| { | |
| "best_metric": 0.0004729398642666638, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 2.9304029304029307, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.014652014652014652, | |
| "grad_norm": 7.069050312042236, | |
| "learning_rate": 1e-05, | |
| "loss": 5.7842, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.014652014652014652, | |
| "eval_loss": 1.6904494762420654, | |
| "eval_runtime": 12.2524, | |
| "eval_samples_per_second": 9.386, | |
| "eval_steps_per_second": 2.367, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.029304029304029304, | |
| "grad_norm": 8.143122673034668, | |
| "learning_rate": 2e-05, | |
| "loss": 5.4593, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.04395604395604396, | |
| "grad_norm": 8.063944816589355, | |
| "learning_rate": 3e-05, | |
| "loss": 6.2371, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.05860805860805861, | |
| "grad_norm": 7.555496692657471, | |
| "learning_rate": 4e-05, | |
| "loss": 6.1426, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.07326007326007326, | |
| "grad_norm": 7.24888277053833, | |
| "learning_rate": 5e-05, | |
| "loss": 6.4686, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.08791208791208792, | |
| "grad_norm": 9.164324760437012, | |
| "learning_rate": 6e-05, | |
| "loss": 5.4746, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.10256410256410256, | |
| "grad_norm": 7.516927242279053, | |
| "learning_rate": 7e-05, | |
| "loss": 5.1506, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.11721611721611722, | |
| "grad_norm": 7.269052982330322, | |
| "learning_rate": 8e-05, | |
| "loss": 4.6344, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.13186813186813187, | |
| "grad_norm": 8.308143615722656, | |
| "learning_rate": 9e-05, | |
| "loss": 4.0138, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.14652014652014653, | |
| "grad_norm": 8.23723316192627, | |
| "learning_rate": 0.0001, | |
| "loss": 3.8622, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.16117216117216118, | |
| "grad_norm": 7.4196906089782715, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 3.573, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.17582417582417584, | |
| "grad_norm": 12.47601318359375, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 2.9025, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.19047619047619047, | |
| "grad_norm": 7.383912086486816, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 2.2465, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.20512820512820512, | |
| "grad_norm": 8.258135795593262, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.728, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.21978021978021978, | |
| "grad_norm": 8.764373779296875, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.879, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.23443223443223443, | |
| "grad_norm": 18.639711380004883, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 2.8013, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.2490842490842491, | |
| "grad_norm": 15.905421257019043, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 2.1102, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.26373626373626374, | |
| "grad_norm": 30.563175201416016, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 5.1491, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.2783882783882784, | |
| "grad_norm": 19.737285614013672, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 3.6496, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.29304029304029305, | |
| "grad_norm": 10.716526985168457, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 2.9596, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.3076923076923077, | |
| "grad_norm": 7.916341304779053, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 2.9004, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.32234432234432236, | |
| "grad_norm": 8.126665115356445, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 2.7632, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.336996336996337, | |
| "grad_norm": 7.077164173126221, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 2.414, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.3516483516483517, | |
| "grad_norm": 7.344516754150391, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.6168, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.3663003663003663, | |
| "grad_norm": 5.7711310386657715, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.3041, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.38095238095238093, | |
| "grad_norm": 3.6991608142852783, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.5436, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.3956043956043956, | |
| "grad_norm": 3.9110922813415527, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.4839, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.41025641025641024, | |
| "grad_norm": 3.7960498332977295, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.4639, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.4249084249084249, | |
| "grad_norm": 5.450497627258301, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.2969, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.43956043956043955, | |
| "grad_norm": 2.457590341567993, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.186, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.4542124542124542, | |
| "grad_norm": 2.6626298427581787, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.1133, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.46886446886446886, | |
| "grad_norm": 2.197824478149414, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.1868, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.4835164835164835, | |
| "grad_norm": 5.061366558074951, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.3182, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.4981684981684982, | |
| "grad_norm": 3.3193469047546387, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.2218, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.5128205128205128, | |
| "grad_norm": 20.37020492553711, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 3.2363, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.5274725274725275, | |
| "grad_norm": 17.228864669799805, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 2.0558, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.5421245421245421, | |
| "grad_norm": 14.611564636230469, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.4301, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.5567765567765568, | |
| "grad_norm": 10.096405982971191, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.2961, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 6.464880466461182, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.8231, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.5860805860805861, | |
| "grad_norm": 9.777287483215332, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.7446, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.6007326007326007, | |
| "grad_norm": 5.354022979736328, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.3958, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.6153846153846154, | |
| "grad_norm": 2.9241244792938232, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.2194, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.63003663003663, | |
| "grad_norm": 3.4176015853881836, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.2476, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.6446886446886447, | |
| "grad_norm": 2.5821731090545654, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.2222, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.6593406593406593, | |
| "grad_norm": 1.424792766571045, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.1326, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.673992673992674, | |
| "grad_norm": 1.426535964012146, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.1272, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.6886446886446886, | |
| "grad_norm": 1.7181131839752197, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.0655, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.7032967032967034, | |
| "grad_norm": 1.1161047220230103, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.0428, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.717948717948718, | |
| "grad_norm": 1.7914754152297974, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.0822, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.7326007326007326, | |
| "grad_norm": 5.58827018737793, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.0789, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.7326007326007326, | |
| "eval_loss": 0.11392630636692047, | |
| "eval_runtime": 12.562, | |
| "eval_samples_per_second": 9.155, | |
| "eval_steps_per_second": 2.309, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.7472527472527473, | |
| "grad_norm": 1.4648451805114746, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.0189, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.7619047619047619, | |
| "grad_norm": 9.68092155456543, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.8904, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.7765567765567766, | |
| "grad_norm": 5.347242832183838, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.8427, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.7912087912087912, | |
| "grad_norm": 4.402308464050293, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.464, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.8058608058608059, | |
| "grad_norm": 3.3170294761657715, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.2698, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.8205128205128205, | |
| "grad_norm": 2.2884583473205566, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.0951, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.8351648351648352, | |
| "grad_norm": 4.389698028564453, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.1663, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.8498168498168498, | |
| "grad_norm": 2.8354127407073975, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.0847, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.8644688644688645, | |
| "grad_norm": 2.3950765132904053, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.072, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.8791208791208791, | |
| "grad_norm": 6.019696235656738, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.1688, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.8937728937728938, | |
| "grad_norm": 3.865851640701294, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.0934, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.9084249084249084, | |
| "grad_norm": 2.18562650680542, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.0477, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.9230769230769231, | |
| "grad_norm": 0.7486851811408997, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.0158, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.9377289377289377, | |
| "grad_norm": 0.42197975516319275, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.0096, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.9523809523809523, | |
| "grad_norm": 0.20389479398727417, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.0063, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.967032967032967, | |
| "grad_norm": 1.4959765672683716, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.0167, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.9816849816849816, | |
| "grad_norm": 0.18284693360328674, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.0048, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.9963369963369964, | |
| "grad_norm": 1.016512155532837, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.0159, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.010989010989011, | |
| "grad_norm": 4.817726135253906, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.6674, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.0256410256410255, | |
| "grad_norm": 5.86989164352417, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.6466, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.0402930402930404, | |
| "grad_norm": 5.039214611053467, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.6084, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.054945054945055, | |
| "grad_norm": 3.513340950012207, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.3211, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.0695970695970696, | |
| "grad_norm": 2.6308839321136475, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.2329, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.0842490842490842, | |
| "grad_norm": 2.06376051902771, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.0905, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.098901098901099, | |
| "grad_norm": 1.184475302696228, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.0292, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.1135531135531136, | |
| "grad_norm": 0.8018744587898254, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.0322, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.1282051282051282, | |
| "grad_norm": 1.0255945920944214, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.032, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.1428571428571428, | |
| "grad_norm": 0.40411192178726196, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.0121, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.1575091575091574, | |
| "grad_norm": 0.1450580507516861, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.0043, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.1721611721611722, | |
| "grad_norm": 0.656509280204773, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.0083, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.1868131868131868, | |
| "grad_norm": 0.24206946790218353, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.0038, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.2014652014652014, | |
| "grad_norm": 0.052082084119319916, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.0019, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.2161172161172162, | |
| "grad_norm": 0.05583946034312248, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.0019, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.2307692307692308, | |
| "grad_norm": 0.585628867149353, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.0076, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.2454212454212454, | |
| "grad_norm": 0.2598362863063812, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.0095, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.26007326007326, | |
| "grad_norm": 6.782217025756836, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.5107, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.2747252747252746, | |
| "grad_norm": 6.780796051025391, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.5346, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.2893772893772895, | |
| "grad_norm": 19.290767669677734, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.5559, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.304029304029304, | |
| "grad_norm": 3.194103240966797, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.1941, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.3186813186813187, | |
| "grad_norm": 1.8198635578155518, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.1214, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 1.0340523719787598, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.0492, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.347985347985348, | |
| "grad_norm": 2.1145784854888916, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.0332, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.3626373626373627, | |
| "grad_norm": 0.3792741000652313, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.0054, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.3772893772893773, | |
| "grad_norm": 1.7770278453826904, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.0091, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.3919413919413919, | |
| "grad_norm": 0.09800445288419724, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.0014, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.4065934065934065, | |
| "grad_norm": 0.08425897359848022, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.0011, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.4212454212454213, | |
| "grad_norm": 0.06335113942623138, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.0012, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.435897435897436, | |
| "grad_norm": 0.4823334515094757, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.0018, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.4505494505494505, | |
| "grad_norm": 0.07274054735898972, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.001, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.4652014652014653, | |
| "grad_norm": 0.0599992610514164, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.0009, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.4652014652014653, | |
| "eval_loss": 0.017204292118549347, | |
| "eval_runtime": 12.5441, | |
| "eval_samples_per_second": 9.168, | |
| "eval_steps_per_second": 2.312, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.47985347985348, | |
| "grad_norm": 0.019661907106637955, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.0007, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.4945054945054945, | |
| "grad_norm": 0.06000884249806404, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.0008, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.5091575091575091, | |
| "grad_norm": 4.516892910003662, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.4002, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.5238095238095237, | |
| "grad_norm": 6.432425498962402, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.2111, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.5384615384615383, | |
| "grad_norm": 3.1115596294403076, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0868, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.5531135531135531, | |
| "grad_norm": 1.811315655708313, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.0364, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.5677655677655677, | |
| "grad_norm": 0.5356123447418213, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.0111, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.5824175824175826, | |
| "grad_norm": 0.3036579191684723, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.0039, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.5970695970695972, | |
| "grad_norm": 0.07567751407623291, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.0017, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.6117216117216118, | |
| "grad_norm": 0.06660040467977524, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.0015, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.6263736263736264, | |
| "grad_norm": 0.0334957018494606, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.0009, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.641025641025641, | |
| "grad_norm": 0.027862094342708588, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.0008, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.6556776556776556, | |
| "grad_norm": 0.250003844499588, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.0017, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.6703296703296702, | |
| "grad_norm": 0.06321448087692261, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.001, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.684981684981685, | |
| "grad_norm": 0.11561797559261322, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.0013, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.6996336996336996, | |
| "grad_norm": 0.01735706254839897, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.0006, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.7142857142857144, | |
| "grad_norm": 0.010219499468803406, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.0005, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.728937728937729, | |
| "grad_norm": 0.032738737761974335, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.0006, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.7435897435897436, | |
| "grad_norm": 0.02502625808119774, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.0005, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.7582417582417582, | |
| "grad_norm": 3.037071466445923, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.1075, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.7728937728937728, | |
| "grad_norm": 2.78204083442688, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.1182, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.7875457875457874, | |
| "grad_norm": 3.9707586765289307, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.1579, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.8021978021978022, | |
| "grad_norm": 1.3319613933563232, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.0229, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.8168498168498168, | |
| "grad_norm": 0.32199329137802124, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.0075, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.8315018315018317, | |
| "grad_norm": 0.13943661749362946, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.0021, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.8461538461538463, | |
| "grad_norm": 0.10824523866176605, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.001, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.8608058608058609, | |
| "grad_norm": 0.020937751978635788, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.0007, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.8754578754578755, | |
| "grad_norm": 0.013725826516747475, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.0006, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.89010989010989, | |
| "grad_norm": 0.21734972298145294, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.0051, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.9047619047619047, | |
| "grad_norm": 0.021338798105716705, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.0006, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.9194139194139193, | |
| "grad_norm": 0.014328362420201302, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.0005, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.934065934065934, | |
| "grad_norm": 0.015361218713223934, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.0005, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.9487179487179487, | |
| "grad_norm": 0.01827828399837017, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.0005, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.9633699633699635, | |
| "grad_norm": 0.024537015706300735, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.0006, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.978021978021978, | |
| "grad_norm": 0.07515154778957367, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.0008, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.9926739926739927, | |
| "grad_norm": 0.30712074041366577, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.0014, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.0073260073260073, | |
| "grad_norm": 0.2810360789299011, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.0132, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 2.021978021978022, | |
| "grad_norm": 1.975380778312683, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.0395, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.0366300366300365, | |
| "grad_norm": 3.420339584350586, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.0776, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 2.051282051282051, | |
| "grad_norm": 1.0953068733215332, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.0215, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.065934065934066, | |
| "grad_norm": 0.35299283266067505, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.0035, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 2.0805860805860807, | |
| "grad_norm": 0.02507457509636879, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.0012, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.0952380952380953, | |
| "grad_norm": 0.21908210217952728, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.001, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 2.10989010989011, | |
| "grad_norm": 0.008136048913002014, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.0005, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.1245421245421245, | |
| "grad_norm": 0.00842053908854723, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.0004, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 2.139194139194139, | |
| "grad_norm": 0.006062902510166168, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.0004, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.1538461538461537, | |
| "grad_norm": 0.004440332762897015, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.0003, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 2.1684981684981683, | |
| "grad_norm": 0.004300225526094437, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.0003, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.183150183150183, | |
| "grad_norm": 0.004314280580729246, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.0003, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 2.197802197802198, | |
| "grad_norm": 0.004214618355035782, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.0003, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.197802197802198, | |
| "eval_loss": 0.0026424131356179714, | |
| "eval_runtime": 12.556, | |
| "eval_samples_per_second": 9.159, | |
| "eval_steps_per_second": 2.31, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.2124542124542126, | |
| "grad_norm": 0.005527649540454149, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.0003, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 2.227106227106227, | |
| "grad_norm": 0.005907935556024313, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.0003, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 2.241758241758242, | |
| "grad_norm": 0.44951891899108887, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.0037, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 2.2564102564102564, | |
| "grad_norm": 0.10671165585517883, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.0047, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 2.271062271062271, | |
| "grad_norm": 3.4973959922790527, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.1065, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 2.2857142857142856, | |
| "grad_norm": 2.766770124435425, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.0328, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 2.3003663003663, | |
| "grad_norm": 0.20955878496170044, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.0033, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 2.315018315018315, | |
| "grad_norm": 0.023399582132697105, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.0012, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 2.32967032967033, | |
| "grad_norm": 0.012029211036860943, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.0008, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 2.3443223443223444, | |
| "grad_norm": 0.006998474709689617, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.0005, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.358974358974359, | |
| "grad_norm": 0.006124721374362707, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.0004, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 2.3736263736263736, | |
| "grad_norm": 0.005482771433889866, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.0004, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 2.3882783882783882, | |
| "grad_norm": 0.0508597195148468, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.0006, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 2.402930402930403, | |
| "grad_norm": 0.005052575375884771, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.0003, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 2.4175824175824174, | |
| "grad_norm": 0.0048966361209750175, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.0003, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 2.4322344322344325, | |
| "grad_norm": 0.003910858649760485, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.0003, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 2.446886446886447, | |
| "grad_norm": 0.004500968847423792, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.0003, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 2.4615384615384617, | |
| "grad_norm": 0.00401561614125967, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.0003, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 2.4761904761904763, | |
| "grad_norm": 0.0040866234339773655, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.0003, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 2.490842490842491, | |
| "grad_norm": 0.004417790565639734, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.0002, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.5054945054945055, | |
| "grad_norm": 0.06679342687129974, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.0033, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 2.52014652014652, | |
| "grad_norm": 1.136623740196228, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.0223, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 2.5347985347985347, | |
| "grad_norm": 0.8348060250282288, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.0116, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 2.5494505494505493, | |
| "grad_norm": 0.13132958114147186, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.0019, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 2.564102564102564, | |
| "grad_norm": 0.02172706089913845, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.0011, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.578754578754579, | |
| "grad_norm": 0.01383768767118454, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.0008, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 2.5934065934065935, | |
| "grad_norm": 0.007891649380326271, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.0005, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 2.608058608058608, | |
| "grad_norm": 0.0061985403299331665, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.0004, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 2.6227106227106227, | |
| "grad_norm": 0.005118181928992271, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.0004, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 2.6373626373626373, | |
| "grad_norm": 0.004196122754365206, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.0003, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.652014652014652, | |
| "grad_norm": 0.004347350914031267, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.0003, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 0.0034958310425281525, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.0003, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 2.6813186813186816, | |
| "grad_norm": 0.003480622312054038, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.0003, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 2.695970695970696, | |
| "grad_norm": 0.004568925127387047, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.0003, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 2.7106227106227108, | |
| "grad_norm": 0.0038007262628525496, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.0003, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 2.7252747252747254, | |
| "grad_norm": 0.004476861096918583, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.0003, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 2.73992673992674, | |
| "grad_norm": 0.0037991313729435205, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.0002, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.7545787545787546, | |
| "grad_norm": 0.05690116435289383, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.003, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 2.769230769230769, | |
| "grad_norm": 0.13506053388118744, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.0066, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 2.7838827838827838, | |
| "grad_norm": 0.20625267922878265, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.0039, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.7985347985347984, | |
| "grad_norm": 0.01528448611497879, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.001, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 2.813186813186813, | |
| "grad_norm": 0.01365617848932743, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.0009, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 2.8278388278388276, | |
| "grad_norm": 0.008216440677642822, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.0006, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 2.8424908424908426, | |
| "grad_norm": 0.007709157653152943, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.0004, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 2.857142857142857, | |
| "grad_norm": 0.005566953215748072, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.0004, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 2.871794871794872, | |
| "grad_norm": 0.004489661194384098, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.0003, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 2.8864468864468864, | |
| "grad_norm": 0.003931612707674503, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.0003, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 2.901098901098901, | |
| "grad_norm": 0.003706548595800996, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.0003, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 2.9157509157509156, | |
| "grad_norm": 0.003506936365738511, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.0003, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 2.9304029304029307, | |
| "grad_norm": 0.003877169918268919, | |
| "learning_rate": 0.0, | |
| "loss": 0.0003, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.9304029304029307, | |
| "eval_loss": 0.0004729398642666638, | |
| "eval_runtime": 12.5603, | |
| "eval_samples_per_second": 9.156, | |
| "eval_steps_per_second": 2.309, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4.6316661491367936e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |