| { | |
| "best_metric": 2.786395788192749, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.012832852101379532, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 6.416426050689766e-05, | |
| "grad_norm": 1.6019502878189087, | |
| "learning_rate": 1.004e-05, | |
| "loss": 7.1708, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 6.416426050689766e-05, | |
| "eval_loss": 3.9179491996765137, | |
| "eval_runtime": 703.5067, | |
| "eval_samples_per_second": 9.329, | |
| "eval_steps_per_second": 2.333, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00012832852101379532, | |
| "grad_norm": 1.6569987535476685, | |
| "learning_rate": 2.008e-05, | |
| "loss": 7.7498, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00019249278152069297, | |
| "grad_norm": 1.8272496461868286, | |
| "learning_rate": 3.012e-05, | |
| "loss": 7.4374, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00025665704202759064, | |
| "grad_norm": 1.8648499250411987, | |
| "learning_rate": 4.016e-05, | |
| "loss": 7.5896, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0003208213025344883, | |
| "grad_norm": 1.9629138708114624, | |
| "learning_rate": 5.02e-05, | |
| "loss": 7.1235, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00038498556304138594, | |
| "grad_norm": 2.0486273765563965, | |
| "learning_rate": 6.024e-05, | |
| "loss": 7.5494, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0004491498235482836, | |
| "grad_norm": 2.0849950313568115, | |
| "learning_rate": 7.028e-05, | |
| "loss": 7.0111, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0005133140840551813, | |
| "grad_norm": 2.051990509033203, | |
| "learning_rate": 8.032e-05, | |
| "loss": 7.0505, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0005774783445620789, | |
| "grad_norm": 2.1753077507019043, | |
| "learning_rate": 9.036000000000001e-05, | |
| "loss": 6.5329, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0006416426050689766, | |
| "grad_norm": 2.141965627670288, | |
| "learning_rate": 0.0001004, | |
| "loss": 6.9547, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0007058068655758742, | |
| "grad_norm": 1.723946213722229, | |
| "learning_rate": 9.987157894736842e-05, | |
| "loss": 6.5572, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0007699711260827719, | |
| "grad_norm": 1.7414573431015015, | |
| "learning_rate": 9.934315789473684e-05, | |
| "loss": 6.3907, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0008341353865896695, | |
| "grad_norm": 2.984980821609497, | |
| "learning_rate": 9.881473684210525e-05, | |
| "loss": 5.9726, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0008982996470965672, | |
| "grad_norm": 2.6435160636901855, | |
| "learning_rate": 9.828631578947369e-05, | |
| "loss": 5.9844, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0009624639076034649, | |
| "grad_norm": 2.249189615249634, | |
| "learning_rate": 9.77578947368421e-05, | |
| "loss": 6.4426, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0010266281681103626, | |
| "grad_norm": 2.079399347305298, | |
| "learning_rate": 9.722947368421052e-05, | |
| "loss": 6.2247, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.00109079242861726, | |
| "grad_norm": 1.9160220623016357, | |
| "learning_rate": 9.670105263157895e-05, | |
| "loss": 6.5336, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0011549566891241579, | |
| "grad_norm": 1.7296383380889893, | |
| "learning_rate": 9.617263157894737e-05, | |
| "loss": 6.0668, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0012191209496310554, | |
| "grad_norm": 2.0181539058685303, | |
| "learning_rate": 9.564421052631579e-05, | |
| "loss": 6.4754, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0012832852101379532, | |
| "grad_norm": 1.6639951467514038, | |
| "learning_rate": 9.511578947368421e-05, | |
| "loss": 5.3426, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.001347449470644851, | |
| "grad_norm": 2.009568214416504, | |
| "learning_rate": 9.458736842105264e-05, | |
| "loss": 6.255, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0014116137311517485, | |
| "grad_norm": 1.925960659980774, | |
| "learning_rate": 9.405894736842106e-05, | |
| "loss": 5.712, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0014757779916586462, | |
| "grad_norm": 1.931323528289795, | |
| "learning_rate": 9.353052631578947e-05, | |
| "loss": 5.7793, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0015399422521655437, | |
| "grad_norm": 2.025932550430298, | |
| "learning_rate": 9.300210526315789e-05, | |
| "loss": 5.5682, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0016041065126724415, | |
| "grad_norm": 1.761017918586731, | |
| "learning_rate": 9.247368421052631e-05, | |
| "loss": 6.1186, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.001668270773179339, | |
| "grad_norm": 2.1195151805877686, | |
| "learning_rate": 9.194526315789473e-05, | |
| "loss": 6.1748, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0017324350336862368, | |
| "grad_norm": 1.9101064205169678, | |
| "learning_rate": 9.141684210526316e-05, | |
| "loss": 5.9105, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0017965992941931343, | |
| "grad_norm": 1.8694759607315063, | |
| "learning_rate": 9.088842105263158e-05, | |
| "loss": 5.6025, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.001860763554700032, | |
| "grad_norm": 1.7502330541610718, | |
| "learning_rate": 9.036000000000001e-05, | |
| "loss": 5.4051, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0019249278152069298, | |
| "grad_norm": 2.0604054927825928, | |
| "learning_rate": 8.983157894736843e-05, | |
| "loss": 6.5442, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0019890920757138276, | |
| "grad_norm": 2.209613800048828, | |
| "learning_rate": 8.930315789473684e-05, | |
| "loss": 6.4631, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.002053256336220725, | |
| "grad_norm": 2.1784732341766357, | |
| "learning_rate": 8.877473684210526e-05, | |
| "loss": 6.2148, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0021174205967276227, | |
| "grad_norm": 2.089324712753296, | |
| "learning_rate": 8.824631578947368e-05, | |
| "loss": 5.6497, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.00218158485723452, | |
| "grad_norm": 2.2112064361572266, | |
| "learning_rate": 8.771789473684211e-05, | |
| "loss": 5.8131, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.002245749117741418, | |
| "grad_norm": 2.4185714721679688, | |
| "learning_rate": 8.718947368421053e-05, | |
| "loss": 6.0275, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0023099133782483157, | |
| "grad_norm": 2.314575433731079, | |
| "learning_rate": 8.666105263157895e-05, | |
| "loss": 5.5578, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0023740776387552133, | |
| "grad_norm": 2.286389112472534, | |
| "learning_rate": 8.613263157894737e-05, | |
| "loss": 5.8923, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.002438241899262111, | |
| "grad_norm": 2.207777976989746, | |
| "learning_rate": 8.560421052631578e-05, | |
| "loss": 5.9219, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0025024061597690088, | |
| "grad_norm": 2.3077611923217773, | |
| "learning_rate": 8.50757894736842e-05, | |
| "loss": 5.7264, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0025665704202759063, | |
| "grad_norm": 2.4047534465789795, | |
| "learning_rate": 8.454736842105263e-05, | |
| "loss": 6.2283, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.002630734680782804, | |
| "grad_norm": 2.7174782752990723, | |
| "learning_rate": 8.401894736842106e-05, | |
| "loss": 6.1087, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.002694898941289702, | |
| "grad_norm": 2.5407960414886475, | |
| "learning_rate": 8.349052631578948e-05, | |
| "loss": 6.0594, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0027590632017965994, | |
| "grad_norm": 2.6896162033081055, | |
| "learning_rate": 8.29621052631579e-05, | |
| "loss": 6.087, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.002823227462303497, | |
| "grad_norm": 2.7748165130615234, | |
| "learning_rate": 8.243368421052632e-05, | |
| "loss": 5.8548, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0028873917228103944, | |
| "grad_norm": 2.458576202392578, | |
| "learning_rate": 8.190526315789474e-05, | |
| "loss": 5.7433, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0029515559833172924, | |
| "grad_norm": 2.6704394817352295, | |
| "learning_rate": 8.137684210526315e-05, | |
| "loss": 5.944, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.00301572024382419, | |
| "grad_norm": 2.7400500774383545, | |
| "learning_rate": 8.084842105263157e-05, | |
| "loss": 5.8527, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0030798845043310875, | |
| "grad_norm": 2.7328145503997803, | |
| "learning_rate": 8.032e-05, | |
| "loss": 5.9252, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.003144048764837985, | |
| "grad_norm": 3.944028615951538, | |
| "learning_rate": 7.979157894736842e-05, | |
| "loss": 6.5863, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.003208213025344883, | |
| "grad_norm": 3.3126437664031982, | |
| "learning_rate": 7.926315789473684e-05, | |
| "loss": 5.661, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.003208213025344883, | |
| "eval_loss": 2.9033212661743164, | |
| "eval_runtime": 705.8123, | |
| "eval_samples_per_second": 9.299, | |
| "eval_steps_per_second": 2.325, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0032723772858517805, | |
| "grad_norm": 1.5550154447555542, | |
| "learning_rate": 7.873473684210526e-05, | |
| "loss": 5.5412, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.003336541546358678, | |
| "grad_norm": 1.3860490322113037, | |
| "learning_rate": 7.820631578947369e-05, | |
| "loss": 5.5933, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.003400705806865576, | |
| "grad_norm": 1.4063161611557007, | |
| "learning_rate": 7.76778947368421e-05, | |
| "loss": 5.9446, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0034648700673724736, | |
| "grad_norm": 1.0651419162750244, | |
| "learning_rate": 7.714947368421052e-05, | |
| "loss": 5.664, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.003529034327879371, | |
| "grad_norm": 1.0154882669448853, | |
| "learning_rate": 7.662105263157896e-05, | |
| "loss": 5.0898, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0035931985883862687, | |
| "grad_norm": 1.1615040302276611, | |
| "learning_rate": 7.609263157894737e-05, | |
| "loss": 6.0429, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0036573628488931666, | |
| "grad_norm": 1.004925012588501, | |
| "learning_rate": 7.556421052631579e-05, | |
| "loss": 5.1923, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.003721527109400064, | |
| "grad_norm": 1.1468150615692139, | |
| "learning_rate": 7.503578947368421e-05, | |
| "loss": 5.5039, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0037856913699069617, | |
| "grad_norm": 1.3373304605484009, | |
| "learning_rate": 7.450736842105263e-05, | |
| "loss": 5.6979, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0038498556304138597, | |
| "grad_norm": 1.1840150356292725, | |
| "learning_rate": 7.397894736842105e-05, | |
| "loss": 5.5324, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.003914019890920757, | |
| "grad_norm": 1.2353522777557373, | |
| "learning_rate": 7.345052631578948e-05, | |
| "loss": 5.8171, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.003978184151427655, | |
| "grad_norm": 1.2683886289596558, | |
| "learning_rate": 7.29221052631579e-05, | |
| "loss": 5.7111, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.004042348411934553, | |
| "grad_norm": 1.293735384941101, | |
| "learning_rate": 7.239368421052631e-05, | |
| "loss": 6.0277, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.00410651267244145, | |
| "grad_norm": 1.3527857065200806, | |
| "learning_rate": 7.186526315789474e-05, | |
| "loss": 5.5717, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.004170676932948348, | |
| "grad_norm": 1.3823940753936768, | |
| "learning_rate": 7.133684210526316e-05, | |
| "loss": 5.7555, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.004234841193455245, | |
| "grad_norm": 1.4544517993927002, | |
| "learning_rate": 7.080842105263158e-05, | |
| "loss": 5.6251, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.004299005453962143, | |
| "grad_norm": 1.45553457736969, | |
| "learning_rate": 7.028e-05, | |
| "loss": 6.0601, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.00436316971446904, | |
| "grad_norm": 1.3593374490737915, | |
| "learning_rate": 6.975157894736843e-05, | |
| "loss": 5.3168, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.004427333974975939, | |
| "grad_norm": 1.7438663244247437, | |
| "learning_rate": 6.922315789473685e-05, | |
| "loss": 6.2382, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.004491498235482836, | |
| "grad_norm": 1.430897831916809, | |
| "learning_rate": 6.869473684210527e-05, | |
| "loss": 5.5089, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.004555662495989734, | |
| "grad_norm": 1.5499473810195923, | |
| "learning_rate": 6.816631578947368e-05, | |
| "loss": 5.5141, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0046198267564966315, | |
| "grad_norm": 1.4696685075759888, | |
| "learning_rate": 6.76378947368421e-05, | |
| "loss": 6.2138, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.004683991017003529, | |
| "grad_norm": 1.6637015342712402, | |
| "learning_rate": 6.710947368421052e-05, | |
| "loss": 5.5004, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0047481552775104265, | |
| "grad_norm": 1.773887276649475, | |
| "learning_rate": 6.658105263157894e-05, | |
| "loss": 5.1702, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.004812319538017324, | |
| "grad_norm": 1.6259382963180542, | |
| "learning_rate": 6.605263157894737e-05, | |
| "loss": 6.0664, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.004876483798524222, | |
| "grad_norm": 2.1414995193481445, | |
| "learning_rate": 6.55242105263158e-05, | |
| "loss": 6.0428, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.00494064805903112, | |
| "grad_norm": 1.9427917003631592, | |
| "learning_rate": 6.499578947368422e-05, | |
| "loss": 5.7616, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.0050048123195380175, | |
| "grad_norm": 1.8610105514526367, | |
| "learning_rate": 6.446736842105264e-05, | |
| "loss": 5.6827, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.005068976580044915, | |
| "grad_norm": 2.151019811630249, | |
| "learning_rate": 6.393894736842105e-05, | |
| "loss": 5.4517, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.005133140840551813, | |
| "grad_norm": 1.7757065296173096, | |
| "learning_rate": 6.341052631578947e-05, | |
| "loss": 5.5975, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.00519730510105871, | |
| "grad_norm": 2.1130731105804443, | |
| "learning_rate": 6.288210526315789e-05, | |
| "loss": 5.7106, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.005261469361565608, | |
| "grad_norm": 1.9320672750473022, | |
| "learning_rate": 6.235368421052632e-05, | |
| "loss": 5.9469, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.005325633622072505, | |
| "grad_norm": 2.143235683441162, | |
| "learning_rate": 6.182526315789474e-05, | |
| "loss": 5.7521, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.005389797882579404, | |
| "grad_norm": 2.1077873706817627, | |
| "learning_rate": 6.129684210526316e-05, | |
| "loss": 5.7203, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.005453962143086301, | |
| "grad_norm": 2.1506385803222656, | |
| "learning_rate": 6.076842105263158e-05, | |
| "loss": 6.1618, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.005518126403593199, | |
| "grad_norm": 2.0908615589141846, | |
| "learning_rate": 6.024e-05, | |
| "loss": 5.4239, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.005582290664100096, | |
| "grad_norm": 2.7268731594085693, | |
| "learning_rate": 5.971157894736842e-05, | |
| "loss": 5.7719, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.005646454924606994, | |
| "grad_norm": 2.2536518573760986, | |
| "learning_rate": 5.9183157894736835e-05, | |
| "loss": 5.7436, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.005710619185113891, | |
| "grad_norm": 2.401059865951538, | |
| "learning_rate": 5.8654736842105267e-05, | |
| "loss": 5.8639, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.005774783445620789, | |
| "grad_norm": 2.6088643074035645, | |
| "learning_rate": 5.8126315789473684e-05, | |
| "loss": 6.0605, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.005838947706127687, | |
| "grad_norm": 2.477994680404663, | |
| "learning_rate": 5.759789473684211e-05, | |
| "loss": 5.752, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.005903111966634585, | |
| "grad_norm": 2.4255878925323486, | |
| "learning_rate": 5.706947368421053e-05, | |
| "loss": 6.0441, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.005967276227141482, | |
| "grad_norm": 2.4979171752929688, | |
| "learning_rate": 5.6541052631578945e-05, | |
| "loss": 5.5721, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.00603144048764838, | |
| "grad_norm": 2.662763833999634, | |
| "learning_rate": 5.601263157894736e-05, | |
| "loss": 5.7948, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.0060956047481552774, | |
| "grad_norm": 2.7771427631378174, | |
| "learning_rate": 5.5484210526315794e-05, | |
| "loss": 5.5672, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.006159769008662175, | |
| "grad_norm": 2.8292243480682373, | |
| "learning_rate": 5.495578947368421e-05, | |
| "loss": 5.5114, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.0062239332691690725, | |
| "grad_norm": 2.8949716091156006, | |
| "learning_rate": 5.442736842105264e-05, | |
| "loss": 5.5617, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.00628809752967597, | |
| "grad_norm": 3.2557790279388428, | |
| "learning_rate": 5.3898947368421055e-05, | |
| "loss": 6.2371, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.0063522617901828685, | |
| "grad_norm": 3.2778756618499756, | |
| "learning_rate": 5.337052631578947e-05, | |
| "loss": 5.3658, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.006416426050689766, | |
| "grad_norm": 3.3429481983184814, | |
| "learning_rate": 5.284210526315789e-05, | |
| "loss": 6.0804, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.006416426050689766, | |
| "eval_loss": 2.8372957706451416, | |
| "eval_runtime": 702.8035, | |
| "eval_samples_per_second": 9.338, | |
| "eval_steps_per_second": 2.335, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0064805903111966635, | |
| "grad_norm": 1.095025658607483, | |
| "learning_rate": 5.231368421052631e-05, | |
| "loss": 5.3962, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.006544754571703561, | |
| "grad_norm": 1.1655224561691284, | |
| "learning_rate": 5.178526315789474e-05, | |
| "loss": 5.6473, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.006608918832210459, | |
| "grad_norm": 1.1299000978469849, | |
| "learning_rate": 5.1256842105263165e-05, | |
| "loss": 5.5021, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.006673083092717356, | |
| "grad_norm": 1.0751549005508423, | |
| "learning_rate": 5.072842105263158e-05, | |
| "loss": 5.6848, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.006737247353224254, | |
| "grad_norm": 1.1250840425491333, | |
| "learning_rate": 5.02e-05, | |
| "loss": 5.7385, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.006801411613731152, | |
| "grad_norm": 1.096813440322876, | |
| "learning_rate": 4.967157894736842e-05, | |
| "loss": 5.4361, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.00686557587423805, | |
| "grad_norm": 1.1172356605529785, | |
| "learning_rate": 4.914315789473684e-05, | |
| "loss": 5.141, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.006929740134744947, | |
| "grad_norm": 1.2574751377105713, | |
| "learning_rate": 4.861473684210526e-05, | |
| "loss": 5.7921, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.006993904395251845, | |
| "grad_norm": 1.1475943326950073, | |
| "learning_rate": 4.8086315789473686e-05, | |
| "loss": 5.3017, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.007058068655758742, | |
| "grad_norm": 1.182779312133789, | |
| "learning_rate": 4.7557894736842104e-05, | |
| "loss": 5.2065, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.00712223291626564, | |
| "grad_norm": 1.1800684928894043, | |
| "learning_rate": 4.702947368421053e-05, | |
| "loss": 5.4792, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.007186397176772537, | |
| "grad_norm": 1.346672773361206, | |
| "learning_rate": 4.6501052631578946e-05, | |
| "loss": 5.9185, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.007250561437279436, | |
| "grad_norm": 1.3497660160064697, | |
| "learning_rate": 4.5972631578947364e-05, | |
| "loss": 5.7997, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.007314725697786333, | |
| "grad_norm": 1.487912654876709, | |
| "learning_rate": 4.544421052631579e-05, | |
| "loss": 6.1196, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.007378889958293231, | |
| "grad_norm": 1.363545298576355, | |
| "learning_rate": 4.4915789473684213e-05, | |
| "loss": 5.2859, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.007443054218800128, | |
| "grad_norm": 1.5003371238708496, | |
| "learning_rate": 4.438736842105263e-05, | |
| "loss": 5.5254, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.007507218479307026, | |
| "grad_norm": 1.685253381729126, | |
| "learning_rate": 4.3858947368421056e-05, | |
| "loss": 5.5345, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.007571382739813923, | |
| "grad_norm": 1.8338501453399658, | |
| "learning_rate": 4.3330526315789474e-05, | |
| "loss": 5.6438, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.007635547000320821, | |
| "grad_norm": 1.5995521545410156, | |
| "learning_rate": 4.280210526315789e-05, | |
| "loss": 5.6032, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.007699711260827719, | |
| "grad_norm": 1.5582114458084106, | |
| "learning_rate": 4.2273684210526317e-05, | |
| "loss": 5.7405, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.007763875521334617, | |
| "grad_norm": 2.019646644592285, | |
| "learning_rate": 4.174526315789474e-05, | |
| "loss": 6.0767, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.007828039781841514, | |
| "grad_norm": 1.9564622640609741, | |
| "learning_rate": 4.121684210526316e-05, | |
| "loss": 6.3192, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.007892204042348411, | |
| "grad_norm": 1.720892310142517, | |
| "learning_rate": 4.068842105263158e-05, | |
| "loss": 5.6769, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.00795636830285531, | |
| "grad_norm": 1.9853098392486572, | |
| "learning_rate": 4.016e-05, | |
| "loss": 5.4357, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.008020532563362208, | |
| "grad_norm": 1.7024343013763428, | |
| "learning_rate": 3.963157894736842e-05, | |
| "loss": 5.605, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.008084696823869105, | |
| "grad_norm": 1.926780343055725, | |
| "learning_rate": 3.9103157894736844e-05, | |
| "loss": 5.5458, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.008148861084376003, | |
| "grad_norm": 1.8911856412887573, | |
| "learning_rate": 3.857473684210526e-05, | |
| "loss": 5.5754, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.0082130253448829, | |
| "grad_norm": 1.9950193166732788, | |
| "learning_rate": 3.804631578947369e-05, | |
| "loss": 5.7808, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.008277189605389798, | |
| "grad_norm": 2.379807472229004, | |
| "learning_rate": 3.7517894736842105e-05, | |
| "loss": 5.5042, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.008341353865896696, | |
| "grad_norm": 1.982927918434143, | |
| "learning_rate": 3.698947368421052e-05, | |
| "loss": 4.8575, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.008405518126403593, | |
| "grad_norm": 2.424283266067505, | |
| "learning_rate": 3.646105263157895e-05, | |
| "loss": 6.3244, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.00846968238691049, | |
| "grad_norm": 2.2915103435516357, | |
| "learning_rate": 3.593263157894737e-05, | |
| "loss": 5.3091, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.008533846647417388, | |
| "grad_norm": 2.128199815750122, | |
| "learning_rate": 3.540421052631579e-05, | |
| "loss": 5.6833, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.008598010907924286, | |
| "grad_norm": 2.3494131565093994, | |
| "learning_rate": 3.4875789473684215e-05, | |
| "loss": 5.435, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.008662175168431183, | |
| "grad_norm": 2.4190518856048584, | |
| "learning_rate": 3.434736842105263e-05, | |
| "loss": 5.7771, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.00872633942893808, | |
| "grad_norm": 2.0826826095581055, | |
| "learning_rate": 3.381894736842105e-05, | |
| "loss": 5.6895, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.008790503689444978, | |
| "grad_norm": 2.357288122177124, | |
| "learning_rate": 3.329052631578947e-05, | |
| "loss": 4.9487, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.008854667949951878, | |
| "grad_norm": 2.6756513118743896, | |
| "learning_rate": 3.27621052631579e-05, | |
| "loss": 5.5356, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.008918832210458775, | |
| "grad_norm": 2.8232810497283936, | |
| "learning_rate": 3.223368421052632e-05, | |
| "loss": 6.3404, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.008982996470965673, | |
| "grad_norm": 2.7257964611053467, | |
| "learning_rate": 3.1705263157894736e-05, | |
| "loss": 5.493, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.00904716073147257, | |
| "grad_norm": 2.5550906658172607, | |
| "learning_rate": 3.117684210526316e-05, | |
| "loss": 5.3718, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.009111324991979468, | |
| "grad_norm": 2.4199259281158447, | |
| "learning_rate": 3.064842105263158e-05, | |
| "loss": 5.3804, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.009175489252486365, | |
| "grad_norm": 2.59352445602417, | |
| "learning_rate": 3.012e-05, | |
| "loss": 5.7932, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.009239653512993263, | |
| "grad_norm": 2.602415084838867, | |
| "learning_rate": 2.9591578947368418e-05, | |
| "loss": 5.0882, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.00930381777350016, | |
| "grad_norm": 3.0608277320861816, | |
| "learning_rate": 2.9063157894736842e-05, | |
| "loss": 5.6549, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.009367982034007058, | |
| "grad_norm": 2.965409994125366, | |
| "learning_rate": 2.8534736842105264e-05, | |
| "loss": 5.5571, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.009432146294513956, | |
| "grad_norm": 2.993109941482544, | |
| "learning_rate": 2.800631578947368e-05, | |
| "loss": 5.1919, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.009496310555020853, | |
| "grad_norm": 3.2940351963043213, | |
| "learning_rate": 2.7477894736842106e-05, | |
| "loss": 4.7934, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.00956047481552775, | |
| "grad_norm": 3.4445641040802, | |
| "learning_rate": 2.6949473684210527e-05, | |
| "loss": 5.3651, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.009624639076034648, | |
| "grad_norm": 3.1870310306549072, | |
| "learning_rate": 2.6421052631578945e-05, | |
| "loss": 5.4287, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.009624639076034648, | |
| "eval_loss": 2.803600311279297, | |
| "eval_runtime": 703.2591, | |
| "eval_samples_per_second": 9.332, | |
| "eval_steps_per_second": 2.333, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.009688803336541546, | |
| "grad_norm": 1.013375997543335, | |
| "learning_rate": 2.589263157894737e-05, | |
| "loss": 5.1892, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.009752967597048443, | |
| "grad_norm": 1.2413551807403564, | |
| "learning_rate": 2.536421052631579e-05, | |
| "loss": 6.1944, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.009817131857555342, | |
| "grad_norm": 1.0705325603485107, | |
| "learning_rate": 2.483578947368421e-05, | |
| "loss": 5.8975, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.00988129611806224, | |
| "grad_norm": 1.1972635984420776, | |
| "learning_rate": 2.430736842105263e-05, | |
| "loss": 5.9433, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.009945460378569138, | |
| "grad_norm": 1.179183840751648, | |
| "learning_rate": 2.3778947368421052e-05, | |
| "loss": 5.5273, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.010009624639076035, | |
| "grad_norm": 1.2335314750671387, | |
| "learning_rate": 2.3250526315789473e-05, | |
| "loss": 5.4563, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.010073788899582933, | |
| "grad_norm": 1.1595032215118408, | |
| "learning_rate": 2.2722105263157894e-05, | |
| "loss": 5.3758, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.01013795316008983, | |
| "grad_norm": 1.3281739950180054, | |
| "learning_rate": 2.2193684210526316e-05, | |
| "loss": 5.7909, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.010202117420596728, | |
| "grad_norm": 1.3218365907669067, | |
| "learning_rate": 2.1665263157894737e-05, | |
| "loss": 5.8669, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.010266281681103625, | |
| "grad_norm": 1.4711360931396484, | |
| "learning_rate": 2.1136842105263158e-05, | |
| "loss": 5.346, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.010330445941610523, | |
| "grad_norm": 1.376670002937317, | |
| "learning_rate": 2.060842105263158e-05, | |
| "loss": 5.9901, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.01039461020211742, | |
| "grad_norm": 1.5264054536819458, | |
| "learning_rate": 2.008e-05, | |
| "loss": 5.2455, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.010458774462624318, | |
| "grad_norm": 1.338756799697876, | |
| "learning_rate": 1.9551578947368422e-05, | |
| "loss": 5.5642, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.010522938723131215, | |
| "grad_norm": 1.5276857614517212, | |
| "learning_rate": 1.9023157894736843e-05, | |
| "loss": 5.7341, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.010587102983638113, | |
| "grad_norm": 1.5230273008346558, | |
| "learning_rate": 1.849473684210526e-05, | |
| "loss": 5.4654, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.01065126724414501, | |
| "grad_norm": 1.5083640813827515, | |
| "learning_rate": 1.7966315789473686e-05, | |
| "loss": 5.2931, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.010715431504651908, | |
| "grad_norm": 1.4739644527435303, | |
| "learning_rate": 1.7437894736842107e-05, | |
| "loss": 5.7144, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.010779595765158807, | |
| "grad_norm": 1.5886435508728027, | |
| "learning_rate": 1.6909473684210525e-05, | |
| "loss": 5.8369, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.010843760025665705, | |
| "grad_norm": 1.5803991556167603, | |
| "learning_rate": 1.638105263157895e-05, | |
| "loss": 5.4256, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.010907924286172602, | |
| "grad_norm": 1.5075167417526245, | |
| "learning_rate": 1.5852631578947368e-05, | |
| "loss": 5.5388, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.0109720885466795, | |
| "grad_norm": 1.7619450092315674, | |
| "learning_rate": 1.532421052631579e-05, | |
| "loss": 5.5654, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.011036252807186397, | |
| "grad_norm": 1.7552381753921509, | |
| "learning_rate": 1.4795789473684209e-05, | |
| "loss": 5.5068, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.011100417067693295, | |
| "grad_norm": 1.588983178138733, | |
| "learning_rate": 1.4267368421052632e-05, | |
| "loss": 5.3193, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.011164581328200193, | |
| "grad_norm": 1.6596215963363647, | |
| "learning_rate": 1.3738947368421053e-05, | |
| "loss": 4.9316, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.01122874558870709, | |
| "grad_norm": 2.2129220962524414, | |
| "learning_rate": 1.3210526315789473e-05, | |
| "loss": 6.2231, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.011292909849213988, | |
| "grad_norm": 1.8372780084609985, | |
| "learning_rate": 1.2682105263157896e-05, | |
| "loss": 6.2628, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.011357074109720885, | |
| "grad_norm": 1.8769925832748413, | |
| "learning_rate": 1.2153684210526315e-05, | |
| "loss": 5.8288, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.011421238370227783, | |
| "grad_norm": 1.7789459228515625, | |
| "learning_rate": 1.1625263157894737e-05, | |
| "loss": 5.4904, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.01148540263073468, | |
| "grad_norm": 2.0547878742218018, | |
| "learning_rate": 1.1096842105263158e-05, | |
| "loss": 6.2145, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.011549566891241578, | |
| "grad_norm": 2.1852498054504395, | |
| "learning_rate": 1.0568421052631579e-05, | |
| "loss": 5.3296, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.011613731151748475, | |
| "grad_norm": 2.358675479888916, | |
| "learning_rate": 1.004e-05, | |
| "loss": 5.3969, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.011677895412255375, | |
| "grad_norm": 2.2292699813842773, | |
| "learning_rate": 9.511578947368422e-06, | |
| "loss": 5.8842, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.011742059672762272, | |
| "grad_norm": 2.164508581161499, | |
| "learning_rate": 8.983157894736843e-06, | |
| "loss": 5.7544, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.01180622393326917, | |
| "grad_norm": 2.151883125305176, | |
| "learning_rate": 8.454736842105263e-06, | |
| "loss": 5.8768, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.011870388193776067, | |
| "grad_norm": 2.0282275676727295, | |
| "learning_rate": 7.926315789473684e-06, | |
| "loss": 5.936, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.011934552454282965, | |
| "grad_norm": 2.4359304904937744, | |
| "learning_rate": 7.397894736842104e-06, | |
| "loss": 5.7367, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.011998716714789862, | |
| "grad_norm": 2.2616195678710938, | |
| "learning_rate": 6.8694736842105265e-06, | |
| "loss": 5.2675, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.01206288097529676, | |
| "grad_norm": 2.1506917476654053, | |
| "learning_rate": 6.341052631578948e-06, | |
| "loss": 5.5162, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.012127045235803657, | |
| "grad_norm": 2.314349889755249, | |
| "learning_rate": 5.812631578947368e-06, | |
| "loss": 5.7775, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.012191209496310555, | |
| "grad_norm": 2.3769171237945557, | |
| "learning_rate": 5.2842105263157896e-06, | |
| "loss": 6.1183, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.012255373756817452, | |
| "grad_norm": 2.4818451404571533, | |
| "learning_rate": 4.755789473684211e-06, | |
| "loss": 5.7682, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.01231953801732435, | |
| "grad_norm": 2.498087167739868, | |
| "learning_rate": 4.227368421052631e-06, | |
| "loss": 5.555, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.012383702277831247, | |
| "grad_norm": 2.425711154937744, | |
| "learning_rate": 3.698947368421052e-06, | |
| "loss": 5.1028, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.012447866538338145, | |
| "grad_norm": 2.878943920135498, | |
| "learning_rate": 3.170526315789474e-06, | |
| "loss": 5.748, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.012512030798845043, | |
| "grad_norm": 2.801713466644287, | |
| "learning_rate": 2.6421052631578948e-06, | |
| "loss": 6.4606, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.01257619505935194, | |
| "grad_norm": 2.964569568634033, | |
| "learning_rate": 2.1136842105263157e-06, | |
| "loss": 5.5059, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.01264035931985884, | |
| "grad_norm": 2.598716974258423, | |
| "learning_rate": 1.585263157894737e-06, | |
| "loss": 5.0059, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.012704523580365737, | |
| "grad_norm": 3.2355284690856934, | |
| "learning_rate": 1.0568421052631578e-06, | |
| "loss": 5.8954, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.012768687840872634, | |
| "grad_norm": 2.76204514503479, | |
| "learning_rate": 5.284210526315789e-07, | |
| "loss": 5.7194, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.012832852101379532, | |
| "grad_norm": 3.6820626258850098, | |
| "learning_rate": 0.0, | |
| "loss": 5.3636, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.012832852101379532, | |
| "eval_loss": 2.786395788192749, | |
| "eval_runtime": 702.4656, | |
| "eval_samples_per_second": 9.343, | |
| "eval_steps_per_second": 2.336, | |
| "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": 5.963334542480179e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |