| { | |
| "best_metric": 3.5797412395477295, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.06587615283267458, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00032938076416337287, | |
| "grad_norm": 0.28886115550994873, | |
| "learning_rate": 1e-05, | |
| "loss": 2.2386, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00032938076416337287, | |
| "eval_loss": 4.896640777587891, | |
| "eval_runtime": 364.4007, | |
| "eval_samples_per_second": 14.031, | |
| "eval_steps_per_second": 3.51, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0006587615283267457, | |
| "grad_norm": 0.47013208270072937, | |
| "learning_rate": 2e-05, | |
| "loss": 2.4952, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0009881422924901185, | |
| "grad_norm": 0.7144032120704651, | |
| "learning_rate": 3e-05, | |
| "loss": 2.5533, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0013175230566534915, | |
| "grad_norm": 0.7053059339523315, | |
| "learning_rate": 4e-05, | |
| "loss": 2.8317, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0016469038208168643, | |
| "grad_norm": 0.6332340240478516, | |
| "learning_rate": 5e-05, | |
| "loss": 2.8384, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.001976284584980237, | |
| "grad_norm": 0.7503998279571533, | |
| "learning_rate": 6e-05, | |
| "loss": 3.1517, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.00230566534914361, | |
| "grad_norm": 0.7341294288635254, | |
| "learning_rate": 7e-05, | |
| "loss": 3.0199, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.002635046113306983, | |
| "grad_norm": 0.7393780946731567, | |
| "learning_rate": 8e-05, | |
| "loss": 3.0835, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0029644268774703555, | |
| "grad_norm": 0.8398771286010742, | |
| "learning_rate": 9e-05, | |
| "loss": 3.1712, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0032938076416337285, | |
| "grad_norm": 0.8553745746612549, | |
| "learning_rate": 0.0001, | |
| "loss": 3.1214, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0036231884057971015, | |
| "grad_norm": 0.9420980215072632, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 3.2433, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.003952569169960474, | |
| "grad_norm": 0.9858407974243164, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 3.1902, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.004281949934123847, | |
| "grad_norm": 0.8567302227020264, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 3.2124, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.00461133069828722, | |
| "grad_norm": 1.1642296314239502, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 3.2996, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.004940711462450593, | |
| "grad_norm": 1.1074599027633667, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 3.5617, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.005270092226613966, | |
| "grad_norm": 1.1594340801239014, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 3.5334, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.005599472990777339, | |
| "grad_norm": 1.0779310464859009, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 3.228, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.005928853754940711, | |
| "grad_norm": 1.0369453430175781, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 3.6012, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.006258234519104084, | |
| "grad_norm": 1.4606853723526, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 3.6378, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.006587615283267457, | |
| "grad_norm": 1.2403980493545532, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 3.4832, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.00691699604743083, | |
| "grad_norm": 1.16143798828125, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 3.527, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.007246376811594203, | |
| "grad_norm": 1.260703206062317, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 3.4012, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.007575757575757576, | |
| "grad_norm": 1.3434325456619263, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 3.6272, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.007905138339920948, | |
| "grad_norm": 1.3341567516326904, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 3.5262, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.008234519104084322, | |
| "grad_norm": 1.2552639245986938, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 3.5141, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.008563899868247694, | |
| "grad_norm": 1.6191933155059814, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 3.5698, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.008893280632411068, | |
| "grad_norm": 1.7251110076904297, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 3.6391, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.00922266139657444, | |
| "grad_norm": 1.3592276573181152, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 3.5411, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.009552042160737812, | |
| "grad_norm": 1.4254640340805054, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 3.3193, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.009881422924901186, | |
| "grad_norm": 1.4740948677062988, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 3.6591, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.010210803689064558, | |
| "grad_norm": 1.5142802000045776, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 3.5248, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.010540184453227932, | |
| "grad_norm": 1.5121207237243652, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 3.7445, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.010869565217391304, | |
| "grad_norm": 1.9710557460784912, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 4.0837, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.011198945981554678, | |
| "grad_norm": 2.0302300453186035, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 3.8267, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.01152832674571805, | |
| "grad_norm": 2.0829148292541504, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 3.773, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.011857707509881422, | |
| "grad_norm": 2.2299692630767822, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 3.7738, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.012187088274044796, | |
| "grad_norm": 2.5526123046875, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 3.7843, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.012516469038208168, | |
| "grad_norm": 2.357074499130249, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 4.2314, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.012845849802371542, | |
| "grad_norm": 2.5620172023773193, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 4.0925, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.013175230566534914, | |
| "grad_norm": 2.9400711059570312, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 4.0555, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.013504611330698288, | |
| "grad_norm": 2.570716381072998, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 3.8589, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.01383399209486166, | |
| "grad_norm": 2.8535444736480713, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 4.3069, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.014163372859025032, | |
| "grad_norm": 2.6688196659088135, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 3.9936, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.014492753623188406, | |
| "grad_norm": 3.4962191581726074, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 4.2251, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.014822134387351778, | |
| "grad_norm": 3.627702236175537, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 4.2905, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.015151515151515152, | |
| "grad_norm": 2.822854518890381, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 4.2637, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.015480895915678524, | |
| "grad_norm": 2.7237679958343506, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 3.8639, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.015810276679841896, | |
| "grad_norm": 2.9576973915100098, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 4.3402, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.016139657444005272, | |
| "grad_norm": 8.693846702575684, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 4.3245, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.016469038208168644, | |
| "grad_norm": 14.129925727844238, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 4.8286, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.016469038208168644, | |
| "eval_loss": 4.443373203277588, | |
| "eval_runtime": 367.2539, | |
| "eval_samples_per_second": 13.922, | |
| "eval_steps_per_second": 3.483, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.016798418972332016, | |
| "grad_norm": 2.6934754848480225, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 2.4826, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.017127799736495388, | |
| "grad_norm": 3.4384169578552246, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 2.7491, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.01745718050065876, | |
| "grad_norm": 2.0442357063293457, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 2.7783, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.017786561264822136, | |
| "grad_norm": 1.0162229537963867, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 2.5059, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.018115942028985508, | |
| "grad_norm": 0.7227568626403809, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 2.7383, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.01844532279314888, | |
| "grad_norm": 0.6420945525169373, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 2.7785, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.018774703557312252, | |
| "grad_norm": 0.6553905010223389, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 3.0646, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.019104084321475624, | |
| "grad_norm": 0.7528241872787476, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 3.1277, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.019433465085639, | |
| "grad_norm": 0.6690371036529541, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 2.8938, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.019762845849802372, | |
| "grad_norm": 0.7149075865745544, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 3.1118, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.020092226613965744, | |
| "grad_norm": 0.7195890545845032, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 2.9614, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.020421607378129116, | |
| "grad_norm": 0.745587944984436, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 3.0415, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.020750988142292492, | |
| "grad_norm": 0.8089593052864075, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 3.1343, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.021080368906455864, | |
| "grad_norm": 0.8166781067848206, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 3.2951, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.021409749670619236, | |
| "grad_norm": 0.8866109251976013, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 3.4326, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.021739130434782608, | |
| "grad_norm": 0.8653653860092163, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 3.3296, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.02206851119894598, | |
| "grad_norm": 0.9505738019943237, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 3.4388, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.022397891963109356, | |
| "grad_norm": 1.1444917917251587, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 3.6358, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.022727272727272728, | |
| "grad_norm": 1.0584211349487305, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 3.3289, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0230566534914361, | |
| "grad_norm": 0.9308885335922241, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 3.3747, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.023386034255599472, | |
| "grad_norm": 1.0567805767059326, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 3.5003, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.023715415019762844, | |
| "grad_norm": 1.118471384048462, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 3.4619, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.02404479578392622, | |
| "grad_norm": 1.0994294881820679, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 3.436, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.024374176548089592, | |
| "grad_norm": 1.1488723754882812, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 3.5807, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.024703557312252964, | |
| "grad_norm": 1.1153106689453125, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 3.4539, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.025032938076416336, | |
| "grad_norm": 1.1607385873794556, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 3.5449, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.025362318840579712, | |
| "grad_norm": 1.296028733253479, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 3.5412, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.025691699604743084, | |
| "grad_norm": 1.382262945175171, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 3.5824, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.026021080368906456, | |
| "grad_norm": 1.3983418941497803, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 3.3038, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.026350461133069828, | |
| "grad_norm": 1.353999137878418, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 3.6433, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0266798418972332, | |
| "grad_norm": 1.3110865354537964, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 3.4723, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.027009222661396576, | |
| "grad_norm": 1.7040696144104004, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 3.9391, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.027338603425559948, | |
| "grad_norm": 1.679714560508728, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 3.8998, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.02766798418972332, | |
| "grad_norm": 1.737741470336914, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 3.7029, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.027997364953886692, | |
| "grad_norm": 1.4635043144226074, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 3.5636, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.028326745718050064, | |
| "grad_norm": 1.8945380449295044, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 3.8428, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.02865612648221344, | |
| "grad_norm": 1.7399659156799316, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 3.5383, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.028985507246376812, | |
| "grad_norm": 1.6550588607788086, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 3.8385, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.029314888010540184, | |
| "grad_norm": 1.8105171918869019, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 3.8848, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.029644268774703556, | |
| "grad_norm": 2.003801107406616, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 4.0658, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.029973649538866932, | |
| "grad_norm": 2.603792428970337, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 3.4235, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.030303030303030304, | |
| "grad_norm": 2.0880866050720215, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 3.7713, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.030632411067193676, | |
| "grad_norm": 2.0831027030944824, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 3.9281, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.030961791831357048, | |
| "grad_norm": 2.485839605331421, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 3.7701, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.031291172595520424, | |
| "grad_norm": 2.451188564300537, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 4.2388, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.03162055335968379, | |
| "grad_norm": 2.601740837097168, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 3.8389, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.03194993412384717, | |
| "grad_norm": 3.030194044113159, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 4.1005, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.032279314888010543, | |
| "grad_norm": 3.9098434448242188, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 3.976, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.03260869565217391, | |
| "grad_norm": 3.508920431137085, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 4.0548, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.03293807641633729, | |
| "grad_norm": 6.9816179275512695, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 4.89, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03293807641633729, | |
| "eval_loss": 4.353458881378174, | |
| "eval_runtime": 366.5166, | |
| "eval_samples_per_second": 13.95, | |
| "eval_steps_per_second": 3.49, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.033267457180500656, | |
| "grad_norm": 1.9534411430358887, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 2.5377, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.03359683794466403, | |
| "grad_norm": 3.9449827671051025, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 3.0191, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.03392621870882741, | |
| "grad_norm": 3.4320390224456787, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 2.6032, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.034255599472990776, | |
| "grad_norm": 2.760798692703247, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 2.862, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.03458498023715415, | |
| "grad_norm": 1.8093721866607666, | |
| "learning_rate": 5e-05, | |
| "loss": 2.6868, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.03491436100131752, | |
| "grad_norm": 1.3576546907424927, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 2.9458, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.035243741765480896, | |
| "grad_norm": 0.7329630255699158, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 2.8029, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.03557312252964427, | |
| "grad_norm": 0.8393235802650452, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 2.8362, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.03590250329380764, | |
| "grad_norm": 0.7931125164031982, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 3.19, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.036231884057971016, | |
| "grad_norm": 0.727262020111084, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 3.0535, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.036561264822134384, | |
| "grad_norm": 0.7600178122520447, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 3.0458, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.03689064558629776, | |
| "grad_norm": 0.7988314032554626, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 3.0866, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.037220026350461136, | |
| "grad_norm": 0.8111890554428101, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 3.3127, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.037549407114624504, | |
| "grad_norm": 0.9296747446060181, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 3.538, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.03787878787878788, | |
| "grad_norm": 0.9645518064498901, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 3.5188, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.03820816864295125, | |
| "grad_norm": 0.877086877822876, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 3.1687, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.038537549407114624, | |
| "grad_norm": 0.9801080822944641, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 3.5552, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.038866930171278, | |
| "grad_norm": 0.9617884159088135, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 3.3693, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.03919631093544137, | |
| "grad_norm": 0.9909358620643616, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 3.3733, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.039525691699604744, | |
| "grad_norm": 0.9788588285446167, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 3.3048, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.03985507246376811, | |
| "grad_norm": 0.9471210837364197, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 3.4333, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.04018445322793149, | |
| "grad_norm": 1.0797932147979736, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 3.5387, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.040513833992094864, | |
| "grad_norm": 1.0987634658813477, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 3.6053, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.04084321475625823, | |
| "grad_norm": 1.1885888576507568, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 3.3075, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.04117259552042161, | |
| "grad_norm": 1.2186295986175537, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 3.8534, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.041501976284584984, | |
| "grad_norm": 1.1846344470977783, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 3.5999, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.04183135704874835, | |
| "grad_norm": 1.1363506317138672, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 3.4929, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.04216073781291173, | |
| "grad_norm": 1.2770732641220093, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 3.5519, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.042490118577075096, | |
| "grad_norm": 1.1716625690460205, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 3.4708, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.04281949934123847, | |
| "grad_norm": 1.38254976272583, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 3.806, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.04314888010540185, | |
| "grad_norm": 1.3303941488265991, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 3.7209, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.043478260869565216, | |
| "grad_norm": 1.4240809679031372, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 3.7252, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.04380764163372859, | |
| "grad_norm": 1.4428542852401733, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 3.4899, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.04413702239789196, | |
| "grad_norm": 1.5575388669967651, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 3.8213, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.044466403162055336, | |
| "grad_norm": 1.4392671585083008, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 3.8798, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.04479578392621871, | |
| "grad_norm": 1.4760370254516602, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 3.762, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.04512516469038208, | |
| "grad_norm": 1.5625317096710205, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 3.8826, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.045454545454545456, | |
| "grad_norm": 1.865438461303711, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 3.8565, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.045783926218708824, | |
| "grad_norm": 1.783351182937622, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 4.1215, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.0461133069828722, | |
| "grad_norm": 2.019209861755371, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 4.3478, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.046442687747035576, | |
| "grad_norm": 1.9931212663650513, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 3.7415, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.046772068511198944, | |
| "grad_norm": 1.941890835762024, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 4.1205, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.04710144927536232, | |
| "grad_norm": 2.1537375450134277, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 3.9631, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.04743083003952569, | |
| "grad_norm": 2.0665690898895264, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 4.0958, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.047760210803689064, | |
| "grad_norm": 2.0371243953704834, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 3.8377, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.04808959156785244, | |
| "grad_norm": 2.9738543033599854, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 4.5223, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.04841897233201581, | |
| "grad_norm": 2.5726799964904785, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 4.5206, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.048748353096179184, | |
| "grad_norm": 2.7862298488616943, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 4.0664, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.04907773386034255, | |
| "grad_norm": 2.910820722579956, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 4.327, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.04940711462450593, | |
| "grad_norm": 5.964183330535889, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 4.7147, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04940711462450593, | |
| "eval_loss": 3.6443803310394287, | |
| "eval_runtime": 367.3681, | |
| "eval_samples_per_second": 13.918, | |
| "eval_steps_per_second": 3.482, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.049736495388669304, | |
| "grad_norm": 0.5568860769271851, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 2.476, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.05006587615283267, | |
| "grad_norm": 1.2807371616363525, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 2.5842, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.05039525691699605, | |
| "grad_norm": 1.090383529663086, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 2.5713, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.050724637681159424, | |
| "grad_norm": 1.2068983316421509, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 2.4996, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.05105401844532279, | |
| "grad_norm": 1.5854796171188354, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 2.8089, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.05138339920948617, | |
| "grad_norm": 1.6575156450271606, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 3.0389, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.051712779973649536, | |
| "grad_norm": 1.6128365993499756, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 3.1951, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.05204216073781291, | |
| "grad_norm": 1.3779178857803345, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 3.2141, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.05237154150197629, | |
| "grad_norm": 1.424267053604126, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 3.1791, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.052700922266139656, | |
| "grad_norm": 1.1836161613464355, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 3.0598, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.05303030303030303, | |
| "grad_norm": 1.2339744567871094, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 3.3081, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.0533596837944664, | |
| "grad_norm": 1.1193962097167969, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 3.2128, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.053689064558629776, | |
| "grad_norm": 0.9499046802520752, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 3.3217, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.05401844532279315, | |
| "grad_norm": 0.8986177444458008, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 3.4255, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.05434782608695652, | |
| "grad_norm": 0.9408730268478394, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 3.5404, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.054677206851119896, | |
| "grad_norm": 0.9029567241668701, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 3.2013, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.055006587615283264, | |
| "grad_norm": 1.044360637664795, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 3.4893, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.05533596837944664, | |
| "grad_norm": 0.9520983099937439, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 3.3199, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.055665349143610016, | |
| "grad_norm": 0.9456550478935242, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 3.2871, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.055994729907773384, | |
| "grad_norm": 1.1052970886230469, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 3.4073, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.05632411067193676, | |
| "grad_norm": 1.234739065170288, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 3.6327, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.05665349143610013, | |
| "grad_norm": 1.2236120700836182, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 3.7213, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.056982872200263504, | |
| "grad_norm": 1.14174485206604, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 3.4071, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.05731225296442688, | |
| "grad_norm": 1.2664788961410522, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 3.7652, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.05764163372859025, | |
| "grad_norm": 1.3992626667022705, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 3.7304, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.057971014492753624, | |
| "grad_norm": 1.489917278289795, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 3.6773, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.058300395256917, | |
| "grad_norm": 1.4404100179672241, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 3.7922, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.05862977602108037, | |
| "grad_norm": 1.3944207429885864, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 3.7745, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.058959156785243744, | |
| "grad_norm": 1.5016889572143555, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 3.8608, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05928853754940711, | |
| "grad_norm": 1.6601581573486328, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 3.979, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.05961791831357049, | |
| "grad_norm": 1.7638131380081177, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 3.6176, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.059947299077733864, | |
| "grad_norm": 1.8236701488494873, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 4.0674, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.06027667984189723, | |
| "grad_norm": 1.713302731513977, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 3.7418, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.06060606060606061, | |
| "grad_norm": 1.7112972736358643, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 3.845, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.060935441370223976, | |
| "grad_norm": 1.9323976039886475, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 3.728, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.06126482213438735, | |
| "grad_norm": 1.8406111001968384, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 3.8027, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.06159420289855073, | |
| "grad_norm": 2.0645482540130615, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 3.8884, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.061923583662714096, | |
| "grad_norm": 2.4051105976104736, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 4.343, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.06225296442687747, | |
| "grad_norm": 2.376185655593872, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 3.9507, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.06258234519104085, | |
| "grad_norm": 2.4887542724609375, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 4.1476, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.06291172595520422, | |
| "grad_norm": 2.6241660118103027, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 4.4219, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.06324110671936758, | |
| "grad_norm": 2.478088140487671, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 3.7916, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.06357048748353096, | |
| "grad_norm": 2.9313812255859375, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 4.1702, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.06389986824769434, | |
| "grad_norm": 3.6325390338897705, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 4.3693, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.06422924901185771, | |
| "grad_norm": 3.3586654663085938, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 4.4021, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.06455862977602109, | |
| "grad_norm": 3.4749622344970703, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 3.9223, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.06488801054018445, | |
| "grad_norm": 4.705071449279785, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 4.5717, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.06521739130434782, | |
| "grad_norm": 4.904595851898193, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 4.7226, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.0655467720685112, | |
| "grad_norm": 5.40534782409668, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 5.0899, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.06587615283267458, | |
| "grad_norm": 7.301121711730957, | |
| "learning_rate": 0.0, | |
| "loss": 5.02, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.06587615283267458, | |
| "eval_loss": 3.5797412395477295, | |
| "eval_runtime": 366.6515, | |
| "eval_samples_per_second": 13.945, | |
| "eval_steps_per_second": 3.488, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "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": 2.8722151443922944e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |