| { | |
| "best_metric": 0.365750253200531, | |
| "best_model_checkpoint": "speecht5_finetuned_voxpopuli_nl\\checkpoint-7820", | |
| "epoch": 94.0, | |
| "eval_steps": 1, | |
| "global_step": 8648, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.5787184834480286, | |
| "eval_runtime": 15.2264, | |
| "eval_samples_per_second": 85.969, | |
| "eval_steps_per_second": 5.385, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.4981578290462494, | |
| "eval_runtime": 16.7099, | |
| "eval_samples_per_second": 78.337, | |
| "eval_steps_per_second": 4.907, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.41747257113456726, | |
| "eval_runtime": 17.157, | |
| "eval_samples_per_second": 76.295, | |
| "eval_steps_per_second": 4.779, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.4034822881221771, | |
| "eval_runtime": 17.4838, | |
| "eval_samples_per_second": 74.869, | |
| "eval_steps_per_second": 4.69, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 0.39801275730133057, | |
| "eval_runtime": 17.172, | |
| "eval_samples_per_second": 76.229, | |
| "eval_steps_per_second": 4.775, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_loss": 0.3948397636413574, | |
| "eval_runtime": 17.324, | |
| "eval_samples_per_second": 75.56, | |
| "eval_steps_per_second": 4.733, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_loss": 0.3893665075302124, | |
| "eval_runtime": 17.263, | |
| "eval_samples_per_second": 75.827, | |
| "eval_steps_per_second": 4.75, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_loss": 0.387258380651474, | |
| "eval_runtime": 17.4411, | |
| "eval_samples_per_second": 75.053, | |
| "eval_steps_per_second": 4.702, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_loss": 0.3847057819366455, | |
| "eval_runtime": 17.0524, | |
| "eval_samples_per_second": 76.763, | |
| "eval_steps_per_second": 4.809, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_loss": 0.38324064016342163, | |
| "eval_runtime": 17.1847, | |
| "eval_samples_per_second": 76.172, | |
| "eval_steps_per_second": 4.772, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 10.869565217391305, | |
| "grad_norm": 1.4818129539489746, | |
| "learning_rate": 9.426436781609197e-06, | |
| "loss": 0.4827, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_loss": 0.38332971930503845, | |
| "eval_runtime": 17.343, | |
| "eval_samples_per_second": 75.477, | |
| "eval_steps_per_second": 4.728, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_loss": 0.38135436177253723, | |
| "eval_runtime": 17.0112, | |
| "eval_samples_per_second": 76.949, | |
| "eval_steps_per_second": 4.82, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_loss": 0.37950488924980164, | |
| "eval_runtime": 17.023, | |
| "eval_samples_per_second": 76.896, | |
| "eval_steps_per_second": 4.817, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_loss": 0.3787429630756378, | |
| "eval_runtime": 17.033, | |
| "eval_samples_per_second": 76.851, | |
| "eval_steps_per_second": 4.814, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_loss": 0.37846681475639343, | |
| "eval_runtime": 16.8405, | |
| "eval_samples_per_second": 77.729, | |
| "eval_steps_per_second": 4.869, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_loss": 0.37663570046424866, | |
| "eval_runtime": 17.1094, | |
| "eval_samples_per_second": 76.508, | |
| "eval_steps_per_second": 4.793, | |
| "step": 1472 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_loss": 0.37665536999702454, | |
| "eval_runtime": 17.1693, | |
| "eval_samples_per_second": 76.241, | |
| "eval_steps_per_second": 4.776, | |
| "step": 1564 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_loss": 0.37617385387420654, | |
| "eval_runtime": 17.035, | |
| "eval_samples_per_second": 76.842, | |
| "eval_steps_per_second": 4.814, | |
| "step": 1656 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_loss": 0.3764387369155884, | |
| "eval_runtime": 17.178, | |
| "eval_samples_per_second": 76.202, | |
| "eval_steps_per_second": 4.774, | |
| "step": 1748 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_loss": 0.37478694319725037, | |
| "eval_runtime": 17.2693, | |
| "eval_samples_per_second": 75.799, | |
| "eval_steps_per_second": 4.748, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_loss": 0.3745799660682678, | |
| "eval_runtime": 17.0415, | |
| "eval_samples_per_second": 76.813, | |
| "eval_steps_per_second": 4.812, | |
| "step": 1932 | |
| }, | |
| { | |
| "epoch": 21.73913043478261, | |
| "grad_norm": 0.6598495244979858, | |
| "learning_rate": 8.277011494252875e-06, | |
| "loss": 0.4132, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_loss": 0.3734322190284729, | |
| "eval_runtime": 17.0828, | |
| "eval_samples_per_second": 76.627, | |
| "eval_steps_per_second": 4.8, | |
| "step": 2024 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_loss": 0.3731089234352112, | |
| "eval_runtime": 17.1544, | |
| "eval_samples_per_second": 76.307, | |
| "eval_steps_per_second": 4.78, | |
| "step": 2116 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_loss": 0.3726828992366791, | |
| "eval_runtime": 19.2205, | |
| "eval_samples_per_second": 68.104, | |
| "eval_steps_per_second": 4.266, | |
| "step": 2208 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_loss": 0.3719342052936554, | |
| "eval_runtime": 17.0324, | |
| "eval_samples_per_second": 76.854, | |
| "eval_steps_per_second": 4.814, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_loss": 0.3717547655105591, | |
| "eval_runtime": 16.6796, | |
| "eval_samples_per_second": 78.479, | |
| "eval_steps_per_second": 4.916, | |
| "step": 2392 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_loss": 0.3717862665653229, | |
| "eval_runtime": 17.605, | |
| "eval_samples_per_second": 74.354, | |
| "eval_steps_per_second": 4.658, | |
| "step": 2484 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_loss": 0.371040940284729, | |
| "eval_runtime": 17.658, | |
| "eval_samples_per_second": 74.131, | |
| "eval_steps_per_second": 4.644, | |
| "step": 2576 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_loss": 0.37053394317626953, | |
| "eval_runtime": 17.788, | |
| "eval_samples_per_second": 73.589, | |
| "eval_steps_per_second": 4.61, | |
| "step": 2668 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_loss": 0.3712168335914612, | |
| "eval_runtime": 18.116, | |
| "eval_samples_per_second": 72.257, | |
| "eval_steps_per_second": 4.526, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_loss": 0.37051597237586975, | |
| "eval_runtime": 18.001, | |
| "eval_samples_per_second": 72.718, | |
| "eval_steps_per_second": 4.555, | |
| "step": 2852 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_loss": 0.37035897374153137, | |
| "eval_runtime": 17.09, | |
| "eval_samples_per_second": 76.595, | |
| "eval_steps_per_second": 4.798, | |
| "step": 2944 | |
| }, | |
| { | |
| "epoch": 32.608695652173914, | |
| "grad_norm": 0.655127227306366, | |
| "learning_rate": 7.127586206896552e-06, | |
| "loss": 0.404, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_loss": 0.370259553194046, | |
| "eval_runtime": 17.957, | |
| "eval_samples_per_second": 72.896, | |
| "eval_steps_per_second": 4.566, | |
| "step": 3036 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_loss": 0.37022680044174194, | |
| "eval_runtime": 17.495, | |
| "eval_samples_per_second": 74.821, | |
| "eval_steps_per_second": 4.687, | |
| "step": 3128 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_loss": 0.36998531222343445, | |
| "eval_runtime": 17.76, | |
| "eval_samples_per_second": 73.705, | |
| "eval_steps_per_second": 4.617, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_loss": 0.36924660205841064, | |
| "eval_runtime": 17.075, | |
| "eval_samples_per_second": 76.662, | |
| "eval_steps_per_second": 4.802, | |
| "step": 3312 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_loss": 0.3693754971027374, | |
| "eval_runtime": 17.1777, | |
| "eval_samples_per_second": 76.204, | |
| "eval_steps_per_second": 4.774, | |
| "step": 3404 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_loss": 0.3692266047000885, | |
| "eval_runtime": 17.4136, | |
| "eval_samples_per_second": 75.171, | |
| "eval_steps_per_second": 4.709, | |
| "step": 3496 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_loss": 0.36862194538116455, | |
| "eval_runtime": 17.1187, | |
| "eval_samples_per_second": 76.466, | |
| "eval_steps_per_second": 4.79, | |
| "step": 3588 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_loss": 0.3685244917869568, | |
| "eval_runtime": 17.1157, | |
| "eval_samples_per_second": 76.48, | |
| "eval_steps_per_second": 4.791, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_loss": 0.36859795451164246, | |
| "eval_runtime": 17.1462, | |
| "eval_samples_per_second": 76.344, | |
| "eval_steps_per_second": 4.782, | |
| "step": 3772 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_loss": 0.3684174120426178, | |
| "eval_runtime": 17.1238, | |
| "eval_samples_per_second": 76.443, | |
| "eval_steps_per_second": 4.789, | |
| "step": 3864 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_loss": 0.3683168888092041, | |
| "eval_runtime": 17.423, | |
| "eval_samples_per_second": 75.131, | |
| "eval_steps_per_second": 4.706, | |
| "step": 3956 | |
| }, | |
| { | |
| "epoch": 43.47826086956522, | |
| "grad_norm": 0.7368005514144897, | |
| "learning_rate": 5.97816091954023e-06, | |
| "loss": 0.3985, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_loss": 0.3692004084587097, | |
| "eval_runtime": 17.3077, | |
| "eval_samples_per_second": 75.631, | |
| "eval_steps_per_second": 4.738, | |
| "step": 4048 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_loss": 0.368026465177536, | |
| "eval_runtime": 16.9027, | |
| "eval_samples_per_second": 77.443, | |
| "eval_steps_per_second": 4.851, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_loss": 0.3682689964771271, | |
| "eval_runtime": 17.0307, | |
| "eval_samples_per_second": 76.861, | |
| "eval_steps_per_second": 4.815, | |
| "step": 4232 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_loss": 0.3681689500808716, | |
| "eval_runtime": 17.0938, | |
| "eval_samples_per_second": 76.577, | |
| "eval_steps_per_second": 4.797, | |
| "step": 4324 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_loss": 0.36807677149772644, | |
| "eval_runtime": 17.1767, | |
| "eval_samples_per_second": 76.208, | |
| "eval_steps_per_second": 4.774, | |
| "step": 4416 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_loss": 0.36805370450019836, | |
| "eval_runtime": 17.2818, | |
| "eval_samples_per_second": 75.744, | |
| "eval_steps_per_second": 4.745, | |
| "step": 4508 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_loss": 0.367729514837265, | |
| "eval_runtime": 17.47, | |
| "eval_samples_per_second": 74.928, | |
| "eval_steps_per_second": 4.694, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_loss": 0.3675946891307831, | |
| "eval_runtime": 17.5358, | |
| "eval_samples_per_second": 74.647, | |
| "eval_steps_per_second": 4.676, | |
| "step": 4692 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_loss": 0.3678540885448456, | |
| "eval_runtime": 18.1335, | |
| "eval_samples_per_second": 72.187, | |
| "eval_steps_per_second": 4.522, | |
| "step": 4784 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_loss": 0.36748984456062317, | |
| "eval_runtime": 17.4305, | |
| "eval_samples_per_second": 75.098, | |
| "eval_steps_per_second": 4.704, | |
| "step": 4876 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_loss": 0.3672032058238983, | |
| "eval_runtime": 17.1008, | |
| "eval_samples_per_second": 76.546, | |
| "eval_steps_per_second": 4.795, | |
| "step": 4968 | |
| }, | |
| { | |
| "epoch": 54.34782608695652, | |
| "grad_norm": 0.8721207976341248, | |
| "learning_rate": 4.828735632183908e-06, | |
| "loss": 0.3956, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_loss": 0.3671948313713074, | |
| "eval_runtime": 16.9098, | |
| "eval_samples_per_second": 77.411, | |
| "eval_steps_per_second": 4.849, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_loss": 0.3672229051589966, | |
| "eval_runtime": 16.6549, | |
| "eval_samples_per_second": 78.596, | |
| "eval_steps_per_second": 4.923, | |
| "step": 5152 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_loss": 0.3669443428516388, | |
| "eval_runtime": 17.9423, | |
| "eval_samples_per_second": 72.956, | |
| "eval_steps_per_second": 4.57, | |
| "step": 5244 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_loss": 0.3672654926776886, | |
| "eval_runtime": 17.494, | |
| "eval_samples_per_second": 74.826, | |
| "eval_steps_per_second": 4.687, | |
| "step": 5336 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_loss": 0.36697065830230713, | |
| "eval_runtime": 17.4709, | |
| "eval_samples_per_second": 74.925, | |
| "eval_steps_per_second": 4.694, | |
| "step": 5428 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_loss": 0.36780840158462524, | |
| "eval_runtime": 17.9242, | |
| "eval_samples_per_second": 73.03, | |
| "eval_steps_per_second": 4.575, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_loss": 0.3665044903755188, | |
| "eval_runtime": 18.0071, | |
| "eval_samples_per_second": 72.693, | |
| "eval_steps_per_second": 4.554, | |
| "step": 5612 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_loss": 0.36677271127700806, | |
| "eval_runtime": 17.9079, | |
| "eval_samples_per_second": 73.096, | |
| "eval_steps_per_second": 4.579, | |
| "step": 5704 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_loss": 0.3672313094139099, | |
| "eval_runtime": 17.739, | |
| "eval_samples_per_second": 73.792, | |
| "eval_steps_per_second": 4.623, | |
| "step": 5796 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_loss": 0.36684995889663696, | |
| "eval_runtime": 17.7169, | |
| "eval_samples_per_second": 73.884, | |
| "eval_steps_per_second": 4.628, | |
| "step": 5888 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_loss": 0.3665069043636322, | |
| "eval_runtime": 17.7005, | |
| "eval_samples_per_second": 73.953, | |
| "eval_steps_per_second": 4.633, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 65.21739130434783, | |
| "grad_norm": 0.6173653602600098, | |
| "learning_rate": 3.680459770114943e-06, | |
| "loss": 0.3932, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_loss": 0.3668159246444702, | |
| "eval_runtime": 17.8325, | |
| "eval_samples_per_second": 73.405, | |
| "eval_steps_per_second": 4.598, | |
| "step": 6072 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_loss": 0.3664255738258362, | |
| "eval_runtime": 17.6879, | |
| "eval_samples_per_second": 74.005, | |
| "eval_steps_per_second": 4.636, | |
| "step": 6164 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_loss": 0.3663348853588104, | |
| "eval_runtime": 17.6052, | |
| "eval_samples_per_second": 74.353, | |
| "eval_steps_per_second": 4.658, | |
| "step": 6256 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_loss": 0.3667348623275757, | |
| "eval_runtime": 17.6021, | |
| "eval_samples_per_second": 74.366, | |
| "eval_steps_per_second": 4.659, | |
| "step": 6348 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_loss": 0.36681902408599854, | |
| "eval_runtime": 16.9539, | |
| "eval_samples_per_second": 77.209, | |
| "eval_steps_per_second": 4.837, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_loss": 0.3668743371963501, | |
| "eval_runtime": 18.6549, | |
| "eval_samples_per_second": 70.169, | |
| "eval_steps_per_second": 4.396, | |
| "step": 6532 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_loss": 0.3665105104446411, | |
| "eval_runtime": 17.1969, | |
| "eval_samples_per_second": 76.118, | |
| "eval_steps_per_second": 4.768, | |
| "step": 6624 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_loss": 0.36701568961143494, | |
| "eval_runtime": 17.4949, | |
| "eval_samples_per_second": 74.822, | |
| "eval_steps_per_second": 4.687, | |
| "step": 6716 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_loss": 0.36646127700805664, | |
| "eval_runtime": 17.2609, | |
| "eval_samples_per_second": 75.836, | |
| "eval_steps_per_second": 4.751, | |
| "step": 6808 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_loss": 0.3664456307888031, | |
| "eval_runtime": 17.4389, | |
| "eval_samples_per_second": 75.062, | |
| "eval_steps_per_second": 4.702, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_loss": 0.3668871819972992, | |
| "eval_runtime": 17.2663, | |
| "eval_samples_per_second": 75.813, | |
| "eval_steps_per_second": 4.749, | |
| "step": 6992 | |
| }, | |
| { | |
| "epoch": 76.08695652173913, | |
| "grad_norm": 0.6145053505897522, | |
| "learning_rate": 2.531034482758621e-06, | |
| "loss": 0.3915, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_loss": 0.366481214761734, | |
| "eval_runtime": 16.9739, | |
| "eval_samples_per_second": 77.118, | |
| "eval_steps_per_second": 4.831, | |
| "step": 7084 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_loss": 0.36652714014053345, | |
| "eval_runtime": 17.1419, | |
| "eval_samples_per_second": 76.362, | |
| "eval_steps_per_second": 4.784, | |
| "step": 7176 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_loss": 0.36669179797172546, | |
| "eval_runtime": 17.1379, | |
| "eval_samples_per_second": 76.38, | |
| "eval_steps_per_second": 4.785, | |
| "step": 7268 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_loss": 0.36666759848594666, | |
| "eval_runtime": 17.1669, | |
| "eval_samples_per_second": 76.251, | |
| "eval_steps_per_second": 4.777, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 81.0, | |
| "eval_loss": 0.36649754643440247, | |
| "eval_runtime": 17.2209, | |
| "eval_samples_per_second": 76.012, | |
| "eval_steps_per_second": 4.762, | |
| "step": 7452 | |
| }, | |
| { | |
| "epoch": 82.0, | |
| "eval_loss": 0.36655521392822266, | |
| "eval_runtime": 17.2849, | |
| "eval_samples_per_second": 75.731, | |
| "eval_steps_per_second": 4.744, | |
| "step": 7544 | |
| }, | |
| { | |
| "epoch": 83.0, | |
| "eval_loss": 0.3659179210662842, | |
| "eval_runtime": 16.9279, | |
| "eval_samples_per_second": 77.328, | |
| "eval_steps_per_second": 4.844, | |
| "step": 7636 | |
| }, | |
| { | |
| "epoch": 84.0, | |
| "eval_loss": 0.3663187325000763, | |
| "eval_runtime": 17.0312, | |
| "eval_samples_per_second": 76.859, | |
| "eval_steps_per_second": 4.815, | |
| "step": 7728 | |
| }, | |
| { | |
| "epoch": 85.0, | |
| "eval_loss": 0.365750253200531, | |
| "eval_runtime": 17.8129, | |
| "eval_samples_per_second": 73.486, | |
| "eval_steps_per_second": 4.603, | |
| "step": 7820 | |
| }, | |
| { | |
| "epoch": 86.0, | |
| "eval_loss": 0.3665299415588379, | |
| "eval_runtime": 17.9302, | |
| "eval_samples_per_second": 73.006, | |
| "eval_steps_per_second": 4.573, | |
| "step": 7912 | |
| }, | |
| { | |
| "epoch": 86.95652173913044, | |
| "grad_norm": 0.615623950958252, | |
| "learning_rate": 1.381609195402299e-06, | |
| "loss": 0.3899, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 87.0, | |
| "eval_loss": 0.36646467447280884, | |
| "eval_runtime": 17.8689, | |
| "eval_samples_per_second": 73.256, | |
| "eval_steps_per_second": 4.589, | |
| "step": 8004 | |
| }, | |
| { | |
| "epoch": 88.0, | |
| "eval_loss": 0.3665202856063843, | |
| "eval_runtime": 17.899, | |
| "eval_samples_per_second": 73.133, | |
| "eval_steps_per_second": 4.581, | |
| "step": 8096 | |
| }, | |
| { | |
| "epoch": 89.0, | |
| "eval_loss": 0.36632445454597473, | |
| "eval_runtime": 17.9289, | |
| "eval_samples_per_second": 73.01, | |
| "eval_steps_per_second": 4.574, | |
| "step": 8188 | |
| }, | |
| { | |
| "epoch": 90.0, | |
| "eval_loss": 0.36636534333229065, | |
| "eval_runtime": 15.092, | |
| "eval_samples_per_second": 86.735, | |
| "eval_steps_per_second": 5.433, | |
| "step": 8280 | |
| }, | |
| { | |
| "epoch": 91.0, | |
| "eval_loss": 0.3663894832134247, | |
| "eval_runtime": 17.2323, | |
| "eval_samples_per_second": 75.962, | |
| "eval_steps_per_second": 4.759, | |
| "step": 8372 | |
| }, | |
| { | |
| "epoch": 92.0, | |
| "eval_loss": 0.36603686213493347, | |
| "eval_runtime": 18.3225, | |
| "eval_samples_per_second": 71.442, | |
| "eval_steps_per_second": 4.475, | |
| "step": 8464 | |
| }, | |
| { | |
| "epoch": 93.0, | |
| "eval_loss": 0.36650174856185913, | |
| "eval_runtime": 17.8928, | |
| "eval_samples_per_second": 73.158, | |
| "eval_steps_per_second": 4.583, | |
| "step": 8556 | |
| }, | |
| { | |
| "epoch": 94.0, | |
| "eval_loss": 0.3667086660861969, | |
| "eval_runtime": 17.2053, | |
| "eval_samples_per_second": 76.081, | |
| "eval_steps_per_second": 4.766, | |
| "step": 8648 | |
| } | |
| ], | |
| "logging_steps": 1000, | |
| "max_steps": 9200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 100, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.5485139128280413e+17, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |