| { | |
| "best_metric": 0.8324354290962219, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.013036534889026497, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 6.518267444513249e-05, | |
| "grad_norm": 4.084769248962402, | |
| "learning_rate": 1.004e-05, | |
| "loss": 1.0714, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 6.518267444513249e-05, | |
| "eval_loss": 3.5102760791778564, | |
| "eval_runtime": 744.1991, | |
| "eval_samples_per_second": 8.68, | |
| "eval_steps_per_second": 2.17, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00013036534889026498, | |
| "grad_norm": 5.269199371337891, | |
| "learning_rate": 2.008e-05, | |
| "loss": 1.5721, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00019554802333539745, | |
| "grad_norm": 5.509669303894043, | |
| "learning_rate": 3.012e-05, | |
| "loss": 1.0653, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00026073069778052995, | |
| "grad_norm": 5.992783546447754, | |
| "learning_rate": 4.016e-05, | |
| "loss": 1.1367, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0003259133722256624, | |
| "grad_norm": 5.95001745223999, | |
| "learning_rate": 5.02e-05, | |
| "loss": 1.4664, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0003910960466707949, | |
| "grad_norm": 6.181279182434082, | |
| "learning_rate": 6.024e-05, | |
| "loss": 1.5197, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0004562787211159274, | |
| "grad_norm": 5.237460613250732, | |
| "learning_rate": 7.028e-05, | |
| "loss": 1.4659, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0005214613955610599, | |
| "grad_norm": 7.470530033111572, | |
| "learning_rate": 8.032e-05, | |
| "loss": 2.5472, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0005866440700061923, | |
| "grad_norm": 7.4010443687438965, | |
| "learning_rate": 9.036000000000001e-05, | |
| "loss": 1.8851, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0006518267444513248, | |
| "grad_norm": 8.14273452758789, | |
| "learning_rate": 0.0001004, | |
| "loss": 2.5444, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0007170094188964573, | |
| "grad_norm": 5.8861165046691895, | |
| "learning_rate": 9.987157894736842e-05, | |
| "loss": 1.9468, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0007821920933415898, | |
| "grad_norm": 5.406398296356201, | |
| "learning_rate": 9.934315789473684e-05, | |
| "loss": 1.0875, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0008473747677867222, | |
| "grad_norm": 6.182877540588379, | |
| "learning_rate": 9.881473684210525e-05, | |
| "loss": 1.5979, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0009125574422318548, | |
| "grad_norm": 8.538721084594727, | |
| "learning_rate": 9.828631578947369e-05, | |
| "loss": 1.8065, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0009777401166769872, | |
| "grad_norm": 5.54311466217041, | |
| "learning_rate": 9.77578947368421e-05, | |
| "loss": 1.0368, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0010429227911221198, | |
| "grad_norm": 11.207380294799805, | |
| "learning_rate": 9.722947368421052e-05, | |
| "loss": 1.5687, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0011081054655672521, | |
| "grad_norm": 6.547550201416016, | |
| "learning_rate": 9.670105263157895e-05, | |
| "loss": 1.0427, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0011732881400123847, | |
| "grad_norm": 5.821341037750244, | |
| "learning_rate": 9.617263157894737e-05, | |
| "loss": 1.2702, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0012384708144575172, | |
| "grad_norm": 8.170310974121094, | |
| "learning_rate": 9.564421052631579e-05, | |
| "loss": 1.4842, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0013036534889026496, | |
| "grad_norm": 7.119532585144043, | |
| "learning_rate": 9.511578947368421e-05, | |
| "loss": 1.2944, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0013688361633477821, | |
| "grad_norm": 6.549607276916504, | |
| "learning_rate": 9.458736842105264e-05, | |
| "loss": 1.1745, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0014340188377929147, | |
| "grad_norm": 5.023319244384766, | |
| "learning_rate": 9.405894736842106e-05, | |
| "loss": 0.7572, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.001499201512238047, | |
| "grad_norm": 6.467738628387451, | |
| "learning_rate": 9.353052631578947e-05, | |
| "loss": 1.0398, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0015643841866831796, | |
| "grad_norm": 5.155070781707764, | |
| "learning_rate": 9.300210526315789e-05, | |
| "loss": 0.6913, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0016295668611283121, | |
| "grad_norm": 5.410099506378174, | |
| "learning_rate": 9.247368421052631e-05, | |
| "loss": 1.116, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0016947495355734445, | |
| "grad_norm": 9.872712135314941, | |
| "learning_rate": 9.194526315789473e-05, | |
| "loss": 1.8271, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.001759932210018577, | |
| "grad_norm": 8.195897102355957, | |
| "learning_rate": 9.141684210526316e-05, | |
| "loss": 1.082, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0018251148844637096, | |
| "grad_norm": 7.203644275665283, | |
| "learning_rate": 9.088842105263158e-05, | |
| "loss": 1.1428, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.001890297558908842, | |
| "grad_norm": 9.383421897888184, | |
| "learning_rate": 9.036000000000001e-05, | |
| "loss": 0.9886, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0019554802333539745, | |
| "grad_norm": 4.832867622375488, | |
| "learning_rate": 8.983157894736843e-05, | |
| "loss": 1.019, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.002020662907799107, | |
| "grad_norm": 8.182337760925293, | |
| "learning_rate": 8.930315789473684e-05, | |
| "loss": 1.0352, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0020858455822442396, | |
| "grad_norm": 3.0797736644744873, | |
| "learning_rate": 8.877473684210526e-05, | |
| "loss": 0.6499, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.002151028256689372, | |
| "grad_norm": 4.614433765411377, | |
| "learning_rate": 8.824631578947368e-05, | |
| "loss": 1.3117, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0022162109311345043, | |
| "grad_norm": 4.8648457527160645, | |
| "learning_rate": 8.771789473684211e-05, | |
| "loss": 0.9818, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.002281393605579637, | |
| "grad_norm": 5.053056716918945, | |
| "learning_rate": 8.718947368421053e-05, | |
| "loss": 1.0728, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0023465762800247694, | |
| "grad_norm": 5.258237838745117, | |
| "learning_rate": 8.666105263157895e-05, | |
| "loss": 1.3109, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.002411758954469902, | |
| "grad_norm": 6.49020528793335, | |
| "learning_rate": 8.613263157894737e-05, | |
| "loss": 1.5304, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0024769416289150345, | |
| "grad_norm": 5.650185585021973, | |
| "learning_rate": 8.560421052631578e-05, | |
| "loss": 1.5754, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.002542124303360167, | |
| "grad_norm": 4.012632846832275, | |
| "learning_rate": 8.50757894736842e-05, | |
| "loss": 1.0375, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.002607306977805299, | |
| "grad_norm": 4.553121089935303, | |
| "learning_rate": 8.454736842105263e-05, | |
| "loss": 1.0791, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0026724896522504317, | |
| "grad_norm": 4.67976188659668, | |
| "learning_rate": 8.401894736842106e-05, | |
| "loss": 0.9995, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0027376723266955643, | |
| "grad_norm": 5.818678855895996, | |
| "learning_rate": 8.349052631578948e-05, | |
| "loss": 0.9256, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.002802855001140697, | |
| "grad_norm": 5.402655601501465, | |
| "learning_rate": 8.29621052631579e-05, | |
| "loss": 1.1302, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0028680376755858294, | |
| "grad_norm": 10.834538459777832, | |
| "learning_rate": 8.243368421052632e-05, | |
| "loss": 0.818, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.002933220350030962, | |
| "grad_norm": 4.870290279388428, | |
| "learning_rate": 8.190526315789474e-05, | |
| "loss": 0.9021, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.002998403024476094, | |
| "grad_norm": 4.933202266693115, | |
| "learning_rate": 8.137684210526315e-05, | |
| "loss": 0.801, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0030635856989212266, | |
| "grad_norm": 6.761687278747559, | |
| "learning_rate": 8.084842105263157e-05, | |
| "loss": 1.3199, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.003128768373366359, | |
| "grad_norm": 6.689528465270996, | |
| "learning_rate": 8.032e-05, | |
| "loss": 1.3152, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0031939510478114917, | |
| "grad_norm": 5.506531238555908, | |
| "learning_rate": 7.979157894736842e-05, | |
| "loss": 1.2228, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0032591337222566243, | |
| "grad_norm": 5.423097133636475, | |
| "learning_rate": 7.926315789473684e-05, | |
| "loss": 1.4467, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0032591337222566243, | |
| "eval_loss": 0.9954817891120911, | |
| "eval_runtime": 745.98, | |
| "eval_samples_per_second": 8.66, | |
| "eval_steps_per_second": 2.165, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.003324316396701757, | |
| "grad_norm": 3.4636971950531006, | |
| "learning_rate": 7.873473684210526e-05, | |
| "loss": 0.6352, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.003389499071146889, | |
| "grad_norm": 3.1654837131500244, | |
| "learning_rate": 7.820631578947369e-05, | |
| "loss": 0.8974, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0034546817455920215, | |
| "grad_norm": 2.099487781524658, | |
| "learning_rate": 7.76778947368421e-05, | |
| "loss": 0.2316, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.003519864420037154, | |
| "grad_norm": 1.574611783027649, | |
| "learning_rate": 7.714947368421052e-05, | |
| "loss": 0.2663, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0035850470944822866, | |
| "grad_norm": 2.4194858074188232, | |
| "learning_rate": 7.662105263157896e-05, | |
| "loss": 0.491, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.003650229768927419, | |
| "grad_norm": 1.953523874282837, | |
| "learning_rate": 7.609263157894737e-05, | |
| "loss": 0.449, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0037154124433725517, | |
| "grad_norm": 2.2616214752197266, | |
| "learning_rate": 7.556421052631579e-05, | |
| "loss": 0.3948, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.003780595117817684, | |
| "grad_norm": 2.246051788330078, | |
| "learning_rate": 7.503578947368421e-05, | |
| "loss": 0.5058, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0038457777922628164, | |
| "grad_norm": 2.9115405082702637, | |
| "learning_rate": 7.450736842105263e-05, | |
| "loss": 0.7811, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.003910960466707949, | |
| "grad_norm": 4.887603759765625, | |
| "learning_rate": 7.397894736842105e-05, | |
| "loss": 0.7756, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.003976143141153081, | |
| "grad_norm": 2.9186997413635254, | |
| "learning_rate": 7.345052631578948e-05, | |
| "loss": 0.4695, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.004041325815598214, | |
| "grad_norm": 4.976742267608643, | |
| "learning_rate": 7.29221052631579e-05, | |
| "loss": 0.868, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.004106508490043346, | |
| "grad_norm": 4.973067760467529, | |
| "learning_rate": 7.239368421052631e-05, | |
| "loss": 1.181, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.004171691164488479, | |
| "grad_norm": 5.87078857421875, | |
| "learning_rate": 7.186526315789474e-05, | |
| "loss": 1.1077, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.004236873838933611, | |
| "grad_norm": 3.100754737854004, | |
| "learning_rate": 7.133684210526316e-05, | |
| "loss": 0.5465, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.004302056513378744, | |
| "grad_norm": 6.342639446258545, | |
| "learning_rate": 7.080842105263158e-05, | |
| "loss": 1.0568, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0043672391878238764, | |
| "grad_norm": 3.8607728481292725, | |
| "learning_rate": 7.028e-05, | |
| "loss": 1.054, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0044324218622690086, | |
| "grad_norm": 4.340756893157959, | |
| "learning_rate": 6.975157894736843e-05, | |
| "loss": 1.2158, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0044976045367141415, | |
| "grad_norm": 2.7851240634918213, | |
| "learning_rate": 6.922315789473685e-05, | |
| "loss": 0.6975, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.004562787211159274, | |
| "grad_norm": 19.89763832092285, | |
| "learning_rate": 6.869473684210527e-05, | |
| "loss": 1.2703, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.004627969885604407, | |
| "grad_norm": 5.404323101043701, | |
| "learning_rate": 6.816631578947368e-05, | |
| "loss": 0.9856, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.004693152560049539, | |
| "grad_norm": 5.077646255493164, | |
| "learning_rate": 6.76378947368421e-05, | |
| "loss": 1.2272, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.004758335234494671, | |
| "grad_norm": 11.702156066894531, | |
| "learning_rate": 6.710947368421052e-05, | |
| "loss": 1.2462, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.004823517908939804, | |
| "grad_norm": 1.9628050327301025, | |
| "learning_rate": 6.658105263157894e-05, | |
| "loss": 0.5371, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.004888700583384936, | |
| "grad_norm": 4.088584899902344, | |
| "learning_rate": 6.605263157894737e-05, | |
| "loss": 0.9884, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.004953883257830069, | |
| "grad_norm": 4.946320533752441, | |
| "learning_rate": 6.55242105263158e-05, | |
| "loss": 1.036, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.005019065932275201, | |
| "grad_norm": 3.6775155067443848, | |
| "learning_rate": 6.499578947368422e-05, | |
| "loss": 1.1745, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.005084248606720334, | |
| "grad_norm": 3.1447501182556152, | |
| "learning_rate": 6.446736842105264e-05, | |
| "loss": 0.7307, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.005149431281165466, | |
| "grad_norm": 3.2460031509399414, | |
| "learning_rate": 6.393894736842105e-05, | |
| "loss": 0.9647, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.005214613955610598, | |
| "grad_norm": 7.355772972106934, | |
| "learning_rate": 6.341052631578947e-05, | |
| "loss": 1.0611, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.005279796630055731, | |
| "grad_norm": 6.00462007522583, | |
| "learning_rate": 6.288210526315789e-05, | |
| "loss": 1.0856, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.0053449793045008635, | |
| "grad_norm": 4.158636569976807, | |
| "learning_rate": 6.235368421052632e-05, | |
| "loss": 0.9878, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0054101619789459965, | |
| "grad_norm": 4.67047643661499, | |
| "learning_rate": 6.182526315789474e-05, | |
| "loss": 1.1027, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.005475344653391129, | |
| "grad_norm": 4.211904525756836, | |
| "learning_rate": 6.129684210526316e-05, | |
| "loss": 0.9119, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.005540527327836261, | |
| "grad_norm": 4.3183064460754395, | |
| "learning_rate": 6.076842105263158e-05, | |
| "loss": 0.991, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.005605710002281394, | |
| "grad_norm": 3.8753435611724854, | |
| "learning_rate": 6.024e-05, | |
| "loss": 0.9348, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.005670892676726526, | |
| "grad_norm": 5.450557231903076, | |
| "learning_rate": 5.971157894736842e-05, | |
| "loss": 1.4856, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.005736075351171659, | |
| "grad_norm": 5.696898460388184, | |
| "learning_rate": 5.9183157894736835e-05, | |
| "loss": 1.1666, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.005801258025616791, | |
| "grad_norm": 3.665428876876831, | |
| "learning_rate": 5.8654736842105267e-05, | |
| "loss": 0.7272, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.005866440700061924, | |
| "grad_norm": 7.14247989654541, | |
| "learning_rate": 5.8126315789473684e-05, | |
| "loss": 1.0759, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.005931623374507056, | |
| "grad_norm": 4.450038909912109, | |
| "learning_rate": 5.759789473684211e-05, | |
| "loss": 1.0025, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.005996806048952188, | |
| "grad_norm": 3.9177703857421875, | |
| "learning_rate": 5.706947368421053e-05, | |
| "loss": 0.7917, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.006061988723397321, | |
| "grad_norm": 5.90571403503418, | |
| "learning_rate": 5.6541052631578945e-05, | |
| "loss": 1.0844, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.006127171397842453, | |
| "grad_norm": 7.018702983856201, | |
| "learning_rate": 5.601263157894736e-05, | |
| "loss": 1.0232, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.006192354072287586, | |
| "grad_norm": 3.4453823566436768, | |
| "learning_rate": 5.5484210526315794e-05, | |
| "loss": 0.7434, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.006257536746732718, | |
| "grad_norm": 6.936615943908691, | |
| "learning_rate": 5.495578947368421e-05, | |
| "loss": 1.3415, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.006322719421177851, | |
| "grad_norm": 11.391502380371094, | |
| "learning_rate": 5.442736842105264e-05, | |
| "loss": 1.4938, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.0063879020956229835, | |
| "grad_norm": 3.610074520111084, | |
| "learning_rate": 5.3898947368421055e-05, | |
| "loss": 0.9637, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.006453084770068116, | |
| "grad_norm": 6.011956214904785, | |
| "learning_rate": 5.337052631578947e-05, | |
| "loss": 1.492, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.006518267444513249, | |
| "grad_norm": 6.378303050994873, | |
| "learning_rate": 5.284210526315789e-05, | |
| "loss": 1.1898, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.006518267444513249, | |
| "eval_loss": 0.9247763156890869, | |
| "eval_runtime": 746.0854, | |
| "eval_samples_per_second": 8.659, | |
| "eval_steps_per_second": 2.165, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.006583450118958381, | |
| "grad_norm": 2.645019769668579, | |
| "learning_rate": 5.231368421052631e-05, | |
| "loss": 0.6677, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.006648632793403514, | |
| "grad_norm": 2.862882375717163, | |
| "learning_rate": 5.178526315789474e-05, | |
| "loss": 0.7984, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.006713815467848646, | |
| "grad_norm": 2.1093852519989014, | |
| "learning_rate": 5.1256842105263165e-05, | |
| "loss": 0.5116, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.006778998142293778, | |
| "grad_norm": 1.6436916589736938, | |
| "learning_rate": 5.072842105263158e-05, | |
| "loss": 0.3384, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.006844180816738911, | |
| "grad_norm": 1.8502347469329834, | |
| "learning_rate": 5.02e-05, | |
| "loss": 0.3001, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.006909363491184043, | |
| "grad_norm": 3.300266742706299, | |
| "learning_rate": 4.967157894736842e-05, | |
| "loss": 0.3394, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.006974546165629176, | |
| "grad_norm": 3.8908684253692627, | |
| "learning_rate": 4.914315789473684e-05, | |
| "loss": 0.8438, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.007039728840074308, | |
| "grad_norm": 2.5882935523986816, | |
| "learning_rate": 4.861473684210526e-05, | |
| "loss": 0.764, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.007104911514519441, | |
| "grad_norm": 2.4956729412078857, | |
| "learning_rate": 4.8086315789473686e-05, | |
| "loss": 0.5656, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.007170094188964573, | |
| "grad_norm": 3.178035020828247, | |
| "learning_rate": 4.7557894736842104e-05, | |
| "loss": 0.5182, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.007235276863409705, | |
| "grad_norm": 2.6953673362731934, | |
| "learning_rate": 4.702947368421053e-05, | |
| "loss": 0.7, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.007300459537854838, | |
| "grad_norm": 4.535560607910156, | |
| "learning_rate": 4.6501052631578946e-05, | |
| "loss": 0.6026, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.0073656422122999705, | |
| "grad_norm": 3.3130948543548584, | |
| "learning_rate": 4.5972631578947364e-05, | |
| "loss": 0.6538, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.0074308248867451035, | |
| "grad_norm": 5.011913776397705, | |
| "learning_rate": 4.544421052631579e-05, | |
| "loss": 1.1627, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.007496007561190236, | |
| "grad_norm": 2.643364906311035, | |
| "learning_rate": 4.4915789473684213e-05, | |
| "loss": 0.6814, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.007561190235635368, | |
| "grad_norm": 4.945113182067871, | |
| "learning_rate": 4.438736842105263e-05, | |
| "loss": 1.4463, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.007626372910080501, | |
| "grad_norm": 4.396795749664307, | |
| "learning_rate": 4.3858947368421056e-05, | |
| "loss": 1.09, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.007691555584525633, | |
| "grad_norm": 4.34941291809082, | |
| "learning_rate": 4.3330526315789474e-05, | |
| "loss": 0.6612, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.007756738258970766, | |
| "grad_norm": 4.358571529388428, | |
| "learning_rate": 4.280210526315789e-05, | |
| "loss": 1.0953, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.007821920933415898, | |
| "grad_norm": 4.529819488525391, | |
| "learning_rate": 4.2273684210526317e-05, | |
| "loss": 1.2583, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.00788710360786103, | |
| "grad_norm": 3.2337148189544678, | |
| "learning_rate": 4.174526315789474e-05, | |
| "loss": 1.2295, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.007952286282306162, | |
| "grad_norm": 3.514124631881714, | |
| "learning_rate": 4.121684210526316e-05, | |
| "loss": 0.7983, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.008017468956751296, | |
| "grad_norm": 4.527546405792236, | |
| "learning_rate": 4.068842105263158e-05, | |
| "loss": 1.1304, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.008082651631196428, | |
| "grad_norm": 3.7314906120300293, | |
| "learning_rate": 4.016e-05, | |
| "loss": 0.9276, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.00814783430564156, | |
| "grad_norm": 3.9742982387542725, | |
| "learning_rate": 3.963157894736842e-05, | |
| "loss": 1.1314, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.008213016980086692, | |
| "grad_norm": 3.210330009460449, | |
| "learning_rate": 3.9103157894736844e-05, | |
| "loss": 0.8935, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.008278199654531826, | |
| "grad_norm": 2.126352310180664, | |
| "learning_rate": 3.857473684210526e-05, | |
| "loss": 0.5475, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.008343382328976958, | |
| "grad_norm": 4.862457275390625, | |
| "learning_rate": 3.804631578947369e-05, | |
| "loss": 1.1201, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.00840856500342209, | |
| "grad_norm": 4.386620044708252, | |
| "learning_rate": 3.7517894736842105e-05, | |
| "loss": 1.243, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.008473747677867223, | |
| "grad_norm": 5.318674564361572, | |
| "learning_rate": 3.698947368421052e-05, | |
| "loss": 1.3332, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.008538930352312355, | |
| "grad_norm": 5.400907039642334, | |
| "learning_rate": 3.646105263157895e-05, | |
| "loss": 1.4768, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.008604113026757489, | |
| "grad_norm": 3.384183406829834, | |
| "learning_rate": 3.593263157894737e-05, | |
| "loss": 1.0143, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.00866929570120262, | |
| "grad_norm": 3.478654384613037, | |
| "learning_rate": 3.540421052631579e-05, | |
| "loss": 0.8405, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.008734478375647753, | |
| "grad_norm": 5.532976150512695, | |
| "learning_rate": 3.4875789473684215e-05, | |
| "loss": 1.3522, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.008799661050092885, | |
| "grad_norm": 3.775825023651123, | |
| "learning_rate": 3.434736842105263e-05, | |
| "loss": 0.9939, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.008864843724538017, | |
| "grad_norm": 4.642473220825195, | |
| "learning_rate": 3.381894736842105e-05, | |
| "loss": 0.7214, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.008930026398983151, | |
| "grad_norm": 27.070499420166016, | |
| "learning_rate": 3.329052631578947e-05, | |
| "loss": 1.0064, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.008995209073428283, | |
| "grad_norm": 4.223586559295654, | |
| "learning_rate": 3.27621052631579e-05, | |
| "loss": 1.1072, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.009060391747873415, | |
| "grad_norm": 2.780446767807007, | |
| "learning_rate": 3.223368421052632e-05, | |
| "loss": 0.7691, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.009125574422318547, | |
| "grad_norm": 3.2398338317871094, | |
| "learning_rate": 3.1705263157894736e-05, | |
| "loss": 0.7892, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.00919075709676368, | |
| "grad_norm": 2.2731664180755615, | |
| "learning_rate": 3.117684210526316e-05, | |
| "loss": 0.581, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.009255939771208813, | |
| "grad_norm": 4.545046329498291, | |
| "learning_rate": 3.064842105263158e-05, | |
| "loss": 0.9304, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.009321122445653945, | |
| "grad_norm": 3.062983512878418, | |
| "learning_rate": 3.012e-05, | |
| "loss": 0.7803, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.009386305120099078, | |
| "grad_norm": 3.362913131713867, | |
| "learning_rate": 2.9591578947368418e-05, | |
| "loss": 0.8837, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.00945148779454421, | |
| "grad_norm": 4.0667877197265625, | |
| "learning_rate": 2.9063157894736842e-05, | |
| "loss": 0.8816, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.009516670468989342, | |
| "grad_norm": 3.24399733543396, | |
| "learning_rate": 2.8534736842105264e-05, | |
| "loss": 0.6107, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.009581853143434476, | |
| "grad_norm": 4.025416374206543, | |
| "learning_rate": 2.800631578947368e-05, | |
| "loss": 0.8663, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.009647035817879608, | |
| "grad_norm": 3.7384579181671143, | |
| "learning_rate": 2.7477894736842106e-05, | |
| "loss": 0.9184, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.00971221849232474, | |
| "grad_norm": 4.581092834472656, | |
| "learning_rate": 2.6949473684210527e-05, | |
| "loss": 1.0749, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.009777401166769872, | |
| "grad_norm": 7.011988162994385, | |
| "learning_rate": 2.6421052631578945e-05, | |
| "loss": 1.1314, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.009777401166769872, | |
| "eval_loss": 0.8550050854682922, | |
| "eval_runtime": 746.3587, | |
| "eval_samples_per_second": 8.655, | |
| "eval_steps_per_second": 2.164, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.009842583841215006, | |
| "grad_norm": 2.339585304260254, | |
| "learning_rate": 2.589263157894737e-05, | |
| "loss": 0.7291, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.009907766515660138, | |
| "grad_norm": 2.318110942840576, | |
| "learning_rate": 2.536421052631579e-05, | |
| "loss": 0.5631, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.00997294919010527, | |
| "grad_norm": 1.509214997291565, | |
| "learning_rate": 2.483578947368421e-05, | |
| "loss": 0.2818, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.010038131864550402, | |
| "grad_norm": 1.4966317415237427, | |
| "learning_rate": 2.430736842105263e-05, | |
| "loss": 0.3039, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.010103314538995534, | |
| "grad_norm": 1.6486977338790894, | |
| "learning_rate": 2.3778947368421052e-05, | |
| "loss": 0.2881, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.010168497213440668, | |
| "grad_norm": 1.94291090965271, | |
| "learning_rate": 2.3250526315789473e-05, | |
| "loss": 0.3744, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.0102336798878858, | |
| "grad_norm": 2.5989363193511963, | |
| "learning_rate": 2.2722105263157894e-05, | |
| "loss": 0.7534, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.010298862562330932, | |
| "grad_norm": 2.6126620769500732, | |
| "learning_rate": 2.2193684210526316e-05, | |
| "loss": 0.7143, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.010364045236776065, | |
| "grad_norm": 3.0555806159973145, | |
| "learning_rate": 2.1665263157894737e-05, | |
| "loss": 0.8661, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.010429227911221197, | |
| "grad_norm": 2.486401319503784, | |
| "learning_rate": 2.1136842105263158e-05, | |
| "loss": 0.7079, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.01049441058566633, | |
| "grad_norm": 2.7490785121917725, | |
| "learning_rate": 2.060842105263158e-05, | |
| "loss": 0.7556, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.010559593260111463, | |
| "grad_norm": 3.5909423828125, | |
| "learning_rate": 2.008e-05, | |
| "loss": 0.8089, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.010624775934556595, | |
| "grad_norm": 2.8206937313079834, | |
| "learning_rate": 1.9551578947368422e-05, | |
| "loss": 0.5847, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.010689958609001727, | |
| "grad_norm": 4.731099605560303, | |
| "learning_rate": 1.9023157894736843e-05, | |
| "loss": 1.1395, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.010755141283446859, | |
| "grad_norm": 3.4779961109161377, | |
| "learning_rate": 1.849473684210526e-05, | |
| "loss": 0.8823, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.010820323957891993, | |
| "grad_norm": 4.600359916687012, | |
| "learning_rate": 1.7966315789473686e-05, | |
| "loss": 0.9135, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.010885506632337125, | |
| "grad_norm": 4.061426162719727, | |
| "learning_rate": 1.7437894736842107e-05, | |
| "loss": 1.0511, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.010950689306782257, | |
| "grad_norm": 4.084474563598633, | |
| "learning_rate": 1.6909473684210525e-05, | |
| "loss": 1.023, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.01101587198122739, | |
| "grad_norm": 4.1737213134765625, | |
| "learning_rate": 1.638105263157895e-05, | |
| "loss": 0.6724, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.011081054655672521, | |
| "grad_norm": 4.323464393615723, | |
| "learning_rate": 1.5852631578947368e-05, | |
| "loss": 0.7298, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.011146237330117655, | |
| "grad_norm": 3.48018217086792, | |
| "learning_rate": 1.532421052631579e-05, | |
| "loss": 0.9651, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.011211420004562787, | |
| "grad_norm": 3.573929786682129, | |
| "learning_rate": 1.4795789473684209e-05, | |
| "loss": 0.9659, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.01127660267900792, | |
| "grad_norm": 2.933013677597046, | |
| "learning_rate": 1.4267368421052632e-05, | |
| "loss": 0.6419, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.011341785353453052, | |
| "grad_norm": 2.557969808578491, | |
| "learning_rate": 1.3738947368421053e-05, | |
| "loss": 0.4679, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.011406968027898185, | |
| "grad_norm": 4.5971360206604, | |
| "learning_rate": 1.3210526315789473e-05, | |
| "loss": 0.8361, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.011472150702343318, | |
| "grad_norm": 2.6714653968811035, | |
| "learning_rate": 1.2682105263157896e-05, | |
| "loss": 0.7694, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.01153733337678845, | |
| "grad_norm": 2.248840570449829, | |
| "learning_rate": 1.2153684210526315e-05, | |
| "loss": 0.5984, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.011602516051233582, | |
| "grad_norm": 3.2296483516693115, | |
| "learning_rate": 1.1625263157894737e-05, | |
| "loss": 0.9471, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.011667698725678714, | |
| "grad_norm": 2.628018856048584, | |
| "learning_rate": 1.1096842105263158e-05, | |
| "loss": 0.6132, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.011732881400123848, | |
| "grad_norm": 2.5440778732299805, | |
| "learning_rate": 1.0568421052631579e-05, | |
| "loss": 0.4686, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.01179806407456898, | |
| "grad_norm": 5.853560447692871, | |
| "learning_rate": 1.004e-05, | |
| "loss": 0.9851, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.011863246749014112, | |
| "grad_norm": 3.293943405151367, | |
| "learning_rate": 9.511578947368422e-06, | |
| "loss": 0.8526, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.011928429423459244, | |
| "grad_norm": 4.2562761306762695, | |
| "learning_rate": 8.983157894736843e-06, | |
| "loss": 1.2255, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.011993612097904376, | |
| "grad_norm": 2.7283856868743896, | |
| "learning_rate": 8.454736842105263e-06, | |
| "loss": 0.7548, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.01205879477234951, | |
| "grad_norm": 3.0538840293884277, | |
| "learning_rate": 7.926315789473684e-06, | |
| "loss": 0.9394, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.012123977446794642, | |
| "grad_norm": 3.4769504070281982, | |
| "learning_rate": 7.397894736842104e-06, | |
| "loss": 0.7831, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.012189160121239774, | |
| "grad_norm": 4.124552249908447, | |
| "learning_rate": 6.8694736842105265e-06, | |
| "loss": 1.0468, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.012254342795684907, | |
| "grad_norm": 5.11678409576416, | |
| "learning_rate": 6.341052631578948e-06, | |
| "loss": 1.5542, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.012319525470130039, | |
| "grad_norm": 5.838722229003906, | |
| "learning_rate": 5.812631578947368e-06, | |
| "loss": 0.9434, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.012384708144575172, | |
| "grad_norm": 3.940918445587158, | |
| "learning_rate": 5.2842105263157896e-06, | |
| "loss": 1.0149, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.012449890819020305, | |
| "grad_norm": 3.6323320865631104, | |
| "learning_rate": 4.755789473684211e-06, | |
| "loss": 1.1067, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.012515073493465437, | |
| "grad_norm": 3.954714775085449, | |
| "learning_rate": 4.227368421052631e-06, | |
| "loss": 0.8935, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.012580256167910569, | |
| "grad_norm": 4.876105785369873, | |
| "learning_rate": 3.698947368421052e-06, | |
| "loss": 1.1635, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.012645438842355703, | |
| "grad_norm": 3.779630661010742, | |
| "learning_rate": 3.170526315789474e-06, | |
| "loss": 1.0547, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.012710621516800835, | |
| "grad_norm": 6.356569290161133, | |
| "learning_rate": 2.6421052631578948e-06, | |
| "loss": 1.4103, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.012775804191245967, | |
| "grad_norm": 5.530601978302002, | |
| "learning_rate": 2.1136842105263157e-06, | |
| "loss": 1.4511, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.012840986865691099, | |
| "grad_norm": 7.286373138427734, | |
| "learning_rate": 1.585263157894737e-06, | |
| "loss": 0.8836, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.012906169540136231, | |
| "grad_norm": 3.2843291759490967, | |
| "learning_rate": 1.0568421052631578e-06, | |
| "loss": 0.8748, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.012971352214581365, | |
| "grad_norm": 5.327868938446045, | |
| "learning_rate": 5.284210526315789e-07, | |
| "loss": 0.8412, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.013036534889026497, | |
| "grad_norm": 5.048506259918213, | |
| "learning_rate": 0.0, | |
| "loss": 1.0769, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.013036534889026497, | |
| "eval_loss": 0.8324354290962219, | |
| "eval_runtime": 746.1953, | |
| "eval_samples_per_second": 8.657, | |
| "eval_steps_per_second": 2.164, | |
| "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": 7.844270609478451e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |