_norm": 0.33909428119659424, + "learning_rate": 0.00048021394175055363, + "loss": 4.7461, + "step": 1507 + }, + { + "epoch": 0.147265625, + "grad_norm": 0.47826051712036133, + "learning_rate": 0.00048018520544466513, + "loss": 4.7422, + "step": 1508 + }, + { + "epoch": 0.14736328125, + "grad_norm": 0.6739643812179565, + "learning_rate": 0.00048015644924746364, + "loss": 4.7031, + "step": 1509 + }, + { + "epoch": 0.1474609375, + "grad_norm": 0.7271323204040527, + "learning_rate": 0.00048012767316173675, + "loss": 4.7461, + "step": 1510 + }, + { + "epoch": 0.14755859375, + "grad_norm": 0.632926881313324, + "learning_rate": 0.0004800988771902742, + "loss": 4.7148, + "step": 1511 + }, + { + "epoch": 0.14765625, + "grad_norm": 0.4113217294216156, + "learning_rate": 0.00048007006133586746, + "loss": 4.7344, + "step": 1512 + }, + { + "epoch": 0.14775390625, + "grad_norm": 0.37485432624816895, + "learning_rate": 0.00048004122560131016, + "loss": 4.7188, + "step": 1513 + }, + { + "epoch": 0.1478515625, + "grad_norm": 0.484247624874115, + "learning_rate": 0.00048001236998939756, + "loss": 4.7305, + "step": 1514 + }, + { + "epoch": 0.14794921875, + "grad_norm": 0.43032926321029663, + "learning_rate": 0.0004799834945029272, + "loss": 4.7344, + "step": 1515 + }, + { + "epoch": 0.148046875, + "grad_norm": 0.5900740623474121, + "learning_rate": 0.00047995459914469826, + "loss": 4.7695, + "step": 1516 + }, + { + "epoch": 0.14814453125, + "grad_norm": 0.31932970881462097, + "learning_rate": 0.0004799256839175119, + "loss": 4.7188, + "step": 1517 + }, + { + "epoch": 0.1482421875, + "grad_norm": 0.3383445739746094, + "learning_rate": 0.0004798967488241714, + "loss": 4.7305, + "step": 1518 + }, + { + "epoch": 0.14833984375, + "grad_norm": 0.3197227120399475, + "learning_rate": 0.00047986779386748166, + "loss": 4.75, + "step": 1519 + }, + { + "epoch": 0.1484375, + "grad_norm": 0.3146499693393707, + "learning_rate": 0.0004798388190502497, + "loss": 4.7422, + "step": 1520 + }, + { + "epoch": 0.14853515625, + "grad_norm": 0.3118762969970703, + "learning_rate": 0.0004798098243752846, + "loss": 4.7266, + "step": 1521 + }, + { + "epoch": 0.1486328125, + "grad_norm": 0.3077457547187805, + "learning_rate": 0.0004797808098453971, + "loss": 4.7305, + "step": 1522 + }, + { + "epoch": 0.14873046875, + "grad_norm": 0.3942987024784088, + "learning_rate": 0.0004797517754633998, + "loss": 4.7148, + "step": 1523 + }, + { + "epoch": 0.148828125, + "grad_norm": 0.6512936353683472, + "learning_rate": 0.0004797227212321076, + "loss": 4.7305, + "step": 1524 + }, + { + "epoch": 0.14892578125, + "grad_norm": 1.190845251083374, + "learning_rate": 0.00047969364715433707, + "loss": 4.7461, + "step": 1525 + }, + { + "epoch": 0.1490234375, + "grad_norm": 0.8728731870651245, + "learning_rate": 0.0004796645532329066, + "loss": 4.7734, + "step": 1526 + }, + { + "epoch": 0.14912109375, + "grad_norm": 0.551252543926239, + "learning_rate": 0.00047963543947063687, + "loss": 4.75, + "step": 1527 + }, + { + "epoch": 0.14921875, + "grad_norm": 0.6479698419570923, + "learning_rate": 0.0004796063058703501, + "loss": 4.7617, + "step": 1528 + }, + { + "epoch": 0.14931640625, + "grad_norm": 0.5585426688194275, + "learning_rate": 0.0004795771524348707, + "loss": 4.7578, + "step": 1529 + }, + { + "epoch": 0.1494140625, + "grad_norm": 0.5894700884819031, + "learning_rate": 0.0004795479791670247, + "loss": 4.7578, + "step": 1530 + }, + { + "epoch": 0.14951171875, + "grad_norm": 0.6230053305625916, + "learning_rate": 0.0004795187860696405, + "loss": 4.7461, + "step": 1531 + }, + { + "epoch": 0.149609375, + "grad_norm": 0.4770023226737976, + "learning_rate": 0.00047948957314554805, + "loss": 4.7305, + "step": 1532 + }, + { + "epoch": 0.14970703125, + "grad_norm": 0.3334863781929016, + "learning_rate": 0.00047946034039757934, + "loss": 4.7656, + "step": 1533 + }, + { + "epoch": 0.1498046875, + "grad_norm": 0.3881434500217438, + "learning_rate": 0.0004794310878285683, + "loss": 4.7227, + "step": 1534 + }, + { + "epoch": 0.14990234375, + "grad_norm": 0.3674808740615845, + "learning_rate": 0.0004794018154413508, + "loss": 4.7422, + "step": 1535 + }, + { + "epoch": 0.15, + "grad_norm": 0.3127332627773285, + "learning_rate": 0.0004793725232387646, + "loss": 4.7344, + "step": 1536 + }, + { + "epoch": 0.15009765625, + "grad_norm": 0.3168219327926636, + "learning_rate": 0.00047934321122364927, + "loss": 4.7344, + "step": 1537 + }, + { + "epoch": 0.1501953125, + "grad_norm": 0.3331605792045593, + "learning_rate": 0.00047931387939884653, + "loss": 4.7305, + "step": 1538 + }, + { + "epoch": 0.15029296875, + "grad_norm": 0.3181023597717285, + "learning_rate": 0.0004792845277671999, + "loss": 4.7188, + "step": 1539 + }, + { + "epoch": 0.150390625, + "grad_norm": 0.3030186891555786, + "learning_rate": 0.0004792551563315547, + "loss": 4.75, + "step": 1540 + }, + { + "epoch": 0.15048828125, + "grad_norm": 0.30499133467674255, + "learning_rate": 0.00047922576509475844, + "loss": 4.7539, + "step": 1541 + }, + { + "epoch": 0.1505859375, + "grad_norm": 0.30657604336738586, + "learning_rate": 0.0004791963540596603, + "loss": 4.7305, + "step": 1542 + }, + { + "epoch": 0.15068359375, + "grad_norm": 0.34554725885391235, + "learning_rate": 0.0004791669232291116, + "loss": 4.7031, + "step": 1543 + }, + { + "epoch": 0.15078125, + "grad_norm": 0.3566804528236389, + "learning_rate": 0.00047913747260596535, + "loss": 4.7031, + "step": 1544 + }, + { + "epoch": 0.15087890625, + "grad_norm": 0.41564902663230896, + "learning_rate": 0.00047910800219307657, + "loss": 4.7109, + "step": 1545 + }, + { + "epoch": 0.1509765625, + "grad_norm": 0.48682817816734314, + "learning_rate": 0.00047907851199330227, + "loss": 4.7188, + "step": 1546 + }, + { + "epoch": 0.15107421875, + "grad_norm": 0.5196331739425659, + "learning_rate": 0.00047904900200950134, + "loss": 4.7578, + "step": 1547 + }, + { + "epoch": 0.151171875, + "grad_norm": 0.47747114300727844, + "learning_rate": 0.0004790194722445345, + "loss": 4.7188, + "step": 1548 + }, + { + "epoch": 0.15126953125, + "grad_norm": 0.36361950635910034, + "learning_rate": 0.0004789899227012646, + "loss": 4.75, + "step": 1549 + }, + { + "epoch": 0.1513671875, + "grad_norm": 0.3436301052570343, + "learning_rate": 0.0004789603533825562, + "loss": 4.7617, + "step": 1550 + }, + { + "epoch": 0.15146484375, + "grad_norm": 0.3718210458755493, + "learning_rate": 0.00047893076429127575, + "loss": 4.7695, + "step": 1551 + }, + { + "epoch": 0.1515625, + "grad_norm": 0.4168568253517151, + "learning_rate": 0.00047890115543029174, + "loss": 4.7461, + "step": 1552 + }, + { + "epoch": 0.15166015625, + "grad_norm": 0.4080406129360199, + "learning_rate": 0.00047887152680247465, + "loss": 4.7188, + "step": 1553 + }, + { + "epoch": 0.1517578125, + "grad_norm": 0.4210541546344757, + "learning_rate": 0.0004788418784106967, + "loss": 4.7383, + "step": 1554 + }, + { + "epoch": 0.15185546875, + "grad_norm": 0.38915780186653137, + "learning_rate": 0.0004788122102578321, + "loss": 4.7266, + "step": 1555 + }, + { + "epoch": 0.151953125, + "grad_norm": 0.40426111221313477, + "learning_rate": 0.0004787825223467571, + "loss": 4.7031, + "step": 1556 + }, + { + "epoch": 0.15205078125, + "grad_norm": 0.3467988669872284, + "learning_rate": 0.0004787528146803495, + "loss": 4.7344, + "step": 1557 + }, + { + "epoch": 0.1521484375, + "grad_norm": 0.30397987365722656, + "learning_rate": 0.00047872308726148945, + "loss": 4.7422, + "step": 1558 + }, + { + "epoch": 0.15224609375, + "grad_norm": 0.2704983055591583, + "learning_rate": 0.00047869334009305873, + "loss": 4.7227, + "step": 1559 + }, + { + "epoch": 0.15234375, + "grad_norm": 0.28865277767181396, + "learning_rate": 0.00047866357317794125, + "loss": 4.7461, + "step": 1560 + }, + { + "epoch": 0.15244140625, + "grad_norm": 0.3506476581096649, + "learning_rate": 0.0004786337865190225, + "loss": 4.7539, + "step": 1561 + }, + { + "epoch": 0.1525390625, + "grad_norm": 0.38137906789779663, + "learning_rate": 0.00047860398011919026, + "loss": 4.7578, + "step": 1562 + }, + { + "epoch": 0.15263671875, + "grad_norm": 0.35326069593429565, + "learning_rate": 0.00047857415398133403, + "loss": 4.7305, + "step": 1563 + }, + { + "epoch": 0.152734375, + "grad_norm": 0.33928734064102173, + "learning_rate": 0.0004785443081083452, + "loss": 4.7266, + "step": 1564 + }, + { + "epoch": 0.15283203125, + "grad_norm": 0.5283599495887756, + "learning_rate": 0.0004785144425031172, + "loss": 4.7188, + "step": 1565 + }, + { + "epoch": 0.1529296875, + "grad_norm": 0.5691794157028198, + "learning_rate": 0.00047848455716854524, + "loss": 4.7266, + "step": 1566 + }, + { + "epoch": 0.15302734375, + "grad_norm": 0.4781295657157898, + "learning_rate": 0.0004784546521075265, + "loss": 4.7578, + "step": 1567 + }, + { + "epoch": 0.153125, + "grad_norm": 0.46981513500213623, + "learning_rate": 0.00047842472732296013, + "loss": 4.7344, + "step": 1568 + }, + { + "epoch": 0.15322265625, + "grad_norm": 0.4642745852470398, + "learning_rate": 0.0004783947828177471, + "loss": 4.7578, + "step": 1569 + }, + { + "epoch": 0.1533203125, + "grad_norm": 0.4203815758228302, + "learning_rate": 0.00047836481859479026, + "loss": 4.7305, + "step": 1570 + }, + { + "epoch": 0.15341796875, + "grad_norm": 0.472530335187912, + "learning_rate": 0.0004783348346569945, + "loss": 4.7578, + "step": 1571 + }, + { + "epoch": 0.153515625, + "grad_norm": 0.5050036311149597, + "learning_rate": 0.00047830483100726656, + "loss": 4.7227, + "step": 1572 + }, + { + "epoch": 0.15361328125, + "grad_norm": 0.4915096163749695, + "learning_rate": 0.0004782748076485151, + "loss": 4.7578, + "step": 1573 + }, + { + "epoch": 0.1537109375, + "grad_norm": 0.38450387120246887, + "learning_rate": 0.00047824476458365074, + "loss": 4.7344, + "step": 1574 + }, + { + "epoch": 0.15380859375, + "grad_norm": 0.3322133719921112, + "learning_rate": 0.0004782147018155858, + "loss": 4.75, + "step": 1575 + }, + { + "epoch": 0.15390625, + "grad_norm": 0.3817479908466339, + "learning_rate": 0.0004781846193472348, + "loss": 4.7461, + "step": 1576 + }, + { + "epoch": 0.15400390625, + "grad_norm": 0.411730021238327, + "learning_rate": 0.00047815451718151387, + "loss": 4.7148, + "step": 1577 + }, + { + "epoch": 0.1541015625, + "grad_norm": 0.43964841961860657, + "learning_rate": 0.00047812439532134144, + "loss": 4.7305, + "step": 1578 + }, + { + "epoch": 0.15419921875, + "grad_norm": 0.39244744181632996, + "learning_rate": 0.0004780942537696374, + "loss": 4.7344, + "step": 1579 + }, + { + "epoch": 0.154296875, + "grad_norm": 0.3104197084903717, + "learning_rate": 0.0004780640925293239, + "loss": 4.7344, + "step": 1580 + }, + { + "epoch": 0.15439453125, + "grad_norm": 0.34375083446502686, + "learning_rate": 0.00047803391160332483, + "loss": 4.7344, + "step": 1581 + }, + { + "epoch": 0.1544921875, + "grad_norm": 0.3174738585948944, + "learning_rate": 0.000478003710994566, + "loss": 4.6992, + "step": 1582 + }, + { + "epoch": 0.15458984375, + "grad_norm": 0.3029673993587494, + "learning_rate": 0.0004779734907059752, + "loss": 4.7227, + "step": 1583 + }, + { + "epoch": 0.1546875, + "grad_norm": 0.30898958444595337, + "learning_rate": 0.0004779432507404821, + "loss": 4.7148, + "step": 1584 + }, + { + "epoch": 0.15478515625, + "grad_norm": 0.31820470094680786, + "learning_rate": 0.00047791299110101814, + "loss": 4.707, + "step": 1585 + }, + { + "epoch": 0.1548828125, + "grad_norm": 0.32890087366104126, + "learning_rate": 0.0004778827117905169, + "loss": 4.7188, + "step": 1586 + }, + { + "epoch": 0.15498046875, + "grad_norm": 0.2951774597167969, + "learning_rate": 0.0004778524128119138, + "loss": 4.7344, + "step": 1587 + }, + { + "epoch": 0.155078125, + "grad_norm": 0.30525726079940796, + "learning_rate": 0.00047782209416814586, + "loss": 4.707, + "step": 1588 + }, + { + "epoch": 0.15517578125, + "grad_norm": 0.29748696088790894, + "learning_rate": 0.0004777917558621526, + "loss": 4.7188, + "step": 1589 + }, + { + "epoch": 0.1552734375, + "grad_norm": 0.35524943470954895, + "learning_rate": 0.0004777613978968749, + "loss": 4.7461, + "step": 1590 + }, + { + "epoch": 0.15537109375, + "grad_norm": 0.49947482347488403, + "learning_rate": 0.0004777310202752558, + "loss": 4.6875, + "step": 1591 + }, + { + "epoch": 0.15546875, + "grad_norm": 0.6647739410400391, + "learning_rate": 0.00047770062300024033, + "loss": 4.7383, + "step": 1592 + }, + { + "epoch": 0.15556640625, + "grad_norm": 0.6732831597328186, + "learning_rate": 0.00047767020607477505, + "loss": 4.75, + "step": 1593 + }, + { + "epoch": 0.1556640625, + "grad_norm": 0.4741779863834381, + "learning_rate": 0.0004776397695018089, + "loss": 4.7188, + "step": 1594 + }, + { + "epoch": 0.15576171875, + "grad_norm": 0.4117373824119568, + "learning_rate": 0.00047760931328429243, + "loss": 4.7383, + "step": 1595 + }, + { + "epoch": 0.155859375, + "grad_norm": 0.4464336037635803, + "learning_rate": 0.0004775788374251782, + "loss": 4.7461, + "step": 1596 + }, + { + "epoch": 0.15595703125, + "grad_norm": 0.3815647065639496, + "learning_rate": 0.00047754834192742046, + "loss": 4.75, + "step": 1597 + }, + { + "epoch": 0.1560546875, + "grad_norm": 0.3951696455478668, + "learning_rate": 0.00047751782679397573, + "loss": 4.7383, + "step": 1598 + }, + { + "epoch": 0.15615234375, + "grad_norm": 0.49301767349243164, + "learning_rate": 0.0004774872920278022, + "loss": 4.7148, + "step": 1599 + }, + { + "epoch": 0.15625, + "grad_norm": 0.5991638898849487, + "learning_rate": 0.0004774567376318599, + "loss": 4.7266, + "step": 1600 + }, + { + "epoch": 0.15634765625, + "grad_norm": 0.5442166328430176, + "learning_rate": 0.00047742616360911105, + "loss": 4.7344, + "step": 1601 + }, + { + "epoch": 0.1564453125, + "grad_norm": 0.474050909280777, + "learning_rate": 0.0004773955699625196, + "loss": 4.6992, + "step": 1602 + }, + { + "epoch": 0.15654296875, + "grad_norm": 0.38881492614746094, + "learning_rate": 0.0004773649566950512, + "loss": 4.7344, + "step": 1603 + }, + { + "epoch": 0.156640625, + "grad_norm": 0.3230612576007843, + "learning_rate": 0.0004773343238096737, + "loss": 4.7383, + "step": 1604 + }, + { + "epoch": 0.15673828125, + "grad_norm": 0.3541710078716278, + "learning_rate": 0.0004773036713093567, + "loss": 4.7305, + "step": 1605 + }, + { + "epoch": 0.1568359375, + "grad_norm": 0.3545769453048706, + "learning_rate": 0.0004772729991970719, + "loss": 4.7148, + "step": 1606 + }, + { + "epoch": 0.15693359375, + "grad_norm": 0.376462459564209, + "learning_rate": 0.0004772423074757926, + "loss": 4.7383, + "step": 1607 + }, + { + "epoch": 0.15703125, + "grad_norm": 0.40618640184402466, + "learning_rate": 0.00047721159614849424, + "loss": 4.7031, + "step": 1608 + }, + { + "epoch": 0.15712890625, + "grad_norm": 0.5097858905792236, + "learning_rate": 0.000477180865218154, + "loss": 4.7695, + "step": 1609 + }, + { + "epoch": 0.1572265625, + "grad_norm": 0.6011024117469788, + "learning_rate": 0.00047715011468775124, + "loss": 4.7188, + "step": 1610 + }, + { + "epoch": 0.15732421875, + "grad_norm": 0.5320466756820679, + "learning_rate": 0.00047711934456026674, + "loss": 4.7305, + "step": 1611 + }, + { + "epoch": 0.157421875, + "grad_norm": 0.3568183481693268, + "learning_rate": 0.0004770885548386835, + "loss": 4.7422, + "step": 1612 + }, + { + "epoch": 0.15751953125, + "grad_norm": 0.35290566086769104, + "learning_rate": 0.0004770577455259865, + "loss": 4.7227, + "step": 1613 + }, + { + "epoch": 0.1576171875, + "grad_norm": 0.3121565878391266, + "learning_rate": 0.0004770269166251625, + "loss": 4.7461, + "step": 1614 + }, + { + "epoch": 0.15771484375, + "grad_norm": 0.3387918174266815, + "learning_rate": 0.00047699606813920005, + "loss": 4.7031, + "step": 1615 + }, + { + "epoch": 0.1578125, + "grad_norm": 0.3331822156906128, + "learning_rate": 0.0004769652000710898, + "loss": 4.7578, + "step": 1616 + }, + { + "epoch": 0.15791015625, + "grad_norm": 0.38116782903671265, + "learning_rate": 0.00047693431242382405, + "loss": 4.7539, + "step": 1617 + }, + { + "epoch": 0.1580078125, + "grad_norm": 0.38010165095329285, + "learning_rate": 0.0004769034052003973, + "loss": 4.7305, + "step": 1618 + }, + { + "epoch": 0.15810546875, + "grad_norm": 0.3130066692829132, + "learning_rate": 0.00047687247840380576, + "loss": 4.7422, + "step": 1619 + }, + { + "epoch": 0.158203125, + "grad_norm": 0.38829630613327026, + "learning_rate": 0.00047684153203704754, + "loss": 4.7734, + "step": 1620 + }, + { + "epoch": 0.15830078125, + "grad_norm": 0.5447909832000732, + "learning_rate": 0.00047681056610312264, + "loss": 4.7383, + "step": 1621 + }, + { + "epoch": 0.1583984375, + "grad_norm": 0.6322894096374512, + "learning_rate": 0.0004767795806050331, + "loss": 4.7305, + "step": 1622 + }, + { + "epoch": 0.15849609375, + "grad_norm": 0.6145194172859192, + "learning_rate": 0.00047674857554578273, + "loss": 4.7148, + "step": 1623 + }, + { + "epoch": 0.15859375, + "grad_norm": 0.3967234492301941, + "learning_rate": 0.0004767175509283772, + "loss": 4.7578, + "step": 1624 + }, + { + "epoch": 0.15869140625, + "grad_norm": 0.40282419323921204, + "learning_rate": 0.00047668650675582413, + "loss": 4.7227, + "step": 1625 + }, + { + "epoch": 0.1587890625, + "grad_norm": 0.37083151936531067, + "learning_rate": 0.00047665544303113314, + "loss": 4.7539, + "step": 1626 + }, + { + "epoch": 0.15888671875, + "grad_norm": 0.376194566488266, + "learning_rate": 0.0004766243597573155, + "loss": 4.7305, + "step": 1627 + }, + { + "epoch": 0.158984375, + "grad_norm": 0.3005055785179138, + "learning_rate": 0.0004765932569373847, + "loss": 4.7305, + "step": 1628 + }, + { + "epoch": 0.15908203125, + "grad_norm": 0.30844777822494507, + "learning_rate": 0.0004765621345743558, + "loss": 4.7188, + "step": 1629 + }, + { + "epoch": 0.1591796875, + "grad_norm": 0.30865785479545593, + "learning_rate": 0.000476530992671246, + "loss": 4.7383, + "step": 1630 + }, + { + "epoch": 0.15927734375, + "grad_norm": 0.34530970454216003, + "learning_rate": 0.0004764998312310742, + "loss": 4.7422, + "step": 1631 + }, + { + "epoch": 0.159375, + "grad_norm": 0.3432547152042389, + "learning_rate": 0.00047646865025686134, + "loss": 4.7383, + "step": 1632 + }, + { + "epoch": 0.15947265625, + "grad_norm": 0.32131803035736084, + "learning_rate": 0.0004764374497516303, + "loss": 4.7109, + "step": 1633 + }, + { + "epoch": 0.1595703125, + "grad_norm": 0.31007057428359985, + "learning_rate": 0.0004764062297184056, + "loss": 4.7383, + "step": 1634 + }, + { + "epoch": 0.15966796875, + "grad_norm": 0.33511513471603394, + "learning_rate": 0.0004763749901602139, + "loss": 4.7266, + "step": 1635 + }, + { + "epoch": 0.159765625, + "grad_norm": 0.41421255469322205, + "learning_rate": 0.0004763437310800837, + "loss": 4.7188, + "step": 1636 + }, + { + "epoch": 0.15986328125, + "grad_norm": 0.3609168231487274, + "learning_rate": 0.0004763124524810452, + "loss": 4.75, + "step": 1637 + }, + { + "epoch": 0.1599609375, + "grad_norm": 0.32799166440963745, + "learning_rate": 0.0004762811543661308, + "loss": 4.6914, + "step": 1638 + }, + { + "epoch": 0.16005859375, + "grad_norm": 0.33859479427337646, + "learning_rate": 0.0004762498367383746, + "loss": 4.7188, + "step": 1639 + }, + { + "epoch": 0.16015625, + "grad_norm": 0.32123857736587524, + "learning_rate": 0.0004762184996008125, + "loss": 4.7266, + "step": 1640 + }, + { + "epoch": 0.16025390625, + "grad_norm": 0.33974528312683105, + "learning_rate": 0.00047618714295648263, + "loss": 4.7617, + "step": 1641 + }, + { + "epoch": 0.1603515625, + "grad_norm": 0.34150657057762146, + "learning_rate": 0.0004761557668084247, + "loss": 4.7148, + "step": 1642 + }, + { + "epoch": 0.16044921875, + "grad_norm": 0.30895042419433594, + "learning_rate": 0.00047612437115968043, + "loss": 4.7109, + "step": 1643 + }, + { + "epoch": 0.160546875, + "grad_norm": 0.30955177545547485, + "learning_rate": 0.0004760929560132934, + "loss": 4.7031, + "step": 1644 + }, + { + "epoch": 0.16064453125, + "grad_norm": 0.2998163104057312, + "learning_rate": 0.00047606152137230905, + "loss": 4.7227, + "step": 1645 + }, + { + "epoch": 0.1607421875, + "grad_norm": 0.3299749195575714, + "learning_rate": 0.0004760300672397749, + "loss": 4.7188, + "step": 1646 + }, + { + "epoch": 0.16083984375, + "grad_norm": 0.3188712000846863, + "learning_rate": 0.0004759985936187401, + "loss": 4.7188, + "step": 1647 + }, + { + "epoch": 0.1609375, + "grad_norm": 0.2848448157310486, + "learning_rate": 0.00047596710051225585, + "loss": 4.7422, + "step": 1648 + }, + { + "epoch": 0.16103515625, + "grad_norm": 0.32674816250801086, + "learning_rate": 0.00047593558792337516, + "loss": 4.7344, + "step": 1649 + }, + { + "epoch": 0.1611328125, + "grad_norm": 0.30364173650741577, + "learning_rate": 0.00047590405585515294, + "loss": 4.7266, + "step": 1650 + }, + { + "epoch": 0.16123046875, + "grad_norm": 0.31012097001075745, + "learning_rate": 0.0004758725043106461, + "loss": 4.75, + "step": 1651 + }, + { + "epoch": 0.161328125, + "grad_norm": 0.3971497416496277, + "learning_rate": 0.0004758409332929133, + "loss": 4.7148, + "step": 1652 + }, + { + "epoch": 0.16142578125, + "grad_norm": 0.47804024815559387, + "learning_rate": 0.0004758093428050151, + "loss": 4.7344, + "step": 1653 + }, + { + "epoch": 0.1615234375, + "grad_norm": 0.6461985111236572, + "learning_rate": 0.00047577773285001395, + "loss": 4.7109, + "step": 1654 + }, + { + "epoch": 0.16162109375, + "grad_norm": 0.692113995552063, + "learning_rate": 0.0004757461034309744, + "loss": 4.7188, + "step": 1655 + }, + { + "epoch": 0.16171875, + "grad_norm": 0.5253254771232605, + "learning_rate": 0.0004757144545509625, + "loss": 4.7188, + "step": 1656 + }, + { + "epoch": 0.16181640625, + "grad_norm": 0.3248235285282135, + "learning_rate": 0.0004756827862130465, + "loss": 4.6953, + "step": 1657 + }, + { + "epoch": 0.1619140625, + "grad_norm": 0.40135297179222107, + "learning_rate": 0.0004756510984202964, + "loss": 4.7148, + "step": 1658 + }, + { + "epoch": 0.16201171875, + "grad_norm": 0.4172717332839966, + "learning_rate": 0.0004756193911757842, + "loss": 4.7305, + "step": 1659 + }, + { + "epoch": 0.162109375, + "grad_norm": 0.3560163974761963, + "learning_rate": 0.00047558766448258357, + "loss": 4.7695, + "step": 1660 + }, + { + "epoch": 0.16220703125, + "grad_norm": 0.28156837821006775, + "learning_rate": 0.0004755559183437703, + "loss": 4.75, + "step": 1661 + }, + { + "epoch": 0.1623046875, + "grad_norm": 0.3130258321762085, + "learning_rate": 0.00047552415276242186, + "loss": 4.7109, + "step": 1662 + }, + { + "epoch": 0.16240234375, + "grad_norm": 0.3561786115169525, + "learning_rate": 0.0004754923677416178, + "loss": 4.7422, + "step": 1663 + }, + { + "epoch": 0.1625, + "grad_norm": 0.3439064621925354, + "learning_rate": 0.00047546056328443947, + "loss": 4.7383, + "step": 1664 + }, + { + "epoch": 0.16259765625, + "grad_norm": 0.3227904438972473, + "learning_rate": 0.00047542873939397006, + "loss": 4.707, + "step": 1665 + }, + { + "epoch": 0.1626953125, + "grad_norm": 0.29672226309776306, + "learning_rate": 0.00047539689607329467, + "loss": 4.707, + "step": 1666 + }, + { + "epoch": 0.16279296875, + "grad_norm": 0.3555067479610443, + "learning_rate": 0.00047536503332550024, + "loss": 4.6875, + "step": 1667 + }, + { + "epoch": 0.162890625, + "grad_norm": 0.3557547330856323, + "learning_rate": 0.00047533315115367577, + "loss": 4.7305, + "step": 1668 + }, + { + "epoch": 0.16298828125, + "grad_norm": 0.28665271401405334, + "learning_rate": 0.000475301249560912, + "loss": 4.7266, + "step": 1669 + }, + { + "epoch": 0.1630859375, + "grad_norm": 0.2968331575393677, + "learning_rate": 0.0004752693285503014, + "loss": 4.7305, + "step": 1670 + }, + { + "epoch": 0.16318359375, + "grad_norm": 0.33975455164909363, + "learning_rate": 0.0004752373881249387, + "loss": 4.7344, + "step": 1671 + }, + { + "epoch": 0.16328125, + "grad_norm": 0.36191651225090027, + "learning_rate": 0.0004752054282879203, + "loss": 4.7148, + "step": 1672 + }, + { + "epoch": 0.16337890625, + "grad_norm": 0.3275497853755951, + "learning_rate": 0.0004751734490423444, + "loss": 4.7227, + "step": 1673 + }, + { + "epoch": 0.1634765625, + "grad_norm": 0.3115062713623047, + "learning_rate": 0.00047514145039131116, + "loss": 4.6953, + "step": 1674 + }, + { + "epoch": 0.16357421875, + "grad_norm": 0.33993664383888245, + "learning_rate": 0.0004751094323379227, + "loss": 4.7188, + "step": 1675 + }, + { + "epoch": 0.163671875, + "grad_norm": 0.33467310667037964, + "learning_rate": 0.0004750773948852829, + "loss": 4.7148, + "step": 1676 + }, + { + "epoch": 0.16376953125, + "grad_norm": 0.3193880021572113, + "learning_rate": 0.00047504533803649756, + "loss": 4.7227, + "step": 1677 + }, + { + "epoch": 0.1638671875, + "grad_norm": 0.30603379011154175, + "learning_rate": 0.0004750132617946745, + "loss": 4.7578, + "step": 1678 + }, + { + "epoch": 0.16396484375, + "grad_norm": 0.2963157594203949, + "learning_rate": 0.0004749811661629231, + "loss": 4.6953, + "step": 1679 + }, + { + "epoch": 0.1640625, + "grad_norm": 0.3166114091873169, + "learning_rate": 0.000474949051144355, + "loss": 4.6914, + "step": 1680 + }, + { + "epoch": 0.16416015625, + "grad_norm": 0.29735228419303894, + "learning_rate": 0.00047491691674208343, + "loss": 4.75, + "step": 1681 + }, + { + "epoch": 0.1642578125, + "grad_norm": 0.3676241636276245, + "learning_rate": 0.0004748847629592236, + "loss": 4.6992, + "step": 1682 + }, + { + "epoch": 0.16435546875, + "grad_norm": 0.4388222396373749, + "learning_rate": 0.00047485258979889267, + "loss": 4.7422, + "step": 1683 + }, + { + "epoch": 0.164453125, + "grad_norm": 0.5815744400024414, + "learning_rate": 0.0004748203972642096, + "loss": 4.6875, + "step": 1684 + }, + { + "epoch": 0.16455078125, + "grad_norm": 0.6494610905647278, + "learning_rate": 0.00047478818535829503, + "loss": 4.7383, + "step": 1685 + }, + { + "epoch": 0.1646484375, + "grad_norm": 0.5286709070205688, + "learning_rate": 0.00047475595408427197, + "loss": 4.707, + "step": 1686 + }, + { + "epoch": 0.16474609375, + "grad_norm": 0.3297274708747864, + "learning_rate": 0.00047472370344526496, + "loss": 4.7188, + "step": 1687 + }, + { + "epoch": 0.16484375, + "grad_norm": 0.3907630741596222, + "learning_rate": 0.0004746914334444003, + "loss": 4.7305, + "step": 1688 + }, + { + "epoch": 0.16494140625, + "grad_norm": 0.4345732033252716, + "learning_rate": 0.00047465914408480653, + "loss": 4.7188, + "step": 1689 + }, + { + "epoch": 0.1650390625, + "grad_norm": 0.3433893024921417, + "learning_rate": 0.00047462683536961383, + "loss": 4.7461, + "step": 1690 + }, + { + "epoch": 0.16513671875, + "grad_norm": 0.2961682379245758, + "learning_rate": 0.0004745945073019543, + "loss": 4.7305, + "step": 1691 + }, + { + "epoch": 0.165234375, + "grad_norm": 0.3548338711261749, + "learning_rate": 0.000474562159884962, + "loss": 4.7188, + "step": 1692 + }, + { + "epoch": 0.16533203125, + "grad_norm": 0.3313615322113037, + "learning_rate": 0.0004745297931217727, + "loss": 4.6914, + "step": 1693 + }, + { + "epoch": 0.1654296875, + "grad_norm": 0.29579681158065796, + "learning_rate": 0.0004744974070155242, + "loss": 4.7305, + "step": 1694 + }, + { + "epoch": 0.16552734375, + "grad_norm": 0.3224891722202301, + "learning_rate": 0.000474465001569356, + "loss": 4.7422, + "step": 1695 + }, + { + "epoch": 0.165625, + "grad_norm": 0.37823259830474854, + "learning_rate": 0.0004744325767864097, + "loss": 4.6875, + "step": 1696 + }, + { + "epoch": 0.16572265625, + "grad_norm": 0.4113744795322418, + "learning_rate": 0.00047440013266982867, + "loss": 4.7109, + "step": 1697 + }, + { + "epoch": 0.1658203125, + "grad_norm": 0.39243578910827637, + "learning_rate": 0.00047436766922275805, + "loss": 4.7461, + "step": 1698 + }, + { + "epoch": 0.16591796875, + "grad_norm": 0.34951266646385193, + "learning_rate": 0.0004743351864483451, + "loss": 4.7227, + "step": 1699 + }, + { + "epoch": 0.166015625, + "grad_norm": 0.29626819491386414, + "learning_rate": 0.00047430268434973873, + "loss": 4.6953, + "step": 1700 + }, + { + "epoch": 0.16611328125, + "grad_norm": 0.29469388723373413, + "learning_rate": 0.0004742701629300896, + "loss": 4.7305, + "step": 1701 + }, + { + "epoch": 0.1662109375, + "grad_norm": 0.2781272232532501, + "learning_rate": 0.0004742376221925508, + "loss": 4.75, + "step": 1702 + }, + { + "epoch": 0.16630859375, + "grad_norm": 0.32316306233406067, + "learning_rate": 0.0004742050621402767, + "loss": 4.6992, + "step": 1703 + }, + { + "epoch": 0.16640625, + "grad_norm": 0.36234429478645325, + "learning_rate": 0.00047417248277642385, + "loss": 4.7031, + "step": 1704 + }, + { + "epoch": 0.16650390625, + "grad_norm": 0.4072574973106384, + "learning_rate": 0.0004741398841041506, + "loss": 4.7266, + "step": 1705 + }, + { + "epoch": 0.1666015625, + "grad_norm": 0.34564459323883057, + "learning_rate": 0.0004741072661266171, + "loss": 4.6953, + "step": 1706 + }, + { + "epoch": 0.16669921875, + "grad_norm": 0.30190330743789673, + "learning_rate": 0.0004740746288469855, + "loss": 4.7188, + "step": 1707 + }, + { + "epoch": 0.166796875, + "grad_norm": 0.33523130416870117, + "learning_rate": 0.00047404197226841977, + "loss": 4.7344, + "step": 1708 + }, + { + "epoch": 0.16689453125, + "grad_norm": 0.3154967725276947, + "learning_rate": 0.0004740092963940858, + "loss": 4.7188, + "step": 1709 + }, + { + "epoch": 0.1669921875, + "grad_norm": 0.29154172539711, + "learning_rate": 0.00047397660122715116, + "loss": 4.6914, + "step": 1710 + }, + { + "epoch": 0.16708984375, + "grad_norm": 0.29590922594070435, + "learning_rate": 0.0004739438867707855, + "loss": 4.7266, + "step": 1711 + }, + { + "epoch": 0.1671875, + "grad_norm": 0.31150999665260315, + "learning_rate": 0.00047391115302816017, + "loss": 4.7461, + "step": 1712 + }, + { + "epoch": 0.16728515625, + "grad_norm": 0.3638792037963867, + "learning_rate": 0.00047387840000244865, + "loss": 4.7227, + "step": 1713 + }, + { + "epoch": 0.1673828125, + "grad_norm": 0.41439005732536316, + "learning_rate": 0.000473845627696826, + "loss": 4.7109, + "step": 1714 + }, + { + "epoch": 0.16748046875, + "grad_norm": 0.4629388451576233, + "learning_rate": 0.0004738128361144693, + "loss": 4.6797, + "step": 1715 + }, + { + "epoch": 0.167578125, + "grad_norm": 0.48057830333709717, + "learning_rate": 0.00047378002525855765, + "loss": 4.6953, + "step": 1716 + }, + { + "epoch": 0.16767578125, + "grad_norm": 0.4065990746021271, + "learning_rate": 0.00047374719513227156, + "loss": 4.7031, + "step": 1717 + }, + { + "epoch": 0.1677734375, + "grad_norm": 0.2860855758190155, + "learning_rate": 0.00047371434573879374, + "loss": 4.7031, + "step": 1718 + }, + { + "epoch": 0.16787109375, + "grad_norm": 0.28443628549575806, + "learning_rate": 0.0004736814770813089, + "loss": 4.707, + "step": 1719 + }, + { + "epoch": 0.16796875, + "grad_norm": 0.3028384745121002, + "learning_rate": 0.00047364858916300323, + "loss": 4.7188, + "step": 1720 + }, + { + "epoch": 0.16806640625, + "grad_norm": 0.33343642950057983, + "learning_rate": 0.00047361568198706506, + "loss": 4.7109, + "step": 1721 + }, + { + "epoch": 0.1681640625, + "grad_norm": 0.30160287022590637, + "learning_rate": 0.0004735827555566846, + "loss": 4.7422, + "step": 1722 + }, + { + "epoch": 0.16826171875, + "grad_norm": 0.31500527262687683, + "learning_rate": 0.00047354980987505377, + "loss": 4.7344, + "step": 1723 + }, + { + "epoch": 0.168359375, + "grad_norm": 0.29347512125968933, + "learning_rate": 0.00047351684494536633, + "loss": 4.6953, + "step": 1724 + }, + { + "epoch": 0.16845703125, + "grad_norm": 0.29177677631378174, + "learning_rate": 0.00047348386077081825, + "loss": 4.7148, + "step": 1725 + }, + { + "epoch": 0.1685546875, + "grad_norm": 0.27817559242248535, + "learning_rate": 0.0004734508573546069, + "loss": 4.7188, + "step": 1726 + }, + { + "epoch": 0.16865234375, + "grad_norm": 0.2869120240211487, + "learning_rate": 0.0004734178346999318, + "loss": 4.7148, + "step": 1727 + }, + { + "epoch": 0.16875, + "grad_norm": 0.27404704689979553, + "learning_rate": 0.00047338479280999434, + "loss": 4.6992, + "step": 1728 + }, + { + "epoch": 0.16884765625, + "grad_norm": 0.30236533284187317, + "learning_rate": 0.0004733517316879977, + "loss": 4.707, + "step": 1729 + }, + { + "epoch": 0.1689453125, + "grad_norm": 0.3438217043876648, + "learning_rate": 0.0004733186513371468, + "loss": 4.7109, + "step": 1730 + }, + { + "epoch": 0.16904296875, + "grad_norm": 0.42064616084098816, + "learning_rate": 0.00047328555176064866, + "loss": 4.7031, + "step": 1731 + }, + { + "epoch": 0.169140625, + "grad_norm": 0.5444338321685791, + "learning_rate": 0.00047325243296171207, + "loss": 4.7227, + "step": 1732 + }, + { + "epoch": 0.16923828125, + "grad_norm": 0.6516023874282837, + "learning_rate": 0.00047321929494354765, + "loss": 4.6953, + "step": 1733 + }, + { + "epoch": 0.1693359375, + "grad_norm": 0.5612179040908813, + "learning_rate": 0.0004731861377093679, + "loss": 4.7266, + "step": 1734 + }, + { + "epoch": 0.16943359375, + "grad_norm": 0.36678099632263184, + "learning_rate": 0.00047315296126238714, + "loss": 4.7188, + "step": 1735 + }, + { + "epoch": 0.16953125, + "grad_norm": 0.35155007243156433, + "learning_rate": 0.0004731197656058218, + "loss": 4.6953, + "step": 1736 + }, + { + "epoch": 0.16962890625, + "grad_norm": 0.4132586419582367, + "learning_rate": 0.0004730865507428897, + "loss": 4.7148, + "step": 1737 + }, + { + "epoch": 0.1697265625, + "grad_norm": 0.3739107847213745, + "learning_rate": 0.00047305331667681094, + "loss": 4.707, + "step": 1738 + }, + { + "epoch": 0.16982421875, + "grad_norm": 0.27712637186050415, + "learning_rate": 0.0004730200634108073, + "loss": 4.707, + "step": 1739 + }, + { + "epoch": 0.169921875, + "grad_norm": 0.3548940122127533, + "learning_rate": 0.00047298679094810253, + "loss": 4.7695, + "step": 1740 + }, + { + "epoch": 0.17001953125, + "grad_norm": 0.39106234908103943, + "learning_rate": 0.0004729534992919221, + "loss": 4.7383, + "step": 1741 + }, + { + "epoch": 0.1701171875, + "grad_norm": 0.31391414999961853, + "learning_rate": 0.00047292018844549347, + "loss": 4.7344, + "step": 1742 + }, + { + "epoch": 0.17021484375, + "grad_norm": 0.31894975900650024, + "learning_rate": 0.00047288685841204576, + "loss": 4.7109, + "step": 1743 + }, + { + "epoch": 0.1703125, + "grad_norm": 0.3426681458950043, + "learning_rate": 0.0004728535091948103, + "loss": 4.7266, + "step": 1744 + }, + { + "epoch": 0.17041015625, + "grad_norm": 0.3944070339202881, + "learning_rate": 0.0004728201407970199, + "loss": 4.7383, + "step": 1745 + }, + { + "epoch": 0.1705078125, + "grad_norm": 0.33504220843315125, + "learning_rate": 0.00047278675322190943, + "loss": 4.7031, + "step": 1746 + }, + { + "epoch": 0.17060546875, + "grad_norm": 0.29810965061187744, + "learning_rate": 0.00047275334647271566, + "loss": 4.7188, + "step": 1747 + }, + { + "epoch": 0.170703125, + "grad_norm": 0.40818923711776733, + "learning_rate": 0.00047271992055267713, + "loss": 4.7344, + "step": 1748 + }, + { + "epoch": 0.17080078125, + "grad_norm": 0.4044819474220276, + "learning_rate": 0.0004726864754650342, + "loss": 4.7227, + "step": 1749 + }, + { + "epoch": 0.1708984375, + "grad_norm": 0.3737597167491913, + "learning_rate": 0.00047265301121302917, + "loss": 4.7031, + "step": 1750 + }, + { + "epoch": 0.17099609375, + "grad_norm": 0.346331387758255, + "learning_rate": 0.0004726195277999062, + "loss": 4.707, + "step": 1751 + }, + { + "epoch": 0.17109375, + "grad_norm": 0.3120872378349304, + "learning_rate": 0.00047258602522891124, + "loss": 4.7109, + "step": 1752 + }, + { + "epoch": 0.17119140625, + "grad_norm": 0.3185526132583618, + "learning_rate": 0.0004725525035032921, + "loss": 4.7109, + "step": 1753 + }, + { + "epoch": 0.1712890625, + "grad_norm": 0.35277417302131653, + "learning_rate": 0.0004725189626262986, + "loss": 4.7148, + "step": 1754 + }, + { + "epoch": 0.17138671875, + "grad_norm": 0.3095724880695343, + "learning_rate": 0.00047248540260118224, + "loss": 4.7109, + "step": 1755 + }, + { + "epoch": 0.171484375, + "grad_norm": 0.30601778626441956, + "learning_rate": 0.0004724518234311964, + "loss": 4.6914, + "step": 1756 + }, + { + "epoch": 0.17158203125, + "grad_norm": 0.2815275490283966, + "learning_rate": 0.0004724182251195964, + "loss": 4.7148, + "step": 1757 + }, + { + "epoch": 0.1716796875, + "grad_norm": 0.2875475585460663, + "learning_rate": 0.0004723846076696395, + "loss": 4.7109, + "step": 1758 + }, + { + "epoch": 0.17177734375, + "grad_norm": 0.3428354263305664, + "learning_rate": 0.00047235097108458434, + "loss": 4.7188, + "step": 1759 + }, + { + "epoch": 0.171875, + "grad_norm": 0.4065704643726349, + "learning_rate": 0.00047231731536769207, + "loss": 4.6875, + "step": 1760 + }, + { + "epoch": 0.17197265625, + "grad_norm": 0.47537028789520264, + "learning_rate": 0.00047228364052222515, + "loss": 4.7383, + "step": 1761 + }, + { + "epoch": 0.1720703125, + "grad_norm": 0.46676430106163025, + "learning_rate": 0.0004722499465514484, + "loss": 4.7266, + "step": 1762 + }, + { + "epoch": 0.17216796875, + "grad_norm": 0.35790807008743286, + "learning_rate": 0.00047221623345862794, + "loss": 4.6992, + "step": 1763 + }, + { + "epoch": 0.172265625, + "grad_norm": 0.30717775225639343, + "learning_rate": 0.00047218250124703224, + "loss": 4.7031, + "step": 1764 + }, + { + "epoch": 0.17236328125, + "grad_norm": 0.36534759402275085, + "learning_rate": 0.00047214874991993136, + "loss": 4.7109, + "step": 1765 + }, + { + "epoch": 0.1724609375, + "grad_norm": 0.39176201820373535, + "learning_rate": 0.0004721149794805971, + "loss": 4.707, + "step": 1766 + }, + { + "epoch": 0.17255859375, + "grad_norm": 0.3514379560947418, + "learning_rate": 0.0004720811899323035, + "loss": 4.7305, + "step": 1767 + }, + { + "epoch": 0.17265625, + "grad_norm": 0.3105420768260956, + "learning_rate": 0.00047204738127832615, + "loss": 4.7344, + "step": 1768 + }, + { + "epoch": 0.17275390625, + "grad_norm": 0.27019309997558594, + "learning_rate": 0.0004720135535219425, + "loss": 4.6953, + "step": 1769 + }, + { + "epoch": 0.1728515625, + "grad_norm": 0.29011520743370056, + "learning_rate": 0.000471979706666432, + "loss": 4.7148, + "step": 1770 + }, + { + "epoch": 0.17294921875, + "grad_norm": 0.3539043068885803, + "learning_rate": 0.00047194584071507577, + "loss": 4.7266, + "step": 1771 + }, + { + "epoch": 0.173046875, + "grad_norm": 0.33324524760246277, + "learning_rate": 0.000471911955671157, + "loss": 4.6914, + "step": 1772 + }, + { + "epoch": 0.17314453125, + "grad_norm": 0.3441627323627472, + "learning_rate": 0.0004718780515379606, + "loss": 4.707, + "step": 1773 + }, + { + "epoch": 0.1732421875, + "grad_norm": 0.3166051506996155, + "learning_rate": 0.0004718441283187733, + "loss": 4.7344, + "step": 1774 + }, + { + "epoch": 0.17333984375, + "grad_norm": 0.32287782430648804, + "learning_rate": 0.0004718101860168837, + "loss": 4.6953, + "step": 1775 + }, + { + "epoch": 0.1734375, + "grad_norm": 0.3744673430919647, + "learning_rate": 0.0004717762246355824, + "loss": 4.7227, + "step": 1776 + }, + { + "epoch": 0.17353515625, + "grad_norm": 0.4147856831550598, + "learning_rate": 0.0004717422441781616, + "loss": 4.707, + "step": 1777 + }, + { + "epoch": 0.1736328125, + "grad_norm": 0.44820207357406616, + "learning_rate": 0.0004717082446479156, + "loss": 4.6992, + "step": 1778 + }, + { + "epoch": 0.17373046875, + "grad_norm": 0.42957448959350586, + "learning_rate": 0.0004716742260481402, + "loss": 4.6992, + "step": 1779 + }, + { + "epoch": 0.173828125, + "grad_norm": 0.35971277952194214, + "learning_rate": 0.00047164018838213346, + "loss": 4.707, + "step": 1780 + }, + { + "epoch": 0.17392578125, + "grad_norm": 0.29600271582603455, + "learning_rate": 0.0004716061316531951, + "loss": 4.6641, + "step": 1781 + }, + { + "epoch": 0.1740234375, + "grad_norm": 0.3103143870830536, + "learning_rate": 0.00047157205586462664, + "loss": 4.7148, + "step": 1782 + }, + { + "epoch": 0.17412109375, + "grad_norm": 0.2851412892341614, + "learning_rate": 0.0004715379610197315, + "loss": 4.7422, + "step": 1783 + }, + { + "epoch": 0.17421875, + "grad_norm": 0.2680378258228302, + "learning_rate": 0.000471503847121815, + "loss": 4.7188, + "step": 1784 + }, + { + "epoch": 0.17431640625, + "grad_norm": 0.2789458632469177, + "learning_rate": 0.00047146971417418417, + "loss": 4.6836, + "step": 1785 + }, + { + "epoch": 0.1744140625, + "grad_norm": 0.3282760679721832, + "learning_rate": 0.0004714355621801479, + "loss": 4.7305, + "step": 1786 + }, + { + "epoch": 0.17451171875, + "grad_norm": 0.35235342383384705, + "learning_rate": 0.00047140139114301715, + "loss": 4.707, + "step": 1787 + }, + { + "epoch": 0.174609375, + "grad_norm": 0.3573938310146332, + "learning_rate": 0.0004713672010661045, + "loss": 4.7148, + "step": 1788 + }, + { + "epoch": 0.17470703125, + "grad_norm": 0.3707910478115082, + "learning_rate": 0.0004713329919527245, + "loss": 4.7227, + "step": 1789 + }, + { + "epoch": 0.1748046875, + "grad_norm": 0.3104837238788605, + "learning_rate": 0.0004712987638061934, + "loss": 4.6914, + "step": 1790 + }, + { + "epoch": 0.17490234375, + "grad_norm": 0.3016248047351837, + "learning_rate": 0.00047126451662982946, + "loss": 4.707, + "step": 1791 + }, + { + "epoch": 0.175, + "grad_norm": 0.299649715423584, + "learning_rate": 0.00047123025042695267, + "loss": 4.6953, + "step": 1792 + }, + { + "epoch": 0.17509765625, + "grad_norm": 0.3024265170097351, + "learning_rate": 0.0004711959652008849, + "loss": 4.707, + "step": 1793 + }, + { + "epoch": 0.1751953125, + "grad_norm": 0.3478015065193176, + "learning_rate": 0.0004711616609549499, + "loss": 4.7109, + "step": 1794 + }, + { + "epoch": 0.17529296875, + "grad_norm": 0.38603729009628296, + "learning_rate": 0.0004711273376924732, + "loss": 4.7188, + "step": 1795 + }, + { + "epoch": 0.175390625, + "grad_norm": 0.48535963892936707, + "learning_rate": 0.0004710929954167823, + "loss": 4.6914, + "step": 1796 + }, + { + "epoch": 0.17548828125, + "grad_norm": 0.5580281019210815, + "learning_rate": 0.0004710586341312064, + "loss": 4.7148, + "step": 1797 + }, + { + "epoch": 0.1755859375, + "grad_norm": 0.44946280121803284, + "learning_rate": 0.0004710242538390765, + "loss": 4.75, + "step": 1798 + }, + { + "epoch": 0.17568359375, + "grad_norm": 0.3055347502231598, + "learning_rate": 0.0004709898545437256, + "loss": 4.707, + "step": 1799 + }, + { + "epoch": 0.17578125, + "grad_norm": 0.39318904280662537, + "learning_rate": 0.00047095543624848857, + "loss": 4.6836, + "step": 1800 + }, + { + "epoch": 0.17587890625, + "grad_norm": 0.4857766330242157, + "learning_rate": 0.0004709209989567019, + "loss": 4.6992, + "step": 1801 + }, + { + "epoch": 0.1759765625, + "grad_norm": 0.3867643177509308, + "learning_rate": 0.00047088654267170413, + "loss": 4.7148, + "step": 1802 + }, + { + "epoch": 0.17607421875, + "grad_norm": 0.28597569465637207, + "learning_rate": 0.00047085206739683547, + "loss": 4.7109, + "step": 1803 + }, + { + "epoch": 0.176171875, + "grad_norm": 0.322764128446579, + "learning_rate": 0.00047081757313543815, + "loss": 4.6914, + "step": 1804 + }, + { + "epoch": 0.17626953125, + "grad_norm": 0.33930692076683044, + "learning_rate": 0.0004707830598908561, + "loss": 4.7109, + "step": 1805 + }, + { + "epoch": 0.1763671875, + "grad_norm": 0.31630805134773254, + "learning_rate": 0.0004707485276664353, + "loss": 4.7031, + "step": 1806 + }, + { + "epoch": 0.17646484375, + "grad_norm": 0.3265106976032257, + "learning_rate": 0.0004707139764655232, + "loss": 4.7148, + "step": 1807 + }, + { + "epoch": 0.1765625, + "grad_norm": 0.320544570684433, + "learning_rate": 0.0004706794062914694, + "loss": 4.7266, + "step": 1808 + }, + { + "epoch": 0.17666015625, + "grad_norm": 0.35693979263305664, + "learning_rate": 0.00047064481714762527, + "loss": 4.7109, + "step": 1809 + }, + { + "epoch": 0.1767578125, + "grad_norm": 0.3931410014629364, + "learning_rate": 0.00047061020903734396, + "loss": 4.7227, + "step": 1810 + }, + { + "epoch": 0.17685546875, + "grad_norm": 0.4092164635658264, + "learning_rate": 0.0004705755819639804, + "loss": 4.6797, + "step": 1811 + }, + { + "epoch": 0.176953125, + "grad_norm": 0.3898935317993164, + "learning_rate": 0.00047054093593089163, + "loss": 4.7109, + "step": 1812 + }, + { + "epoch": 0.17705078125, + "grad_norm": 0.30018892884254456, + "learning_rate": 0.0004705062709414363, + "loss": 4.7344, + "step": 1813 + }, + { + "epoch": 0.1771484375, + "grad_norm": 0.3135061264038086, + "learning_rate": 0.00047047158699897485, + "loss": 4.7461, + "step": 1814 + }, + { + "epoch": 0.17724609375, + "grad_norm": 0.33332744240760803, + "learning_rate": 0.00047043688410686977, + "loss": 4.6875, + "step": 1815 + }, + { + "epoch": 0.17734375, + "grad_norm": 0.3224795460700989, + "learning_rate": 0.0004704021622684851, + "loss": 4.7148, + "step": 1816 + }, + { + "epoch": 0.17744140625, + "grad_norm": 0.3023662865161896, + "learning_rate": 0.0004703674214871871, + "loss": 4.7148, + "step": 1817 + }, + { + "epoch": 0.1775390625, + "grad_norm": 0.32896143198013306, + "learning_rate": 0.00047033266176634356, + "loss": 4.6992, + "step": 1818 + }, + { + "epoch": 0.17763671875, + "grad_norm": 0.27983352541923523, + "learning_rate": 0.00047029788310932406, + "loss": 4.7266, + "step": 1819 + }, + { + "epoch": 0.177734375, + "grad_norm": 0.29561662673950195, + "learning_rate": 0.0004702630855195003, + "loss": 4.6914, + "step": 1820 + }, + { + "epoch": 0.17783203125, + "grad_norm": 0.2955554127693176, + "learning_rate": 0.0004702282690002458, + "loss": 4.6875, + "step": 1821 + }, + { + "epoch": 0.1779296875, + "grad_norm": 0.31012076139450073, + "learning_rate": 0.00047019343355493554, + "loss": 4.7148, + "step": 1822 + }, + { + "epoch": 0.17802734375, + "grad_norm": 0.3344959318637848, + "learning_rate": 0.00047015857918694665, + "loss": 4.7578, + "step": 1823 + }, + { + "epoch": 0.178125, + "grad_norm": 0.40076538920402527, + "learning_rate": 0.00047012370589965814, + "loss": 4.7227, + "step": 1824 + }, + { + "epoch": 0.17822265625, + "grad_norm": 0.42246541380882263, + "learning_rate": 0.0004700888136964506, + "loss": 4.707, + "step": 1825 + }, + { + "epoch": 0.1783203125, + "grad_norm": 0.4614593982696533, + "learning_rate": 0.00047005390258070663, + "loss": 4.6836, + "step": 1826 + }, + { + "epoch": 0.17841796875, + "grad_norm": 0.4689350426197052, + "learning_rate": 0.0004700189725558107, + "loss": 4.6875, + "step": 1827 + }, + { + "epoch": 0.178515625, + "grad_norm": 0.4179334044456482, + "learning_rate": 0.00046998402362514893, + "loss": 4.6797, + "step": 1828 + }, + { + "epoch": 0.17861328125, + "grad_norm": 0.3429284989833832, + "learning_rate": 0.00046994905579210937, + "loss": 4.707, + "step": 1829 + }, + { + "epoch": 0.1787109375, + "grad_norm": 0.3537569046020508, + "learning_rate": 0.00046991406906008213, + "loss": 4.7305, + "step": 1830 + }, + { + "epoch": 0.17880859375, + "grad_norm": 0.38520169258117676, + "learning_rate": 0.00046987906343245865, + "loss": 4.7031, + "step": 1831 + }, + { + "epoch": 0.17890625, + "grad_norm": 0.3677808940410614, + "learning_rate": 0.0004698440389126327, + "loss": 4.6953, + "step": 1832 + }, + { + "epoch": 0.17900390625, + "grad_norm": 0.2886373698711395, + "learning_rate": 0.0004698089955039995, + "loss": 4.7383, + "step": 1833 + }, + { + "epoch": 0.1791015625, + "grad_norm": 0.2950337529182434, + "learning_rate": 0.0004697739332099564, + "loss": 4.7227, + "step": 1834 + }, + { + "epoch": 0.17919921875, + "grad_norm": 0.37891969084739685, + "learning_rate": 0.00046973885203390245, + "loss": 4.707, + "step": 1835 + }, + { + "epoch": 0.179296875, + "grad_norm": 0.3522384464740753, + "learning_rate": 0.0004697037519792384, + "loss": 4.7031, + "step": 1836 + }, + { + "epoch": 0.17939453125, + "grad_norm": 0.35117805004119873, + "learning_rate": 0.0004696686330493672, + "loss": 4.6953, + "step": 1837 + }, + { + "epoch": 0.1794921875, + "grad_norm": 0.30079415440559387, + "learning_rate": 0.0004696334952476931, + "loss": 4.7266, + "step": 1838 + }, + { + "epoch": 0.17958984375, + "grad_norm": 0.32153111696243286, + "learning_rate": 0.0004695983385776227, + "loss": 4.7617, + "step": 1839 + }, + { + "epoch": 0.1796875, + "grad_norm": 0.3145742416381836, + "learning_rate": 0.0004695631630425641, + "loss": 4.7344, + "step": 1840 + }, + { + "epoch": 0.17978515625, + "grad_norm": 0.30176612734794617, + "learning_rate": 0.00046952796864592727, + "loss": 4.7383, + "step": 1841 + }, + { + "epoch": 0.1798828125, + "grad_norm": 0.3110758364200592, + "learning_rate": 0.00046949275539112423, + "loss": 4.6797, + "step": 1842 + }, + { + "epoch": 0.17998046875, + "grad_norm": 0.29200172424316406, + "learning_rate": 0.0004694575232815686, + "loss": 4.7109, + "step": 1843 + }, + { + "epoch": 0.180078125, + "grad_norm": 0.311064213514328, + "learning_rate": 0.0004694222723206759, + "loss": 4.6992, + "step": 1844 + }, + { + "epoch": 0.18017578125, + "grad_norm": 0.29677876830101013, + "learning_rate": 0.0004693870025118633, + "loss": 4.6719, + "step": 1845 + }, + { + "epoch": 0.1802734375, + "grad_norm": 0.31672269105911255, + "learning_rate": 0.00046935171385855025, + "loss": 4.6602, + "step": 1846 + }, + { + "epoch": 0.18037109375, + "grad_norm": 0.34578704833984375, + "learning_rate": 0.00046931640636415755, + "loss": 4.6875, + "step": 1847 + }, + { + "epoch": 0.18046875, + "grad_norm": 0.3357842266559601, + "learning_rate": 0.00046928108003210805, + "loss": 4.6914, + "step": 1848 + }, + { + "epoch": 0.18056640625, + "grad_norm": 0.35017260909080505, + "learning_rate": 0.0004692457348658265, + "loss": 4.707, + "step": 1849 + }, + { + "epoch": 0.1806640625, + "grad_norm": 0.38192495703697205, + "learning_rate": 0.00046921037086873927, + "loss": 4.6719, + "step": 1850 + }, + { + "epoch": 0.18076171875, + "grad_norm": 0.38436266779899597, + "learning_rate": 0.0004691749880442747, + "loss": 4.7109, + "step": 1851 + }, + { + "epoch": 0.180859375, + "grad_norm": 0.3527016341686249, + "learning_rate": 0.00046913958639586295, + "loss": 4.6914, + "step": 1852 + }, + { + "epoch": 0.18095703125, + "grad_norm": 0.3024858236312866, + "learning_rate": 0.0004691041659269359, + "loss": 4.6875, + "step": 1853 + }, + { + "epoch": 0.1810546875, + "grad_norm": 0.2871468663215637, + "learning_rate": 0.00046906872664092734, + "loss": 4.6875, + "step": 1854 + }, + { + "epoch": 0.18115234375, + "grad_norm": 0.3229468762874603, + "learning_rate": 0.00046903326854127287, + "loss": 4.7109, + "step": 1855 + }, + { + "epoch": 0.18125, + "grad_norm": 0.3446221351623535, + "learning_rate": 0.0004689977916314099, + "loss": 4.6914, + "step": 1856 + }, + { + "epoch": 0.18134765625, + "grad_norm": 0.39268386363983154, + "learning_rate": 0.0004689622959147778, + "loss": 4.7031, + "step": 1857 + }, + { + "epoch": 0.1814453125, + "grad_norm": 0.4465988874435425, + "learning_rate": 0.00046892678139481744, + "loss": 4.7109, + "step": 1858 + }, + { + "epoch": 0.18154296875, + "grad_norm": 0.44179853796958923, + "learning_rate": 0.0004688912480749718, + "loss": 4.6953, + "step": 1859 + }, + { + "epoch": 0.181640625, + "grad_norm": 0.3754976987838745, + "learning_rate": 0.0004688556959586857, + "loss": 4.6836, + "step": 1860 + }, + { + "epoch": 0.18173828125, + "grad_norm": 0.3733668029308319, + "learning_rate": 0.0004688201250494055, + "loss": 4.6914, + "step": 1861 + }, + { + "epoch": 0.1818359375, + "grad_norm": 0.33280256390571594, + "learning_rate": 0.00046878453535057965, + "loss": 4.6875, + "step": 1862 + }, + { + "epoch": 0.18193359375, + "grad_norm": 0.36261236667633057, + "learning_rate": 0.00046874892686565834, + "loss": 4.6992, + "step": 1863 + }, + { + "epoch": 0.18203125, + "grad_norm": 0.3954784572124481, + "learning_rate": 0.0004687132995980935, + "loss": 4.7031, + "step": 1864 + }, + { + "epoch": 0.18212890625, + "grad_norm": 0.3743366003036499, + "learning_rate": 0.00046867765355133905, + "loss": 4.7266, + "step": 1865 + }, + { + "epoch": 0.1822265625, + "grad_norm": 0.325110524892807, + "learning_rate": 0.0004686419887288506, + "loss": 4.7188, + "step": 1866 + }, + { + "epoch": 0.18232421875, + "grad_norm": 0.28901422023773193, + "learning_rate": 0.0004686063051340856, + "loss": 4.7031, + "step": 1867 + }, + { + "epoch": 0.182421875, + "grad_norm": 0.26248887181282043, + "learning_rate": 0.00046857060277050324, + "loss": 4.6875, + "step": 1868 + }, + { + "epoch": 0.18251953125, + "grad_norm": 0.29704171419143677, + "learning_rate": 0.00046853488164156476, + "loss": 4.6953, + "step": 1869 + }, + { + "epoch": 0.1826171875, + "grad_norm": 0.3862922787666321, + "learning_rate": 0.00046849914175073305, + "loss": 4.6758, + "step": 1870 + }, + { + "epoch": 0.18271484375, + "grad_norm": 0.44130098819732666, + "learning_rate": 0.0004684633831014728, + "loss": 4.6992, + "step": 1871 + }, + { + "epoch": 0.1828125, + "grad_norm": 0.44974687695503235, + "learning_rate": 0.00046842760569725065, + "loss": 4.7188, + "step": 1872 + }, + { + "epoch": 0.18291015625, + "grad_norm": 0.3364246189594269, + "learning_rate": 0.00046839180954153485, + "loss": 4.6836, + "step": 1873 + }, + { + "epoch": 0.1830078125, + "grad_norm": 0.2785428762435913, + "learning_rate": 0.00046835599463779573, + "loss": 4.6914, + "step": 1874 + }, + { + "epoch": 0.18310546875, + "grad_norm": 0.3570761978626251, + "learning_rate": 0.0004683201609895052, + "loss": 4.7031, + "step": 1875 + }, + { + "epoch": 0.183203125, + "grad_norm": 0.4263145923614502, + "learning_rate": 0.0004682843086001371, + "loss": 4.6875, + "step": 1876 + }, + { + "epoch": 0.18330078125, + "grad_norm": 0.4154239594936371, + "learning_rate": 0.00046824843747316717, + "loss": 4.7031, + "step": 1877 + }, + { + "epoch": 0.1833984375, + "grad_norm": 0.34865015745162964, + "learning_rate": 0.0004682125476120728, + "loss": 4.7109, + "step": 1878 + }, + { + "epoch": 0.18349609375, + "grad_norm": 0.3163270950317383, + "learning_rate": 0.00046817663902033323, + "loss": 4.6953, + "step": 1879 + }, + { + "epoch": 0.18359375, + "grad_norm": 0.3564985394477844, + "learning_rate": 0.00046814071170142964, + "loss": 4.6914, + "step": 1880 + }, + { + "epoch": 0.18369140625, + "grad_norm": 0.36978796124458313, + "learning_rate": 0.0004681047656588449, + "loss": 4.6914, + "step": 1881 + }, + { + "epoch": 0.1837890625, + "grad_norm": 0.310011625289917, + "learning_rate": 0.00046806880089606375, + "loss": 4.7109, + "step": 1882 + }, + { + "epoch": 0.18388671875, + "grad_norm": 0.25916099548339844, + "learning_rate": 0.00046803281741657264, + "loss": 4.7031, + "step": 1883 + }, + { + "epoch": 0.183984375, + "grad_norm": 0.3122245669364929, + "learning_rate": 0.00046799681522386013, + "loss": 4.6953, + "step": 1884 + }, + { + "epoch": 0.18408203125, + "grad_norm": 0.3948974013328552, + "learning_rate": 0.0004679607943214162, + "loss": 4.7383, + "step": 1885 + }, + { + "epoch": 0.1841796875, + "grad_norm": 0.38093316555023193, + "learning_rate": 0.00046792475471273283, + "loss": 4.6719, + "step": 1886 + }, + { + "epoch": 0.18427734375, + "grad_norm": 0.33030474185943604, + "learning_rate": 0.0004678886964013039, + "loss": 4.7188, + "step": 1887 + }, + { + "epoch": 0.184375, + "grad_norm": 0.3063085675239563, + "learning_rate": 0.000467852619390625, + "loss": 4.6719, + "step": 1888 + }, + { + "epoch": 0.18447265625, + "grad_norm": 0.3019408881664276, + "learning_rate": 0.0004678165236841936, + "loss": 4.7031, + "step": 1889 + }, + { + "epoch": 0.1845703125, + "grad_norm": 0.32429105043411255, + "learning_rate": 0.0004677804092855088, + "loss": 4.7031, + "step": 1890 + }, + { + "epoch": 0.18466796875, + "grad_norm": 0.3127208650112152, + "learning_rate": 0.00046774427619807176, + "loss": 4.6875, + "step": 1891 + }, + { + "epoch": 0.184765625, + "grad_norm": 0.33778461813926697, + "learning_rate": 0.0004677081244253853, + "loss": 4.6523, + "step": 1892 + }, + { + "epoch": 0.18486328125, + "grad_norm": 0.32395920157432556, + "learning_rate": 0.00046767195397095406, + "loss": 4.7031, + "step": 1893 + }, + { + "epoch": 0.1849609375, + "grad_norm": 0.27252617478370667, + "learning_rate": 0.0004676357648382846, + "loss": 4.6992, + "step": 1894 + }, + { + "epoch": 0.18505859375, + "grad_norm": 0.2853223979473114, + "learning_rate": 0.0004675995570308852, + "loss": 4.6992, + "step": 1895 + }, + { + "epoch": 0.18515625, + "grad_norm": 0.2883679270744324, + "learning_rate": 0.0004675633305522658, + "loss": 4.6719, + "step": 1896 + }, + { + "epoch": 0.18525390625, + "grad_norm": 0.25876960158348083, + "learning_rate": 0.00046752708540593853, + "loss": 4.6875, + "step": 1897 + }, + { + "epoch": 0.1853515625, + "grad_norm": 0.2769707143306732, + "learning_rate": 0.00046749082159541696, + "loss": 4.707, + "step": 1898 + }, + { + "epoch": 0.18544921875, + "grad_norm": 0.3227038085460663, + "learning_rate": 0.0004674545391242167, + "loss": 4.707, + "step": 1899 + }, + { + "epoch": 0.185546875, + "grad_norm": 0.38810545206069946, + "learning_rate": 0.0004674182379958551, + "loss": 4.7305, + "step": 1900 + }, + { + "epoch": 0.18564453125, + "grad_norm": 0.479206383228302, + "learning_rate": 0.0004673819182138512, + "loss": 4.7266, + "step": 1901 + }, + { + "epoch": 0.1857421875, + "grad_norm": 0.531310498714447, + "learning_rate": 0.0004673455797817261, + "loss": 4.7109, + "step": 1902 + }, + { + "epoch": 0.18583984375, + "grad_norm": 0.4943915009498596, + "learning_rate": 0.0004673092227030024, + "loss": 4.7109, + "step": 1903 + }, + { + "epoch": 0.1859375, + "grad_norm": 0.3636752963066101, + "learning_rate": 0.0004672728469812049, + "loss": 4.7148, + "step": 1904 + }, + { + "epoch": 0.18603515625, + "grad_norm": 0.30614086985588074, + "learning_rate": 0.0004672364526198598, + "loss": 4.6953, + "step": 1905 + }, + { + "epoch": 0.1861328125, + "grad_norm": 0.36997532844543457, + "learning_rate": 0.00046720003962249525, + "loss": 4.6758, + "step": 1906 + }, + { + "epoch": 0.18623046875, + "grad_norm": 0.3727617859840393, + "learning_rate": 0.00046716360799264135, + "loss": 4.6875, + "step": 1907 + }, + { + "epoch": 0.186328125, + "grad_norm": 0.3147006034851074, + "learning_rate": 0.00046712715773382986, + "loss": 4.7109, + "step": 1908 + }, + { + "epoch": 0.18642578125, + "grad_norm": 0.3325425088405609, + "learning_rate": 0.0004670906888495945, + "loss": 4.7109, + "step": 1909 + }, + { + "epoch": 0.1865234375, + "grad_norm": 0.34814977645874023, + "learning_rate": 0.0004670542013434705, + "loss": 4.7031, + "step": 1910 + }, + { + "epoch": 0.18662109375, + "grad_norm": 0.395072340965271, + "learning_rate": 0.0004670176952189952, + "loss": 4.6914, + "step": 1911 + }, + { + "epoch": 0.18671875, + "grad_norm": 0.3087127208709717, + "learning_rate": 0.0004669811704797075, + "loss": 4.6875, + "step": 1912 + }, + { + "epoch": 0.18681640625, + "grad_norm": 0.2690333425998688, + "learning_rate": 0.00046694462712914837, + "loss": 4.6797, + "step": 1913 + }, + { + "epoch": 0.1869140625, + "grad_norm": 0.2913764417171478, + "learning_rate": 0.0004669080651708604, + "loss": 4.707, + "step": 1914 + }, + { + "epoch": 0.18701171875, + "grad_norm": 0.28636500239372253, + "learning_rate": 0.0004668714846083879, + "loss": 4.6953, + "step": 1915 + }, + { + "epoch": 0.187109375, + "grad_norm": 0.27798396348953247, + "learning_rate": 0.0004668348854452772, + "loss": 4.6953, + "step": 1916 + }, + { + "epoch": 0.18720703125, + "grad_norm": 0.3135933578014374, + "learning_rate": 0.00046679826768507646, + "loss": 4.7344, + "step": 1917 + }, + { + "epoch": 0.1873046875, + "grad_norm": 0.2792069911956787, + "learning_rate": 0.0004667616313313353, + "loss": 4.6562, + "step": 1918 + }, + { + "epoch": 0.18740234375, + "grad_norm": 0.2715051472187042, + "learning_rate": 0.00046672497638760555, + "loss": 4.707, + "step": 1919 + }, + { + "epoch": 0.1875, + "grad_norm": 0.2932470142841339, + "learning_rate": 0.0004666883028574405, + "loss": 4.7109, + "step": 1920 + }, + { + "epoch": 0.18759765625, + "grad_norm": 0.3258669078350067, + "learning_rate": 0.0004666516107443956, + "loss": 4.6953, + "step": 1921 + }, + { + "epoch": 0.1876953125, + "grad_norm": 0.35614919662475586, + "learning_rate": 0.0004666149000520277, + "loss": 4.6953, + "step": 1922 + }, + { + "epoch": 0.18779296875, + "grad_norm": 0.35679420828819275, + "learning_rate": 0.0004665781707838957, + "loss": 4.707, + "step": 1923 + }, + { + "epoch": 0.187890625, + "grad_norm": 0.3622376024723053, + "learning_rate": 0.00046654142294356033, + "loss": 4.7109, + "step": 1924 + }, + { + "epoch": 0.18798828125, + "grad_norm": 0.400008887052536, + "learning_rate": 0.00046650465653458404, + "loss": 4.6836, + "step": 1925 + }, + { + "epoch": 0.1880859375, + "grad_norm": 0.40335577726364136, + "learning_rate": 0.00046646787156053097, + "loss": 4.6875, + "step": 1926 + }, + { + "epoch": 0.18818359375, + "grad_norm": 0.37399107217788696, + "learning_rate": 0.0004664310680249673, + "loss": 4.7031, + "step": 1927 + }, + { + "epoch": 0.18828125, + "grad_norm": 0.34047555923461914, + "learning_rate": 0.0004663942459314608, + "loss": 4.6836, + "step": 1928 + }, + { + "epoch": 0.18837890625, + "grad_norm": 0.2900930643081665, + "learning_rate": 0.00046635740528358125, + "loss": 4.6914, + "step": 1929 + }, + { + "epoch": 0.1884765625, + "grad_norm": 0.3214733600616455, + "learning_rate": 0.0004663205460848999, + "loss": 4.6445, + "step": 1930 + }, + { + "epoch": 0.18857421875, + "grad_norm": 0.29667600989341736, + "learning_rate": 0.0004662836683389901, + "loss": 4.6992, + "step": 1931 + }, + { + "epoch": 0.188671875, + "grad_norm": 0.28890740871429443, + "learning_rate": 0.0004662467720494269, + "loss": 4.6836, + "step": 1932 + }, + { + "epoch": 0.18876953125, + "grad_norm": 0.3385658264160156, + "learning_rate": 0.0004662098572197872, + "loss": 4.6914, + "step": 1933 + }, + { + "epoch": 0.1888671875, + "grad_norm": 0.26928526163101196, + "learning_rate": 0.00046617292385364956, + "loss": 4.7188, + "step": 1934 + }, + { + "epoch": 0.18896484375, + "grad_norm": 0.3358452618122101, + "learning_rate": 0.00046613597195459445, + "loss": 4.6875, + "step": 1935 + }, + { + "epoch": 0.1890625, + "grad_norm": 0.3048867881298065, + "learning_rate": 0.0004660990015262041, + "loss": 4.6836, + "step": 1936 + }, + { + "epoch": 0.18916015625, + "grad_norm": 0.2835708558559418, + "learning_rate": 0.00046606201257206253, + "loss": 4.7031, + "step": 1937 + }, + { + "epoch": 0.1892578125, + "grad_norm": 0.29097795486450195, + "learning_rate": 0.0004660250050957556, + "loss": 4.6875, + "step": 1938 + }, + { + "epoch": 0.18935546875, + "grad_norm": 0.3146597445011139, + "learning_rate": 0.00046598797910087086, + "loss": 4.6992, + "step": 1939 + }, + { + "epoch": 0.189453125, + "grad_norm": 0.37111765146255493, + "learning_rate": 0.00046595093459099793, + "loss": 4.668, + "step": 1940 + }, + { + "epoch": 0.18955078125, + "grad_norm": 0.4816044867038727, + "learning_rate": 0.0004659138715697278, + "loss": 4.6641, + "step": 1941 + }, + { + "epoch": 0.1896484375, + "grad_norm": 0.6332393884658813, + "learning_rate": 0.0004658767900406535, + "loss": 4.7031, + "step": 1942 + }, + { + "epoch": 0.18974609375, + "grad_norm": 0.598022997379303, + "learning_rate": 0.00046583969000737, + "loss": 4.7148, + "step": 1943 + }, + { + "epoch": 0.18984375, + "grad_norm": 0.37639319896698, + "learning_rate": 0.00046580257147347366, + "loss": 4.6797, + "step": 1944 + }, + { + "epoch": 0.18994140625, + "grad_norm": 0.411358505487442, + "learning_rate": 0.00046576543444256307, + "loss": 4.6875, + "step": 1945 + }, + { + "epoch": 0.1900390625, + "grad_norm": 0.5302320718765259, + "learning_rate": 0.00046572827891823833, + "loss": 4.6992, + "step": 1946 + }, + { + "epoch": 0.19013671875, + "grad_norm": 0.41106823086738586, + "learning_rate": 0.0004656911049041014, + "loss": 4.6602, + "step": 1947 + }, + { + "epoch": 0.190234375, + "grad_norm": 0.2646370232105255, + "learning_rate": 0.00046565391240375607, + "loss": 4.7031, + "step": 1948 + }, + { + "epoch": 0.19033203125, + "grad_norm": 0.34403374791145325, + "learning_rate": 0.000465616701420808, + "loss": 4.7422, + "step": 1949 + }, + { + "epoch": 0.1904296875, + "grad_norm": 0.32895779609680176, + "learning_rate": 0.00046557947195886433, + "loss": 4.6992, + "step": 1950 + }, + { + "epoch": 0.19052734375, + "grad_norm": 0.2670215666294098, + "learning_rate": 0.0004655422240215344, + "loss": 4.6992, + "step": 1951 + }, + { + "epoch": 0.190625, + "grad_norm": 0.2915528118610382, + "learning_rate": 0.00046550495761242907, + "loss": 4.6875, + "step": 1952 + }, + { + "epoch": 0.19072265625, + "grad_norm": 0.30449408292770386, + "learning_rate": 0.000465467672735161, + "loss": 4.6523, + "step": 1953 + }, + { + "epoch": 0.1908203125, + "grad_norm": 0.30098956823349, + "learning_rate": 0.00046543036939334476, + "loss": 4.6914, + "step": 1954 + }, + { + "epoch": 0.19091796875, + "grad_norm": 0.31237950921058655, + "learning_rate": 0.0004653930475905967, + "loss": 4.7031, + "step": 1955 + }, + { + "epoch": 0.191015625, + "grad_norm": 0.3021314740180969, + "learning_rate": 0.0004653557073305349, + "loss": 4.6562, + "step": 1956 + }, + { + "epoch": 0.19111328125, + "grad_norm": 0.29421451687812805, + "learning_rate": 0.0004653183486167792, + "loss": 4.6758, + "step": 1957 + }, + { + "epoch": 0.1912109375, + "grad_norm": 0.3067154884338379, + "learning_rate": 0.0004652809714529512, + "loss": 4.7344, + "step": 1958 + }, + { + "epoch": 0.19130859375, + "grad_norm": 0.3098396360874176, + "learning_rate": 0.0004652435758426746, + "loss": 4.6836, + "step": 1959 + }, + { + "epoch": 0.19140625, + "grad_norm": 0.3248749375343323, + "learning_rate": 0.00046520616178957454, + "loss": 4.6797, + "step": 1960 + }, + { + "epoch": 0.19150390625, + "grad_norm": 0.36084792017936707, + "learning_rate": 0.00046516872929727786, + "loss": 4.6875, + "step": 1961 + }, + { + "epoch": 0.1916015625, + "grad_norm": 0.3105924725532532, + "learning_rate": 0.0004651312783694137, + "loss": 4.707, + "step": 1962 + }, + { + "epoch": 0.19169921875, + "grad_norm": 0.2764946520328522, + "learning_rate": 0.0004650938090096125, + "loss": 4.6797, + "step": 1963 + }, + { + "epoch": 0.191796875, + "grad_norm": 0.2781250476837158, + "learning_rate": 0.0004650563212215066, + "loss": 4.6914, + "step": 1964 + }, + { + "epoch": 0.19189453125, + "grad_norm": 0.3017931282520294, + "learning_rate": 0.00046501881500873036, + "loss": 4.6914, + "step": 1965 + }, + { + "epoch": 0.1919921875, + "grad_norm": 0.32291722297668457, + "learning_rate": 0.0004649812903749196, + "loss": 4.6562, + "step": 1966 + }, + { + "epoch": 0.19208984375, + "grad_norm": 0.30429041385650635, + "learning_rate": 0.0004649437473237122, + "loss": 4.6953, + "step": 1967 + }, + { + "epoch": 0.1921875, + "grad_norm": 0.30845269560813904, + "learning_rate": 0.0004649061858587476, + "loss": 4.6914, + "step": 1968 + }, + { + "epoch": 0.19228515625, + "grad_norm": 0.28626877069473267, + "learning_rate": 0.0004648686059836672, + "loss": 4.7188, + "step": 1969 + }, + { + "epoch": 0.1923828125, + "grad_norm": 0.2860487997531891, + "learning_rate": 0.0004648310077021141, + "loss": 4.6875, + "step": 1970 + }, + { + "epoch": 0.19248046875, + "grad_norm": 0.2899281978607178, + "learning_rate": 0.00046479339101773313, + "loss": 4.7266, + "step": 1971 + }, + { + "epoch": 0.192578125, + "grad_norm": 0.3460501432418823, + "learning_rate": 0.0004647557559341712, + "loss": 4.6797, + "step": 1972 + }, + { + "epoch": 0.19267578125, + "grad_norm": 0.32573822140693665, + "learning_rate": 0.00046471810245507646, + "loss": 4.6562, + "step": 1973 + }, + { + "epoch": 0.1927734375, + "grad_norm": 0.31463614106178284, + "learning_rate": 0.00046468043058409933, + "loss": 4.668, + "step": 1974 + }, + { + "epoch": 0.19287109375, + "grad_norm": 0.34310510754585266, + "learning_rate": 0.0004646427403248919, + "loss": 4.6914, + "step": 1975 + }, + { + "epoch": 0.19296875, + "grad_norm": 0.43290114402770996, + "learning_rate": 0.0004646050316811078, + "loss": 4.6758, + "step": 1976 + }, + { + "epoch": 0.19306640625, + "grad_norm": 0.5048022270202637, + "learning_rate": 0.00046456730465640274, + "loss": 4.6719, + "step": 1977 + }, + { + "epoch": 0.1931640625, + "grad_norm": 0.5019853711128235, + "learning_rate": 0.00046452955925443414, + "loss": 4.6992, + "step": 1978 + }, + { + "epoch": 0.19326171875, + "grad_norm": 0.354935884475708, + "learning_rate": 0.00046449179547886104, + "loss": 4.6602, + "step": 1979 + }, + { + "epoch": 0.193359375, + "grad_norm": 0.29509827494621277, + "learning_rate": 0.00046445401333334457, + "loss": 4.6758, + "step": 1980 + }, + { + "epoch": 0.19345703125, + "grad_norm": 0.3451140224933624, + "learning_rate": 0.00046441621282154727, + "loss": 4.7148, + "step": 1981 + }, + { + "epoch": 0.1935546875, + "grad_norm": 0.3382677435874939, + "learning_rate": 0.00046437839394713364, + "loss": 4.6992, + "step": 1982 + }, + { + "epoch": 0.19365234375, + "grad_norm": 0.32868877053260803, + "learning_rate": 0.0004643405567137702, + "loss": 4.6914, + "step": 1983 + }, + { + "epoch": 0.19375, + "grad_norm": 0.3491230607032776, + "learning_rate": 0.00046430270112512474, + "loss": 4.6953, + "step": 1984 + }, + { + "epoch": 0.19384765625, + "grad_norm": 0.36351510882377625, + "learning_rate": 0.00046426482718486725, + "loss": 4.6797, + "step": 1985 + }, + { + "epoch": 0.1939453125, + "grad_norm": 0.32560887932777405, + "learning_rate": 0.00046422693489666923, + "loss": 4.6758, + "step": 1986 + }, + { + "epoch": 0.19404296875, + "grad_norm": 0.28270575404167175, + "learning_rate": 0.0004641890242642042, + "loss": 4.707, + "step": 1987 + }, + { + "epoch": 0.194140625, + "grad_norm": 0.2650778889656067, + "learning_rate": 0.0004641510952911473, + "loss": 4.6602, + "step": 1988 + }, + { + "epoch": 0.19423828125, + "grad_norm": 0.2566971182823181, + "learning_rate": 0.00046411314798117543, + "loss": 4.6797, + "step": 1989 + }, + { + "epoch": 0.1943359375, + "grad_norm": 0.25341349840164185, + "learning_rate": 0.00046407518233796747, + "loss": 4.6641, + "step": 1990 + }, + { + "epoch": 0.19443359375, + "grad_norm": 0.24855858087539673, + "learning_rate": 0.0004640371983652038, + "loss": 4.6914, + "step": 1991 + }, + { + "epoch": 0.19453125, + "grad_norm": 0.2411157190799713, + "learning_rate": 0.0004639991960665668, + "loss": 4.6875, + "step": 1992 + }, + { + "epoch": 0.19462890625, + "grad_norm": 0.28344714641571045, + "learning_rate": 0.00046396117544574033, + "loss": 4.6719, + "step": 1993 + }, + { + "epoch": 0.1947265625, + "grad_norm": 0.26913028955459595, + "learning_rate": 0.00046392313650641054, + "loss": 4.6953, + "step": 1994 + }, + { + "epoch": 0.19482421875, + "grad_norm": 0.26939257979393005, + "learning_rate": 0.0004638850792522649, + "loss": 4.7266, + "step": 1995 + }, + { + "epoch": 0.194921875, + "grad_norm": 0.26686719059944153, + "learning_rate": 0.0004638470036869927, + "loss": 4.6836, + "step": 1996 + }, + { + "epoch": 0.19501953125, + "grad_norm": 0.27638185024261475, + "learning_rate": 0.00046380890981428523, + "loss": 4.6797, + "step": 1997 + }, + { + "epoch": 0.1951171875, + "grad_norm": 0.3258664608001709, + "learning_rate": 0.00046377079763783535, + "loss": 4.6562, + "step": 1998 + }, + { + "epoch": 0.19521484375, + "grad_norm": 0.3155481219291687, + "learning_rate": 0.0004637326671613379, + "loss": 4.6836, + "step": 1999 + }, + { + "epoch": 0.1953125, + "grad_norm": 0.3607693314552307, + "learning_rate": 0.0004636945183884893, + "loss": 4.6641, + "step": 2000 + } + ], + "logging_steps": 1.0, + "max_steps": 10240, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 200, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 1.2245011890438144e+19, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +}