| { | |
| "best_metric": 0.9390347003936768, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-100", | |
| "epoch": 1.0049751243781095, | |
| "eval_steps": 100, | |
| "global_step": 101, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.009950248756218905, | |
| "grad_norm": 2.370561361312866, | |
| "learning_rate": 5e-06, | |
| "loss": 1.0474, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.009950248756218905, | |
| "eval_loss": 1.0728834867477417, | |
| "eval_runtime": 7.9605, | |
| "eval_samples_per_second": 21.355, | |
| "eval_steps_per_second": 5.402, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01990049751243781, | |
| "grad_norm": 2.4022700786590576, | |
| "learning_rate": 1e-05, | |
| "loss": 1.0829, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.029850746268656716, | |
| "grad_norm": 3.453360080718994, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.0655, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.03980099502487562, | |
| "grad_norm": 2.4925131797790527, | |
| "learning_rate": 2e-05, | |
| "loss": 1.0848, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.04975124378109453, | |
| "grad_norm": 2.5615079402923584, | |
| "learning_rate": 2.5e-05, | |
| "loss": 1.0388, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.05970149253731343, | |
| "grad_norm": 2.178732395172119, | |
| "learning_rate": 3e-05, | |
| "loss": 1.0828, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.06965174129353234, | |
| "grad_norm": 2.5980308055877686, | |
| "learning_rate": 3.5e-05, | |
| "loss": 1.075, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.07960199004975124, | |
| "grad_norm": 2.408590078353882, | |
| "learning_rate": 4e-05, | |
| "loss": 1.076, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.08955223880597014, | |
| "grad_norm": 2.9705393314361572, | |
| "learning_rate": 4.5e-05, | |
| "loss": 1.0433, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.09950248756218906, | |
| "grad_norm": 2.3930296897888184, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0522, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.10945273631840796, | |
| "grad_norm": 2.078711748123169, | |
| "learning_rate": 5.500000000000001e-05, | |
| "loss": 1.0705, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.11940298507462686, | |
| "grad_norm": 2.475327253341675, | |
| "learning_rate": 6e-05, | |
| "loss": 1.0754, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.12935323383084577, | |
| "grad_norm": 2.961998701095581, | |
| "learning_rate": 6.500000000000001e-05, | |
| "loss": 1.0453, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.13930348258706468, | |
| "grad_norm": 4.242774963378906, | |
| "learning_rate": 7e-05, | |
| "loss": 1.0418, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.14925373134328357, | |
| "grad_norm": 2.8345038890838623, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 1.0769, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.15920398009950248, | |
| "grad_norm": 2.97365403175354, | |
| "learning_rate": 8e-05, | |
| "loss": 1.0672, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.1691542288557214, | |
| "grad_norm": 3.3158862590789795, | |
| "learning_rate": 8.5e-05, | |
| "loss": 1.0577, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.1791044776119403, | |
| "grad_norm": 3.4335522651672363, | |
| "learning_rate": 9e-05, | |
| "loss": 1.0795, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.1890547263681592, | |
| "grad_norm": 2.685619592666626, | |
| "learning_rate": 9.5e-05, | |
| "loss": 1.0594, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.19900497512437812, | |
| "grad_norm": 2.6879546642303467, | |
| "learning_rate": 0.0001, | |
| "loss": 1.102, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.208955223880597, | |
| "grad_norm": 3.9734950065612793, | |
| "learning_rate": 9.996239762521151e-05, | |
| "loss": 1.0358, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.21890547263681592, | |
| "grad_norm": 3.007880687713623, | |
| "learning_rate": 9.98496470583896e-05, | |
| "loss": 1.0693, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.22885572139303484, | |
| "grad_norm": 4.0449042320251465, | |
| "learning_rate": 9.966191788709716e-05, | |
| "loss": 1.0339, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.23880597014925373, | |
| "grad_norm": 2.7142257690429688, | |
| "learning_rate": 9.939949247384046e-05, | |
| "loss": 1.0377, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.24875621890547264, | |
| "grad_norm": 2.980034589767456, | |
| "learning_rate": 9.906276553136923e-05, | |
| "loss": 1.0465, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.25870646766169153, | |
| "grad_norm": 3.321119546890259, | |
| "learning_rate": 9.865224352899119e-05, | |
| "loss": 0.9891, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.26865671641791045, | |
| "grad_norm": 2.879805564880371, | |
| "learning_rate": 9.816854393079403e-05, | |
| "loss": 1.0101, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.27860696517412936, | |
| "grad_norm": 2.740520715713501, | |
| "learning_rate": 9.761239426692077e-05, | |
| "loss": 1.0325, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.2885572139303483, | |
| "grad_norm": 2.686652898788452, | |
| "learning_rate": 9.698463103929542e-05, | |
| "loss": 0.9961, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.29850746268656714, | |
| "grad_norm": 2.9260804653167725, | |
| "learning_rate": 9.628619846344454e-05, | |
| "loss": 0.9861, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.30845771144278605, | |
| "grad_norm": 3.1635563373565674, | |
| "learning_rate": 9.551814704830734e-05, | |
| "loss": 1.0263, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.31840796019900497, | |
| "grad_norm": 3.2495548725128174, | |
| "learning_rate": 9.468163201617062e-05, | |
| "loss": 0.9937, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.3283582089552239, | |
| "grad_norm": 2.594937801361084, | |
| "learning_rate": 9.377791156510455e-05, | |
| "loss": 1.0044, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.3383084577114428, | |
| "grad_norm": 2.953624963760376, | |
| "learning_rate": 9.280834497651334e-05, | |
| "loss": 1.0048, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.3482587064676617, | |
| "grad_norm": 2.600581169128418, | |
| "learning_rate": 9.177439057064683e-05, | |
| "loss": 0.984, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.3582089552238806, | |
| "grad_norm": 2.8330259323120117, | |
| "learning_rate": 9.067760351314838e-05, | |
| "loss": 1.0075, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.3681592039800995, | |
| "grad_norm": 2.660095691680908, | |
| "learning_rate": 8.951963347593797e-05, | |
| "loss": 1.0064, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.3781094527363184, | |
| "grad_norm": 2.9839465618133545, | |
| "learning_rate": 8.83022221559489e-05, | |
| "loss": 0.9916, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.3880597014925373, | |
| "grad_norm": 2.646791934967041, | |
| "learning_rate": 8.702720065545024e-05, | |
| "loss": 0.991, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.39800995024875624, | |
| "grad_norm": 2.615619659423828, | |
| "learning_rate": 8.569648672789497e-05, | |
| "loss": 1.0294, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.4079601990049751, | |
| "grad_norm": 2.4245409965515137, | |
| "learning_rate": 8.43120818934367e-05, | |
| "loss": 0.981, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.417910447761194, | |
| "grad_norm": 2.807609796524048, | |
| "learning_rate": 8.28760684284532e-05, | |
| "loss": 0.9982, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.42786069651741293, | |
| "grad_norm": 2.5878968238830566, | |
| "learning_rate": 8.139060623360493e-05, | |
| "loss": 0.994, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.43781094527363185, | |
| "grad_norm": 2.3299400806427, | |
| "learning_rate": 7.985792958513931e-05, | |
| "loss": 1.0052, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.44776119402985076, | |
| "grad_norm": 2.4846181869506836, | |
| "learning_rate": 7.828034377432693e-05, | |
| "loss": 0.9972, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.4577114427860697, | |
| "grad_norm": 2.498612642288208, | |
| "learning_rate": 7.666022164008457e-05, | |
| "loss": 1.0202, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.46766169154228854, | |
| "grad_norm": 2.713153123855591, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.9935, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.47761194029850745, | |
| "grad_norm": 2.8727245330810547, | |
| "learning_rate": 7.330217598512695e-05, | |
| "loss": 0.9894, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.48756218905472637, | |
| "grad_norm": 3.1435141563415527, | |
| "learning_rate": 7.156930328406268e-05, | |
| "loss": 0.9975, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.4975124378109453, | |
| "grad_norm": 2.586751699447632, | |
| "learning_rate": 6.980398830195785e-05, | |
| "loss": 1.0167, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.5074626865671642, | |
| "grad_norm": 2.858091354370117, | |
| "learning_rate": 6.800888624023553e-05, | |
| "loss": 0.9704, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.5174129353233831, | |
| "grad_norm": 2.097409963607788, | |
| "learning_rate": 6.618669710291606e-05, | |
| "loss": 1.0067, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.527363184079602, | |
| "grad_norm": 2.168245315551758, | |
| "learning_rate": 6.434016163555452e-05, | |
| "loss": 0.9568, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.5373134328358209, | |
| "grad_norm": 2.109443187713623, | |
| "learning_rate": 6.247205720289907e-05, | |
| "loss": 0.9757, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.5472636815920398, | |
| "grad_norm": 2.3406951427459717, | |
| "learning_rate": 6.058519361147055e-05, | |
| "loss": 0.9724, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.5572139303482587, | |
| "grad_norm": 2.2367491722106934, | |
| "learning_rate": 5.868240888334653e-05, | |
| "loss": 0.9787, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.5671641791044776, | |
| "grad_norm": 2.1488230228424072, | |
| "learning_rate": 5.6766564987506566e-05, | |
| "loss": 0.9817, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.5771144278606966, | |
| "grad_norm": 2.1089084148406982, | |
| "learning_rate": 5.484054353515896e-05, | |
| "loss": 0.9883, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.5870646766169154, | |
| "grad_norm": 2.262054681777954, | |
| "learning_rate": 5.290724144552379e-05, | |
| "loss": 0.9448, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.5970149253731343, | |
| "grad_norm": 2.093430519104004, | |
| "learning_rate": 5.096956658859122e-05, | |
| "loss": 0.9696, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.6069651741293532, | |
| "grad_norm": 3.0342745780944824, | |
| "learning_rate": 4.903043341140879e-05, | |
| "loss": 0.9654, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.6169154228855721, | |
| "grad_norm": 3.3050591945648193, | |
| "learning_rate": 4.709275855447621e-05, | |
| "loss": 1.0399, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.6268656716417911, | |
| "grad_norm": 2.6303319931030273, | |
| "learning_rate": 4.515945646484105e-05, | |
| "loss": 0.9839, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.6368159203980099, | |
| "grad_norm": 2.3729300498962402, | |
| "learning_rate": 4.323343501249346e-05, | |
| "loss": 0.9815, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.6467661691542289, | |
| "grad_norm": 2.4030983448028564, | |
| "learning_rate": 4.131759111665349e-05, | |
| "loss": 0.9548, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.6567164179104478, | |
| "grad_norm": 2.4516849517822266, | |
| "learning_rate": 3.941480638852948e-05, | |
| "loss": 0.9653, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 2.461177110671997, | |
| "learning_rate": 3.752794279710094e-05, | |
| "loss": 1.0094, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.6766169154228856, | |
| "grad_norm": 2.4054040908813477, | |
| "learning_rate": 3.5659838364445505e-05, | |
| "loss": 0.9782, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.6865671641791045, | |
| "grad_norm": 2.4171035289764404, | |
| "learning_rate": 3.381330289708396e-05, | |
| "loss": 0.9738, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.6965174129353234, | |
| "grad_norm": 2.410240650177002, | |
| "learning_rate": 3.199111375976449e-05, | |
| "loss": 0.9673, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.7064676616915423, | |
| "grad_norm": 2.5898170471191406, | |
| "learning_rate": 3.019601169804216e-05, | |
| "loss": 1.0031, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.7164179104477612, | |
| "grad_norm": 2.106549024581909, | |
| "learning_rate": 2.8430696715937337e-05, | |
| "loss": 0.9615, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.7263681592039801, | |
| "grad_norm": 3.1593167781829834, | |
| "learning_rate": 2.6697824014873075e-05, | |
| "loss": 0.9408, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.736318407960199, | |
| "grad_norm": 2.201892852783203, | |
| "learning_rate": 2.500000000000001e-05, | |
| "loss": 1.0046, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.746268656716418, | |
| "grad_norm": 2.6017942428588867, | |
| "learning_rate": 2.333977835991545e-05, | |
| "loss": 1.0041, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.7562189054726368, | |
| "grad_norm": 2.420989513397217, | |
| "learning_rate": 2.171965622567308e-05, | |
| "loss": 0.956, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.7661691542288557, | |
| "grad_norm": 2.340715169906616, | |
| "learning_rate": 2.0142070414860704e-05, | |
| "loss": 0.9744, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.7761194029850746, | |
| "grad_norm": 2.212714910507202, | |
| "learning_rate": 1.8609393766395085e-05, | |
| "loss": 0.9479, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.7860696517412935, | |
| "grad_norm": 2.3441693782806396, | |
| "learning_rate": 1.7123931571546827e-05, | |
| "loss": 0.9634, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.7960199004975125, | |
| "grad_norm": 2.3936691284179688, | |
| "learning_rate": 1.5687918106563326e-05, | |
| "loss": 0.9946, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.8059701492537313, | |
| "grad_norm": 2.836846351623535, | |
| "learning_rate": 1.4303513272105057e-05, | |
| "loss": 0.9847, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.8159203980099502, | |
| "grad_norm": 2.2729079723358154, | |
| "learning_rate": 1.297279934454978e-05, | |
| "loss": 0.95, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.8258706467661692, | |
| "grad_norm": 3.8941750526428223, | |
| "learning_rate": 1.1697777844051105e-05, | |
| "loss": 0.9585, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.835820895522388, | |
| "grad_norm": 3.08699631690979, | |
| "learning_rate": 1.0480366524062042e-05, | |
| "loss": 0.9774, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.845771144278607, | |
| "grad_norm": 2.5296778678894043, | |
| "learning_rate": 9.322396486851626e-06, | |
| "loss": 0.9695, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.8557213930348259, | |
| "grad_norm": 2.3693594932556152, | |
| "learning_rate": 8.225609429353187e-06, | |
| "loss": 0.9583, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.8656716417910447, | |
| "grad_norm": 2.286224126815796, | |
| "learning_rate": 7.191655023486682e-06, | |
| "loss": 1.0009, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.8756218905472637, | |
| "grad_norm": 2.6686084270477295, | |
| "learning_rate": 6.222088434895462e-06, | |
| "loss": 0.9626, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.8855721393034826, | |
| "grad_norm": 2.4388320446014404, | |
| "learning_rate": 5.318367983829392e-06, | |
| "loss": 0.9736, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.8955223880597015, | |
| "grad_norm": 2.7601969242095947, | |
| "learning_rate": 4.4818529516926726e-06, | |
| "loss": 0.977, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.9054726368159204, | |
| "grad_norm": 2.3179821968078613, | |
| "learning_rate": 3.7138015365554833e-06, | |
| "loss": 0.964, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.9154228855721394, | |
| "grad_norm": 2.7723841667175293, | |
| "learning_rate": 3.0153689607045845e-06, | |
| "loss": 0.9642, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.9253731343283582, | |
| "grad_norm": 2.443444013595581, | |
| "learning_rate": 2.3876057330792346e-06, | |
| "loss": 0.9535, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.9353233830845771, | |
| "grad_norm": 2.286829948425293, | |
| "learning_rate": 1.8314560692059835e-06, | |
| "loss": 0.9691, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.945273631840796, | |
| "grad_norm": 3.324568271636963, | |
| "learning_rate": 1.3477564710088098e-06, | |
| "loss": 0.9715, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.9552238805970149, | |
| "grad_norm": 2.308687210083008, | |
| "learning_rate": 9.372344686307655e-07, | |
| "loss": 0.9482, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.9651741293532339, | |
| "grad_norm": 2.547358274459839, | |
| "learning_rate": 6.005075261595494e-07, | |
| "loss": 0.9975, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.9751243781094527, | |
| "grad_norm": 2.5918407440185547, | |
| "learning_rate": 3.380821129028489e-07, | |
| "loss": 1.0035, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.9850746268656716, | |
| "grad_norm": 2.4646270275115967, | |
| "learning_rate": 1.503529416103988e-07, | |
| "loss": 0.9942, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.9950248756218906, | |
| "grad_norm": 2.581437826156616, | |
| "learning_rate": 3.760237478849793e-08, | |
| "loss": 1.0177, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.9950248756218906, | |
| "eval_loss": 0.9390347003936768, | |
| "eval_runtime": 8.0407, | |
| "eval_samples_per_second": 21.143, | |
| "eval_steps_per_second": 5.348, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.0049751243781095, | |
| "grad_norm": 4.282508373260498, | |
| "learning_rate": 0.0, | |
| "loss": 1.4958, | |
| "step": 101 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 101, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 100, | |
| "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": 5.44348402876416e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |