| { | |
| "best_metric": 0.7420778274536133, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 2.26628895184136, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0113314447592068, | |
| "grad_norm": 1.4940721988677979, | |
| "learning_rate": 1e-05, | |
| "loss": 1.0261, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0113314447592068, | |
| "eval_loss": 1.1746464967727661, | |
| "eval_runtime": 11.1835, | |
| "eval_samples_per_second": 13.323, | |
| "eval_steps_per_second": 3.398, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0226628895184136, | |
| "grad_norm": 1.7041244506835938, | |
| "learning_rate": 2e-05, | |
| "loss": 1.1102, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0339943342776204, | |
| "grad_norm": 1.8132150173187256, | |
| "learning_rate": 3e-05, | |
| "loss": 1.1672, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0453257790368272, | |
| "grad_norm": 1.4703266620635986, | |
| "learning_rate": 4e-05, | |
| "loss": 1.0816, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.056657223796033995, | |
| "grad_norm": 0.9214251637458801, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0017, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0679886685552408, | |
| "grad_norm": 0.5252280235290527, | |
| "learning_rate": 6e-05, | |
| "loss": 0.98, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.07932011331444759, | |
| "grad_norm": 0.7931201457977295, | |
| "learning_rate": 7e-05, | |
| "loss": 1.0041, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0906515580736544, | |
| "grad_norm": 0.9411036372184753, | |
| "learning_rate": 8e-05, | |
| "loss": 1.0307, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.10198300283286119, | |
| "grad_norm": 0.6796821355819702, | |
| "learning_rate": 9e-05, | |
| "loss": 1.0361, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.11331444759206799, | |
| "grad_norm": 0.48324570059776306, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9964, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.12464589235127478, | |
| "grad_norm": 0.37594151496887207, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.9495, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.1359773371104816, | |
| "grad_norm": 0.47099214792251587, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.9703, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.14730878186968838, | |
| "grad_norm": 0.4177071452140808, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.9866, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.15864022662889518, | |
| "grad_norm": 0.4068276584148407, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.9656, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.16997167138810199, | |
| "grad_norm": 0.37006956338882446, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.0134, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.1813031161473088, | |
| "grad_norm": 0.3679398000240326, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.9677, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.19263456090651557, | |
| "grad_norm": 0.352753609418869, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.9623, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.20396600566572237, | |
| "grad_norm": 0.32562100887298584, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.9627, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.21529745042492918, | |
| "grad_norm": 0.3547748327255249, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.9764, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.22662889518413598, | |
| "grad_norm": 0.4081428050994873, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.022, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.23796033994334279, | |
| "grad_norm": 0.4699523150920868, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.879, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.24929178470254956, | |
| "grad_norm": 1.1098036766052246, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.91, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.26062322946175637, | |
| "grad_norm": 0.3437006175518036, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.8645, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.2719546742209632, | |
| "grad_norm": 0.32103949785232544, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.8644, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.28328611898017, | |
| "grad_norm": 0.339089572429657, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.9377, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.29461756373937675, | |
| "grad_norm": 0.3423368036746979, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.9106, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.3059490084985836, | |
| "grad_norm": 0.2850959599018097, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.8449, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.31728045325779036, | |
| "grad_norm": 0.29601606726646423, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.919, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.3286118980169972, | |
| "grad_norm": 0.30124685168266296, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.9002, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.33994334277620397, | |
| "grad_norm": 0.2979277968406677, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.8651, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.35127478753541075, | |
| "grad_norm": 0.2953733503818512, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.9512, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.3626062322946176, | |
| "grad_norm": 0.2760586440563202, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.9007, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.37393767705382436, | |
| "grad_norm": 0.2794976532459259, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.9375, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.38526912181303113, | |
| "grad_norm": 0.3017430305480957, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.936, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.39660056657223797, | |
| "grad_norm": 0.316226601600647, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.9133, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.40793201133144474, | |
| "grad_norm": 0.2935938835144043, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.8939, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.4192634560906516, | |
| "grad_norm": 0.296953409910202, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.9582, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.43059490084985835, | |
| "grad_norm": 0.3016681373119354, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.9365, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.44192634560906513, | |
| "grad_norm": 0.2930508852005005, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.8891, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.45325779036827196, | |
| "grad_norm": 0.32635068893432617, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.8874, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.46458923512747874, | |
| "grad_norm": 0.3173358738422394, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.8823, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.47592067988668557, | |
| "grad_norm": 0.31149113178253174, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.8917, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.48725212464589235, | |
| "grad_norm": 0.38314029574394226, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.9208, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.4985835694050991, | |
| "grad_norm": 1.4880560636520386, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.5576, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.509915014164306, | |
| "grad_norm": 0.25400444865226746, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.7591, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.5212464589235127, | |
| "grad_norm": 0.26966241002082825, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.8677, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.5325779036827195, | |
| "grad_norm": 0.27806952595710754, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.8674, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.5439093484419264, | |
| "grad_norm": 0.26129311323165894, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.8704, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.5552407932011332, | |
| "grad_norm": 0.28313538432121277, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.8913, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.56657223796034, | |
| "grad_norm": 0.2700784504413605, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.8516, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.56657223796034, | |
| "eval_loss": 0.8685831427574158, | |
| "eval_runtime": 11.3624, | |
| "eval_samples_per_second": 13.113, | |
| "eval_steps_per_second": 3.344, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.5779036827195467, | |
| "grad_norm": 0.2857685983181, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.8437, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.5892351274787535, | |
| "grad_norm": 0.28413069248199463, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.9037, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.6005665722379604, | |
| "grad_norm": 0.27404722571372986, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.9212, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.6118980169971672, | |
| "grad_norm": 0.2649526000022888, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.8784, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.623229461756374, | |
| "grad_norm": 0.2947290539741516, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.8644, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.6345609065155807, | |
| "grad_norm": 0.27699536085128784, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.8749, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.6458923512747875, | |
| "grad_norm": 0.2775574326515198, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.8695, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.6572237960339944, | |
| "grad_norm": 0.2814454436302185, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.8573, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.6685552407932012, | |
| "grad_norm": 0.2803303301334381, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.891, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.6798866855524079, | |
| "grad_norm": 0.28773200511932373, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.8753, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.6912181303116147, | |
| "grad_norm": 0.29724711179733276, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.9011, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.7025495750708215, | |
| "grad_norm": 0.32904359698295593, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.9286, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.7138810198300283, | |
| "grad_norm": 0.3503016531467438, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.8743, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.7252124645892352, | |
| "grad_norm": 0.3214200437068939, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.8733, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.7365439093484419, | |
| "grad_norm": 0.3665679693222046, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.8802, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.7478753541076487, | |
| "grad_norm": 0.737060010433197, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.9375, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.7592067988668555, | |
| "grad_norm": 0.3130401372909546, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.841, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.7705382436260623, | |
| "grad_norm": 0.2984524369239807, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.8193, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.7818696883852692, | |
| "grad_norm": 0.29503563046455383, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.8554, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.7932011331444759, | |
| "grad_norm": 0.28448018431663513, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.8187, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.8045325779036827, | |
| "grad_norm": 0.295426607131958, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.8892, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.8158640226628895, | |
| "grad_norm": 0.29137861728668213, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.8098, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.8271954674220963, | |
| "grad_norm": 0.29083821177482605, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.8292, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.8385269121813032, | |
| "grad_norm": 0.2817254066467285, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.8449, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.8498583569405099, | |
| "grad_norm": 0.2846548557281494, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.8838, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.8611898016997167, | |
| "grad_norm": 0.2862998843193054, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.838, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.8725212464589235, | |
| "grad_norm": 0.2817937433719635, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.8634, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.8838526912181303, | |
| "grad_norm": 0.29679593443870544, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.8735, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.8951841359773371, | |
| "grad_norm": 0.2932203710079193, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.8345, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.9065155807365439, | |
| "grad_norm": 0.29441186785697937, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.8419, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.9178470254957507, | |
| "grad_norm": 0.29725492000579834, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.8608, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.9291784702549575, | |
| "grad_norm": 0.3129102289676666, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.8679, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.9405099150141643, | |
| "grad_norm": 0.32254207134246826, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.8656, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.9518413597733711, | |
| "grad_norm": 0.31700399518013, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.8419, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.9631728045325779, | |
| "grad_norm": 0.36216357350349426, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.8015, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.9745042492917847, | |
| "grad_norm": 0.3494912087917328, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.8515, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.9858356940509915, | |
| "grad_norm": 0.373202383518219, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.7379, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.9971671388101983, | |
| "grad_norm": 0.7268677353858948, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.6286, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.0084985835694051, | |
| "grad_norm": 0.654803991317749, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.3468, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.019830028328612, | |
| "grad_norm": 0.26078087091445923, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.7179, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.0311614730878187, | |
| "grad_norm": 0.26947835087776184, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.7735, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.0424929178470255, | |
| "grad_norm": 0.27178534865379333, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.7403, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.0538243626062322, | |
| "grad_norm": 0.27182912826538086, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.7573, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.065155807365439, | |
| "grad_norm": 0.29053595662117004, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.7674, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.0764872521246458, | |
| "grad_norm": 0.3091391324996948, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.7174, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.0878186968838528, | |
| "grad_norm": 0.2862534821033478, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.719, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.0991501416430596, | |
| "grad_norm": 0.2900041937828064, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.7355, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.1104815864022664, | |
| "grad_norm": 0.2984274625778198, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.6907, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.1218130311614731, | |
| "grad_norm": 0.31093156337738037, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.7234, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.13314447592068, | |
| "grad_norm": 0.32328957319259644, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.7473, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.13314447592068, | |
| "eval_loss": 0.8013013601303101, | |
| "eval_runtime": 11.3593, | |
| "eval_samples_per_second": 13.117, | |
| "eval_steps_per_second": 3.345, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.1444759206798867, | |
| "grad_norm": 0.31163015961647034, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.6997, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.1558073654390935, | |
| "grad_norm": 0.3146006762981415, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.7283, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.1671388101983002, | |
| "grad_norm": 0.3301478326320648, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.7262, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.178470254957507, | |
| "grad_norm": 0.3398320972919464, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.7361, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.1898016997167138, | |
| "grad_norm": 0.36190029978752136, | |
| "learning_rate": 5e-05, | |
| "loss": 0.7502, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.2011331444759206, | |
| "grad_norm": 0.3504740595817566, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.6816, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.2124645892351276, | |
| "grad_norm": 0.40286341309547424, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.7556, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.2237960339943343, | |
| "grad_norm": 0.4191229045391083, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.7204, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.2351274787535411, | |
| "grad_norm": 0.42308875918388367, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.7062, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.246458923512748, | |
| "grad_norm": 0.6090673208236694, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.622, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.2577903682719547, | |
| "grad_norm": 0.3646659851074219, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.5627, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.2691218130311614, | |
| "grad_norm": 0.4557804465293884, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.7121, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.2804532577903682, | |
| "grad_norm": 0.3935948610305786, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.7158, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.291784702549575, | |
| "grad_norm": 0.3726375102996826, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.774, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.3031161473087818, | |
| "grad_norm": 0.3537135422229767, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.7672, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.3144475920679888, | |
| "grad_norm": 0.35993891954421997, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.7034, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.3257790368271953, | |
| "grad_norm": 0.3584566116333008, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.7001, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.3371104815864023, | |
| "grad_norm": 0.37049055099487305, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.7125, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.348441926345609, | |
| "grad_norm": 0.35479503870010376, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.7248, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.3597733711048159, | |
| "grad_norm": 0.3559660315513611, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.7629, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.3711048158640227, | |
| "grad_norm": 0.34223610162734985, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.6785, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.3824362606232294, | |
| "grad_norm": 0.3551385998725891, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.7203, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.3937677053824362, | |
| "grad_norm": 0.3625800907611847, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.7097, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.405099150141643, | |
| "grad_norm": 0.3674543797969818, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.7275, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.41643059490085, | |
| "grad_norm": 0.35478031635284424, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.6774, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.4277620396600565, | |
| "grad_norm": 0.39804625511169434, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.7227, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.4390934844192635, | |
| "grad_norm": 0.402412474155426, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.6975, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.4504249291784703, | |
| "grad_norm": 0.3918543756008148, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.6646, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.461756373937677, | |
| "grad_norm": 0.4310785233974457, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.6273, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.4730878186968839, | |
| "grad_norm": 0.47671714425086975, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.5883, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.4844192634560907, | |
| "grad_norm": 0.4919561445713043, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.5803, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.4957507082152974, | |
| "grad_norm": 0.7754276990890503, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.4437, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.5070821529745042, | |
| "grad_norm": 0.36054930090904236, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.5229, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.5184135977337112, | |
| "grad_norm": 0.4418894946575165, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.6609, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.5297450424929178, | |
| "grad_norm": 0.5525121092796326, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.7335, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.5410764872521248, | |
| "grad_norm": 0.4579167068004608, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.6938, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.5524079320113313, | |
| "grad_norm": 0.39674249291419983, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.6535, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.5637393767705383, | |
| "grad_norm": 0.4052182734012604, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.6701, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.575070821529745, | |
| "grad_norm": 0.4056691527366638, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.7056, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.5864022662889519, | |
| "grad_norm": 0.37060877680778503, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.6789, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.5977337110481586, | |
| "grad_norm": 0.3834247291088104, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.7551, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.6090651558073654, | |
| "grad_norm": 0.37477755546569824, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.685, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.6203966005665722, | |
| "grad_norm": 0.3884474039077759, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.7001, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.631728045325779, | |
| "grad_norm": 0.39254558086395264, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.6391, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.643059490084986, | |
| "grad_norm": 0.3775780498981476, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.6829, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.6543909348441925, | |
| "grad_norm": 0.3820933699607849, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.6711, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.6657223796033995, | |
| "grad_norm": 0.38209760189056396, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.6857, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.677053824362606, | |
| "grad_norm": 0.3991166055202484, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.6806, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.688385269121813, | |
| "grad_norm": 0.4100518226623535, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.6901, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.6997167138810199, | |
| "grad_norm": 0.4116714596748352, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.6445, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.6997167138810199, | |
| "eval_loss": 0.7521026134490967, | |
| "eval_runtime": 11.3497, | |
| "eval_samples_per_second": 13.128, | |
| "eval_steps_per_second": 3.348, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.7110481586402266, | |
| "grad_norm": 0.44452881813049316, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.6401, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.7223796033994334, | |
| "grad_norm": 0.447711706161499, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.6424, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.7337110481586402, | |
| "grad_norm": 0.48492729663848877, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.6073, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.7450424929178472, | |
| "grad_norm": 0.8506810069084167, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.5074, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.7563739376770537, | |
| "grad_norm": 0.3347216844558716, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.5402, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.7677053824362607, | |
| "grad_norm": 0.4024171531200409, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.7443, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.7790368271954673, | |
| "grad_norm": 0.41074129939079285, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.7573, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.7903682719546743, | |
| "grad_norm": 0.3707183003425598, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.6854, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.801699716713881, | |
| "grad_norm": 0.39143168926239014, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.698, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.8130311614730878, | |
| "grad_norm": 0.3817121982574463, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.6419, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.8243626062322946, | |
| "grad_norm": 0.38814979791641235, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.6402, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.8356940509915014, | |
| "grad_norm": 0.3999271094799042, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.6806, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.8470254957507082, | |
| "grad_norm": 0.3903258442878723, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.7103, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.858356940509915, | |
| "grad_norm": 0.39158394932746887, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.7021, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.869688385269122, | |
| "grad_norm": 0.3828256130218506, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.6773, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.8810198300283285, | |
| "grad_norm": 0.40929779410362244, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.6684, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.8923512747875355, | |
| "grad_norm": 0.37998712062835693, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.6873, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.903682719546742, | |
| "grad_norm": 0.37896665930747986, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.6508, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.915014164305949, | |
| "grad_norm": 0.40087971091270447, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.6933, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.9263456090651558, | |
| "grad_norm": 0.4156991243362427, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.6888, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.9376770538243626, | |
| "grad_norm": 0.4046257734298706, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.6119, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.9490084985835694, | |
| "grad_norm": 0.3981984555721283, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.5937, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.9603399433427762, | |
| "grad_norm": 0.4312017261981964, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.6388, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.9716713881019832, | |
| "grad_norm": 0.44606897234916687, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.6094, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.9830028328611897, | |
| "grad_norm": 0.4601079523563385, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.6116, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.9943342776203967, | |
| "grad_norm": 0.7336193919181824, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.4964, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 2.0056657223796033, | |
| "grad_norm": 0.8930611610412598, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.9594, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 2.0169971671388103, | |
| "grad_norm": 0.30886948108673096, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.6154, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 2.028328611898017, | |
| "grad_norm": 0.3591644763946533, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.621, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 2.039660056657224, | |
| "grad_norm": 0.3551977574825287, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.6448, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.0509915014164304, | |
| "grad_norm": 0.3618852198123932, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.6173, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 2.0623229461756374, | |
| "grad_norm": 0.3655000329017639, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.6326, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 2.0736543909348444, | |
| "grad_norm": 0.3681482672691345, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.6006, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 2.084985835694051, | |
| "grad_norm": 0.3835912346839905, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.6109, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 2.096317280453258, | |
| "grad_norm": 0.3807328939437866, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.6465, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 2.1076487252124645, | |
| "grad_norm": 0.35800546407699585, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.6601, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 2.1189801699716715, | |
| "grad_norm": 0.36241623759269714, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.6047, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.130311614730878, | |
| "grad_norm": 0.3821946084499359, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.5959, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 2.141643059490085, | |
| "grad_norm": 0.38033100962638855, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.6331, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 2.1529745042492916, | |
| "grad_norm": 0.39716804027557373, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.5891, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.1643059490084986, | |
| "grad_norm": 0.3756254017353058, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.6068, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 2.1756373937677056, | |
| "grad_norm": 0.395888090133667, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.6375, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 2.186968838526912, | |
| "grad_norm": 0.3884637653827667, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.6068, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 2.198300283286119, | |
| "grad_norm": 0.4123586416244507, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.5634, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 2.2096317280453257, | |
| "grad_norm": 0.4333895742893219, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.5411, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 2.2209631728045327, | |
| "grad_norm": 0.5002503991127014, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.5667, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 2.2322946175637393, | |
| "grad_norm": 0.5092443823814392, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.5003, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 2.2436260623229463, | |
| "grad_norm": 0.7444424629211426, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.5318, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 2.254957507082153, | |
| "grad_norm": 0.2637498676776886, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.3755, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 2.26628895184136, | |
| "grad_norm": 0.3529297411441803, | |
| "learning_rate": 0.0, | |
| "loss": 0.6149, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.26628895184136, | |
| "eval_loss": 0.7420778274536133, | |
| "eval_runtime": 11.3419, | |
| "eval_samples_per_second": 13.137, | |
| "eval_steps_per_second": 3.35, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.038712040186184e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |