| { | |
| "best_metric": 9.30453315959312e-05, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-150", | |
| "epoch": 2.9850746268656714, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.014925373134328358, | |
| "grad_norm": 106.12474060058594, | |
| "learning_rate": 1e-05, | |
| "loss": 2.7804, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.014925373134328358, | |
| "eval_loss": 1.059448480606079, | |
| "eval_runtime": 10.5654, | |
| "eval_samples_per_second": 10.695, | |
| "eval_steps_per_second": 2.745, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.029850746268656716, | |
| "grad_norm": 122.70915985107422, | |
| "learning_rate": 2e-05, | |
| "loss": 3.1292, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.04477611940298507, | |
| "grad_norm": 73.10962677001953, | |
| "learning_rate": 3e-05, | |
| "loss": 1.9939, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.05970149253731343, | |
| "grad_norm": 47.595909118652344, | |
| "learning_rate": 4e-05, | |
| "loss": 1.2567, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.07462686567164178, | |
| "grad_norm": 20.38334083557129, | |
| "learning_rate": 5e-05, | |
| "loss": 0.3167, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.08955223880597014, | |
| "grad_norm": 2.4722652435302734, | |
| "learning_rate": 6e-05, | |
| "loss": 0.0357, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.1044776119402985, | |
| "grad_norm": 0.6679202914237976, | |
| "learning_rate": 7e-05, | |
| "loss": 0.0054, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.11940298507462686, | |
| "grad_norm": 0.11702053248882294, | |
| "learning_rate": 8e-05, | |
| "loss": 0.0003, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.13432835820895522, | |
| "grad_norm": 11.30651569366455, | |
| "learning_rate": 9e-05, | |
| "loss": 0.0162, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.14925373134328357, | |
| "grad_norm": 0.13928619027137756, | |
| "learning_rate": 0.0001, | |
| "loss": 0.0004, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.16417910447761194, | |
| "grad_norm": 4.454522609710693, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.0328, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.1791044776119403, | |
| "grad_norm": 0.007104151416569948, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.0001, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.19402985074626866, | |
| "grad_norm": 0.06303039938211441, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.0002, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.208955223880597, | |
| "grad_norm": 1.3901987075805664, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.0044, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.22388059701492538, | |
| "grad_norm": 0.004481032025068998, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.0001, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.23880597014925373, | |
| "grad_norm": 0.13158084452152252, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.0002, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.2537313432835821, | |
| "grad_norm": 3.2775821685791016, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.027, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.26865671641791045, | |
| "grad_norm": 1575.59228515625, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 12.1851, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.2835820895522388, | |
| "grad_norm": 13.348243713378906, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.0201, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.29850746268656714, | |
| "grad_norm": 13.7633056640625, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.0506, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.31343283582089554, | |
| "grad_norm": 16.16105079650879, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.0836, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.3283582089552239, | |
| "grad_norm": 0.6847590804100037, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.0011, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.34328358208955223, | |
| "grad_norm": 1.2273591756820679, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.0035, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.3582089552238806, | |
| "grad_norm": 2.6755146980285645, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.0057, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.373134328358209, | |
| "grad_norm": 1.4577181339263916, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.0077, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.3880597014925373, | |
| "grad_norm": 6.496804237365723, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.0719, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.40298507462686567, | |
| "grad_norm": 3.4276888370513916, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.0452, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.417910447761194, | |
| "grad_norm": 1.0464049577713013, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.0027, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.43283582089552236, | |
| "grad_norm": 4.585959434509277, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.0244, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.44776119402985076, | |
| "grad_norm": 0.3710354268550873, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.0006, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.4626865671641791, | |
| "grad_norm": 0.129180908203125, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.0003, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.47761194029850745, | |
| "grad_norm": 0.23275743424892426, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.0003, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.4925373134328358, | |
| "grad_norm": 0.7053936123847961, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.0055, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.5074626865671642, | |
| "grad_norm": 3.2974233627319336, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.0138, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.5223880597014925, | |
| "grad_norm": 1.7037897109985352, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.0076, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.5373134328358209, | |
| "grad_norm": 7.839724540710449, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.011, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.5522388059701493, | |
| "grad_norm": 0.11264707893133163, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.0004, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.5671641791044776, | |
| "grad_norm": 1.493961215019226, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.0036, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.582089552238806, | |
| "grad_norm": 0.5191388130187988, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.0024, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.5970149253731343, | |
| "grad_norm": 0.8491466641426086, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.0039, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.6119402985074627, | |
| "grad_norm": 10.151022911071777, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.019, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.6268656716417911, | |
| "grad_norm": 9.693743705749512, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.1235, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.6417910447761194, | |
| "grad_norm": 100.07304382324219, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.2999, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.6567164179104478, | |
| "grad_norm": 15.32607364654541, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.1382, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.6716417910447762, | |
| "grad_norm": 2.9328484535217285, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.0732, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.6865671641791045, | |
| "grad_norm": 3.014009714126587, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.0079, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.7014925373134329, | |
| "grad_norm": 92.01437377929688, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 2.3133, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.7164179104477612, | |
| "grad_norm": 0.48266130685806274, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.004, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.7313432835820896, | |
| "grad_norm": 2.2023372650146484, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.0204, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.746268656716418, | |
| "grad_norm": 0.6010475158691406, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.0079, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.746268656716418, | |
| "eval_loss": 0.005077690351754427, | |
| "eval_runtime": 10.7656, | |
| "eval_samples_per_second": 10.496, | |
| "eval_steps_per_second": 2.694, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.7611940298507462, | |
| "grad_norm": 1.3045402765274048, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.0125, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.7761194029850746, | |
| "grad_norm": 4.1738715171813965, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.0448, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.7910447761194029, | |
| "grad_norm": 2.5954270362854004, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.0335, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.8059701492537313, | |
| "grad_norm": 23.488723754882812, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.0879, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.8208955223880597, | |
| "grad_norm": 17.428266525268555, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.0904, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.835820895522388, | |
| "grad_norm": 20.230976104736328, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.0459, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.8507462686567164, | |
| "grad_norm": 2.9111521244049072, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.0122, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.8656716417910447, | |
| "grad_norm": 5.954577922821045, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.0405, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.8805970149253731, | |
| "grad_norm": 110.38929748535156, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.6805, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.8955223880597015, | |
| "grad_norm": 25.989730834960938, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.1605, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.9104477611940298, | |
| "grad_norm": 10.852027893066406, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.0209, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.9253731343283582, | |
| "grad_norm": 3.224698781967163, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.0148, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.9402985074626866, | |
| "grad_norm": 1.6246620416641235, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.0104, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.9552238805970149, | |
| "grad_norm": 3.8351142406463623, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.0869, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.9701492537313433, | |
| "grad_norm": 7.816077709197998, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.1743, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.9850746268656716, | |
| "grad_norm": 1.3039127588272095, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.0151, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.9009143114089966, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.0066, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.0149253731343284, | |
| "grad_norm": 2.292771577835083, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.0597, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.0298507462686568, | |
| "grad_norm": 1.3948904275894165, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.0052, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.044776119402985, | |
| "grad_norm": 2.6687114238739014, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.0166, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.0597014925373134, | |
| "grad_norm": 5.295480728149414, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.1218, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.0746268656716418, | |
| "grad_norm": 8.071635246276855, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.0315, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.0895522388059702, | |
| "grad_norm": 3.1014323234558105, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.0052, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.1044776119402986, | |
| "grad_norm": 0.33262988924980164, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.0032, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.1194029850746268, | |
| "grad_norm": 1.3474750518798828, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.0095, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.1343283582089552, | |
| "grad_norm": 0.0673995241522789, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.0009, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.1492537313432836, | |
| "grad_norm": 0.03149837255477905, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.0004, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.164179104477612, | |
| "grad_norm": 0.018193544819951057, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.0002, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.1791044776119404, | |
| "grad_norm": 0.12023726105690002, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.0008, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.1940298507462686, | |
| "grad_norm": 1.058858871459961, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.021, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.208955223880597, | |
| "grad_norm": 0.45185062289237976, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.0017, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.2238805970149254, | |
| "grad_norm": 5.356001377105713, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.0491, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.2388059701492538, | |
| "grad_norm": 0.28147175908088684, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.0027, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.2537313432835822, | |
| "grad_norm": 0.5419133305549622, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.0044, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.2686567164179103, | |
| "grad_norm": 0.005355765577405691, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.0001, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.2835820895522387, | |
| "grad_norm": 0.009597934782505035, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.0001, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.2985074626865671, | |
| "grad_norm": 0.9960870146751404, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.0034, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.3134328358208955, | |
| "grad_norm": 0.8029950261116028, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.0017, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.328358208955224, | |
| "grad_norm": 1.726015329360962, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.0703, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.3432835820895521, | |
| "grad_norm": 0.2299676239490509, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.0008, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.3582089552238805, | |
| "grad_norm": 0.0269209872931242, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.0002, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.373134328358209, | |
| "grad_norm": 0.008052011020481586, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.0001, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.3880597014925373, | |
| "grad_norm": 0.07679415494203568, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.0006, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.4029850746268657, | |
| "grad_norm": 0.14484962821006775, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.0008, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.417910447761194, | |
| "grad_norm": 0.022719435393810272, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.0002, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.4328358208955223, | |
| "grad_norm": 0.015714868903160095, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.0003, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.4477611940298507, | |
| "grad_norm": 0.18392543494701385, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.0003, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.462686567164179, | |
| "grad_norm": 0.3576776087284088, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.0017, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.4776119402985075, | |
| "grad_norm": 0.02203579992055893, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.0002, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.4925373134328357, | |
| "grad_norm": 0.02050737477838993, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.0003, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.4925373134328357, | |
| "eval_loss": 0.0038506302516907454, | |
| "eval_runtime": 10.7703, | |
| "eval_samples_per_second": 10.492, | |
| "eval_steps_per_second": 2.693, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.5074626865671643, | |
| "grad_norm": 0.3960582911968231, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.0015, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.5223880597014925, | |
| "grad_norm": 0.20707403123378754, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.0011, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.537313432835821, | |
| "grad_norm": 0.013787700794637203, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.0002, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.5522388059701493, | |
| "grad_norm": 0.5520917773246765, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.0014, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.5671641791044775, | |
| "grad_norm": 0.009354157373309135, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0002, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.582089552238806, | |
| "grad_norm": 0.021780215203762054, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.0002, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.5970149253731343, | |
| "grad_norm": 0.007954766973853111, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.0001, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.6119402985074627, | |
| "grad_norm": 0.0062783993780612946, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.0001, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.626865671641791, | |
| "grad_norm": 0.4941234290599823, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.0045, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.6417910447761193, | |
| "grad_norm": 0.4348292350769043, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.0013, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.6567164179104479, | |
| "grad_norm": 0.022044768556952477, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.0002, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.671641791044776, | |
| "grad_norm": 0.007155040744692087, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.0001, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.6865671641791045, | |
| "grad_norm": 0.0037332745268940926, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.0001, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.7014925373134329, | |
| "grad_norm": 0.12028840184211731, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.0003, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.716417910447761, | |
| "grad_norm": 0.0013203342678025365, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.0, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.7313432835820897, | |
| "grad_norm": 4.437915802001953, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.0046, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.7462686567164178, | |
| "grad_norm": 0.045842915773391724, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.0002, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.7611940298507462, | |
| "grad_norm": 0.37353402376174927, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.0015, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.7761194029850746, | |
| "grad_norm": 0.2279956191778183, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.0004, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.7910447761194028, | |
| "grad_norm": 0.6662240624427795, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.0017, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.8059701492537314, | |
| "grad_norm": 0.006625319365411997, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.0001, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.8208955223880596, | |
| "grad_norm": 1.3940486907958984, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.033, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.835820895522388, | |
| "grad_norm": 0.006158743519335985, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.0001, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.8507462686567164, | |
| "grad_norm": 0.02983230911195278, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.0003, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.8656716417910446, | |
| "grad_norm": 0.003336232155561447, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.0001, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.8805970149253732, | |
| "grad_norm": 0.01575973629951477, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.0002, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.8955223880597014, | |
| "grad_norm": 0.03806147351861, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.0004, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.9104477611940298, | |
| "grad_norm": 0.19246968626976013, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.0008, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.9253731343283582, | |
| "grad_norm": 0.08919795602560043, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.0012, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.9402985074626866, | |
| "grad_norm": 0.02700089104473591, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.0003, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.955223880597015, | |
| "grad_norm": 0.6637530326843262, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.0104, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.9701492537313432, | |
| "grad_norm": 0.018300116062164307, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.0003, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.9850746268656716, | |
| "grad_norm": 0.14137744903564453, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.0007, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.07243705540895462, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.0006, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 2.014925373134328, | |
| "grad_norm": 0.03928264603018761, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.0003, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 2.029850746268657, | |
| "grad_norm": 0.09312094002962112, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.0008, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.044776119402985, | |
| "grad_norm": 0.005409217439591885, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.0001, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 2.0597014925373136, | |
| "grad_norm": 0.11604958772659302, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.0009, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.074626865671642, | |
| "grad_norm": 0.02191762626171112, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.0003, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 2.08955223880597, | |
| "grad_norm": 0.464058518409729, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.0104, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.1044776119402986, | |
| "grad_norm": 0.01005578227341175, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.0001, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 2.1194029850746268, | |
| "grad_norm": 0.012159494683146477, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.0002, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.1343283582089554, | |
| "grad_norm": 0.004815397784113884, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.0001, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 2.1492537313432836, | |
| "grad_norm": 0.03488089516758919, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.0005, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.1641791044776117, | |
| "grad_norm": 0.014678386971354485, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.0001, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 2.1791044776119404, | |
| "grad_norm": 0.013062779791653156, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.0002, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.1940298507462686, | |
| "grad_norm": 1.9010485410690308, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.0055, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 2.208955223880597, | |
| "grad_norm": 0.0029188445769250393, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.0001, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.2238805970149254, | |
| "grad_norm": 0.007598489988595247, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.0001, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 2.2388059701492535, | |
| "grad_norm": 0.0036820729728788137, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.0001, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.2388059701492535, | |
| "eval_loss": 9.30453315959312e-05, | |
| "eval_runtime": 10.759, | |
| "eval_samples_per_second": 10.503, | |
| "eval_steps_per_second": 2.695, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.253731343283582, | |
| "grad_norm": 0.014250511303544044, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.0002, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 2.2686567164179103, | |
| "grad_norm": 2.3947012424468994, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.0782, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 2.283582089552239, | |
| "grad_norm": 0.9007366299629211, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.0458, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 2.298507462686567, | |
| "grad_norm": 1.875738501548767, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.0086, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 2.3134328358208958, | |
| "grad_norm": 0.004122923593968153, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.0001, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 2.328358208955224, | |
| "grad_norm": 0.004269476979970932, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.0001, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 2.343283582089552, | |
| "grad_norm": 0.021002259105443954, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.0001, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 2.3582089552238807, | |
| "grad_norm": 0.003884516656398773, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.0001, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 2.373134328358209, | |
| "grad_norm": 0.012166302651166916, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.0001, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 2.388059701492537, | |
| "grad_norm": 0.0653151199221611, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.0003, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.4029850746268657, | |
| "grad_norm": 0.006849711295217276, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.0001, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 2.417910447761194, | |
| "grad_norm": 0.04303903505206108, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.0005, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 2.4328358208955225, | |
| "grad_norm": 0.006036329548805952, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.0001, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 2.4477611940298507, | |
| "grad_norm": 0.057249877601861954, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.0002, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 2.4626865671641793, | |
| "grad_norm": 0.00814924854785204, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.0001, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 2.4776119402985075, | |
| "grad_norm": 0.009180199354887009, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.0002, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 2.4925373134328357, | |
| "grad_norm": 0.051900748163461685, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.0008, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 2.5074626865671643, | |
| "grad_norm": 0.007953710854053497, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.0001, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 2.5223880597014925, | |
| "grad_norm": 0.0539824403822422, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.0003, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 2.5373134328358207, | |
| "grad_norm": 0.24127671122550964, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.0014, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.5522388059701493, | |
| "grad_norm": 0.06358372420072556, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.0006, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 2.5671641791044775, | |
| "grad_norm": 0.03995596617460251, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.0004, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 2.582089552238806, | |
| "grad_norm": 0.9266185760498047, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.0023, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 2.5970149253731343, | |
| "grad_norm": 0.005565286614000797, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.0001, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 2.611940298507463, | |
| "grad_norm": 0.007816660217940807, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.0001, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.626865671641791, | |
| "grad_norm": 0.02459590695798397, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.0002, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 2.6417910447761193, | |
| "grad_norm": 0.008609198965132236, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.0001, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 2.656716417910448, | |
| "grad_norm": 0.008006337098777294, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.0001, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 2.671641791044776, | |
| "grad_norm": 0.005286152940243483, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.0001, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 2.6865671641791042, | |
| "grad_norm": 0.029769370332360268, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.0002, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.701492537313433, | |
| "grad_norm": 0.6887739896774292, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.001, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 2.716417910447761, | |
| "grad_norm": 0.0050988467410206795, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.0001, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 2.7313432835820897, | |
| "grad_norm": 0.4306039810180664, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.0013, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 2.746268656716418, | |
| "grad_norm": 0.004355176351964474, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.0001, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 2.7611940298507465, | |
| "grad_norm": 0.00952870212495327, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.0001, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 2.7761194029850746, | |
| "grad_norm": 0.011709846556186676, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.0002, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 2.791044776119403, | |
| "grad_norm": 0.01787348836660385, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.0002, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.8059701492537314, | |
| "grad_norm": 0.002467110985890031, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.0001, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 2.8208955223880596, | |
| "grad_norm": 0.382357120513916, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.0008, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 2.835820895522388, | |
| "grad_norm": 0.0831623300909996, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.0005, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.8507462686567164, | |
| "grad_norm": 0.052397552877664566, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.0004, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 2.8656716417910446, | |
| "grad_norm": 0.010271397419273853, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.0001, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 2.8805970149253732, | |
| "grad_norm": 0.005534847266972065, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.0001, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 2.8955223880597014, | |
| "grad_norm": 0.0054015819914639, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.0001, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 2.91044776119403, | |
| "grad_norm": 0.010067780502140522, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.0002, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 2.925373134328358, | |
| "grad_norm": 0.01627379283308983, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.0002, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 2.9402985074626864, | |
| "grad_norm": 0.005502472631633282, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.0001, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 2.955223880597015, | |
| "grad_norm": 0.013814440928399563, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.0002, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 2.970149253731343, | |
| "grad_norm": 0.012036381289362907, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.0002, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 2.9850746268656714, | |
| "grad_norm": 0.12174969911575317, | |
| "learning_rate": 0.0, | |
| "loss": 0.0018, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.9850746268656714, | |
| "eval_loss": 0.00014888614532537758, | |
| "eval_runtime": 10.7497, | |
| "eval_samples_per_second": 10.512, | |
| "eval_steps_per_second": 2.698, | |
| "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": 1 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.8691510080202342e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |