Instructions to use chomeed/robometer_board_insertion_lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use chomeed/robometer_board_insertion_lora with PEFT:
Task type is invalid.
- Transformers
How to use chomeed/robometer_board_insertion_lora with Transformers:
# Load model directly from transformers import AutoProcessor, RBM processor = AutoProcessor.from_pretrained("chomeed/robometer_board_insertion_lora") model = RBM.from_pretrained("chomeed/robometer_board_insertion_lora", device_map="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.018475750577367205, | |
| "eval_steps": 100, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.8475750577367205e-05, | |
| "grad_norm": 22.456323623657227, | |
| "learning_rate": 0.0, | |
| "loss": 2.4233, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 3.695150115473441e-05, | |
| "grad_norm": 14.245879173278809, | |
| "learning_rate": 2.0000000000000002e-07, | |
| "loss": 1.4031, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 5.542725173210162e-05, | |
| "grad_norm": 4.1844964027404785, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.5747, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 7.390300230946882e-05, | |
| "grad_norm": 10.97097110748291, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 0.8075, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 9.237875288683603e-05, | |
| "grad_norm": 6.398587226867676, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 0.6758, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00011085450346420324, | |
| "grad_norm": 6.019951820373535, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.7705, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.00012933025404157045, | |
| "grad_norm": 7.282938480377197, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 0.5209, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.00014780600461893764, | |
| "grad_norm": 15.357967376708984, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.4928, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.00016628175519630485, | |
| "grad_norm": 4.623424530029297, | |
| "learning_rate": 1.6000000000000001e-06, | |
| "loss": 0.4328, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00018475750577367207, | |
| "grad_norm": 12.172774314880371, | |
| "learning_rate": 1.8000000000000001e-06, | |
| "loss": 0.4725, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.00020323325635103926, | |
| "grad_norm": 12.907696723937988, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.398, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.00022170900692840647, | |
| "grad_norm": 23.12137222290039, | |
| "learning_rate": 2.2e-06, | |
| "loss": 1.5045, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.00024018475750577366, | |
| "grad_norm": 11.940054893493652, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.7562, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0002586605080831409, | |
| "grad_norm": 18.102041244506836, | |
| "learning_rate": 2.6e-06, | |
| "loss": 1.6719, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0002771362586605081, | |
| "grad_norm": 9.714462280273438, | |
| "learning_rate": 2.8000000000000003e-06, | |
| "loss": 0.7902, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0002956120092378753, | |
| "grad_norm": 8.976316452026367, | |
| "learning_rate": 3e-06, | |
| "loss": 1.0058, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0003140877598152425, | |
| "grad_norm": 9.411298751831055, | |
| "learning_rate": 3.2000000000000003e-06, | |
| "loss": 0.972, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0003325635103926097, | |
| "grad_norm": 16.300657272338867, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 1.3468, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0003510392609699769, | |
| "grad_norm": 28.11629295349121, | |
| "learning_rate": 3.6000000000000003e-06, | |
| "loss": 1.5726, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.00036951501154734414, | |
| "grad_norm": 7.2270402908325195, | |
| "learning_rate": 3.8000000000000005e-06, | |
| "loss": 0.4829, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0003879907621247113, | |
| "grad_norm": 25.163707733154297, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 1.5316, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0004064665127020785, | |
| "grad_norm": 4.467431545257568, | |
| "learning_rate": 4.2000000000000004e-06, | |
| "loss": 0.4397, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0004249422632794457, | |
| "grad_norm": 15.441145896911621, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.9487, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.00044341801385681294, | |
| "grad_norm": 13.387373924255371, | |
| "learning_rate": 4.600000000000001e-06, | |
| "loss": 0.8103, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.00046189376443418013, | |
| "grad_norm": 9.357813835144043, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 0.744, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0004803695150115473, | |
| "grad_norm": 6.9000067710876465, | |
| "learning_rate": 5e-06, | |
| "loss": 0.744, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0004988452655889145, | |
| "grad_norm": 8.554380416870117, | |
| "learning_rate": 5.2e-06, | |
| "loss": 1.1066, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0005173210161662818, | |
| "grad_norm": 7.549210548400879, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.5524, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.000535796766743649, | |
| "grad_norm": 8.536428451538086, | |
| "learning_rate": 5.600000000000001e-06, | |
| "loss": 0.4924, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0005542725173210162, | |
| "grad_norm": 13.4658842086792, | |
| "learning_rate": 5.8e-06, | |
| "loss": 1.0435, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0005727482678983834, | |
| "grad_norm": 11.949090003967285, | |
| "learning_rate": 6e-06, | |
| "loss": 0.798, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0005912240184757506, | |
| "grad_norm": 6.550503253936768, | |
| "learning_rate": 6.200000000000001e-06, | |
| "loss": 0.4978, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0006096997690531177, | |
| "grad_norm": 12.27457332611084, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 1.1826, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.000628175519630485, | |
| "grad_norm": 14.90542984008789, | |
| "learning_rate": 6.600000000000001e-06, | |
| "loss": 0.5991, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0006466512702078522, | |
| "grad_norm": 15.954743385314941, | |
| "learning_rate": 6.800000000000001e-06, | |
| "loss": 1.2791, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0006651270207852194, | |
| "grad_norm": 8.662019729614258, | |
| "learning_rate": 7e-06, | |
| "loss": 0.4723, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0006836027713625866, | |
| "grad_norm": 13.14725112915039, | |
| "learning_rate": 7.2000000000000005e-06, | |
| "loss": 1.1883, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0007020785219399538, | |
| "grad_norm": 11.641953468322754, | |
| "learning_rate": 7.4e-06, | |
| "loss": 1.1936, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.000720554272517321, | |
| "grad_norm": 4.696055889129639, | |
| "learning_rate": 7.600000000000001e-06, | |
| "loss": 0.4239, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0007390300230946883, | |
| "grad_norm": 8.642461776733398, | |
| "learning_rate": 7.800000000000002e-06, | |
| "loss": 0.6516, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0007575057736720555, | |
| "grad_norm": 15.560766220092773, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.6935, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0007759815242494227, | |
| "grad_norm": 11.985705375671387, | |
| "learning_rate": 8.2e-06, | |
| "loss": 0.5668, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0007944572748267898, | |
| "grad_norm": 6.386831760406494, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.3256, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.000812933025404157, | |
| "grad_norm": 6.858893871307373, | |
| "learning_rate": 8.6e-06, | |
| "loss": 0.373, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0008314087759815242, | |
| "grad_norm": 6.545617580413818, | |
| "learning_rate": 8.8e-06, | |
| "loss": 0.5203, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0008498845265588914, | |
| "grad_norm": 6.690424919128418, | |
| "learning_rate": 9e-06, | |
| "loss": 0.6821, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0008683602771362587, | |
| "grad_norm": 2.724168062210083, | |
| "learning_rate": 9.200000000000002e-06, | |
| "loss": 0.3584, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0008868360277136259, | |
| "grad_norm": 4.201177597045898, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.3028, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0009053117782909931, | |
| "grad_norm": 3.529240131378174, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 0.3624, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0009237875288683603, | |
| "grad_norm": 9.743088722229004, | |
| "learning_rate": 9.800000000000001e-06, | |
| "loss": 0.5687, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0009422632794457275, | |
| "grad_norm": 3.6546342372894287, | |
| "learning_rate": 1e-05, | |
| "loss": 0.397, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0009607390300230946, | |
| "grad_norm": 7.545839786529541, | |
| "learning_rate": 1.02e-05, | |
| "loss": 0.4474, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0009792147806004618, | |
| "grad_norm": 3.845768928527832, | |
| "learning_rate": 1.04e-05, | |
| "loss": 0.3596, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.000997690531177829, | |
| "grad_norm": 4.227540493011475, | |
| "learning_rate": 1.0600000000000002e-05, | |
| "loss": 0.351, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0010161662817551962, | |
| "grad_norm": 4.890717029571533, | |
| "learning_rate": 1.0800000000000002e-05, | |
| "loss": 0.3199, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0010346420323325636, | |
| "grad_norm": 4.556331157684326, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 0.4166, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0010531177829099308, | |
| "grad_norm": 1.8914759159088135, | |
| "learning_rate": 1.1200000000000001e-05, | |
| "loss": 0.1817, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.001071593533487298, | |
| "grad_norm": 6.4960551261901855, | |
| "learning_rate": 1.14e-05, | |
| "loss": 0.5272, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0010900692840646652, | |
| "grad_norm": 2.595431327819824, | |
| "learning_rate": 1.16e-05, | |
| "loss": 0.2435, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0011085450346420324, | |
| "grad_norm": 7.035714149475098, | |
| "learning_rate": 1.18e-05, | |
| "loss": 0.5757, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0011270207852193995, | |
| "grad_norm": 4.48577880859375, | |
| "learning_rate": 1.2e-05, | |
| "loss": 0.3932, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0011454965357967667, | |
| "grad_norm": 3.052825450897217, | |
| "learning_rate": 1.22e-05, | |
| "loss": 0.2144, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.001163972286374134, | |
| "grad_norm": 1.5160571336746216, | |
| "learning_rate": 1.2400000000000002e-05, | |
| "loss": 0.2012, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0011824480369515011, | |
| "grad_norm": 5.697556495666504, | |
| "learning_rate": 1.2600000000000001e-05, | |
| "loss": 0.4743, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0012009237875288683, | |
| "grad_norm": 4.945090293884277, | |
| "learning_rate": 1.2800000000000001e-05, | |
| "loss": 0.3702, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0012193995381062355, | |
| "grad_norm": 4.655821800231934, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "loss": 0.4235, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0012378752886836027, | |
| "grad_norm": 5.6088056564331055, | |
| "learning_rate": 1.3200000000000002e-05, | |
| "loss": 0.3487, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.00125635103926097, | |
| "grad_norm": 2.1730315685272217, | |
| "learning_rate": 1.3400000000000002e-05, | |
| "loss": 0.2189, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0012748267898383373, | |
| "grad_norm": 10.11043930053711, | |
| "learning_rate": 1.3600000000000002e-05, | |
| "loss": 0.7641, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0012933025404157045, | |
| "grad_norm": 2.4763123989105225, | |
| "learning_rate": 1.38e-05, | |
| "loss": 0.2229, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0013117782909930716, | |
| "grad_norm": 7.365203380584717, | |
| "learning_rate": 1.4e-05, | |
| "loss": 0.535, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0013302540415704388, | |
| "grad_norm": 6.852386951446533, | |
| "learning_rate": 1.4200000000000001e-05, | |
| "loss": 0.408, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.001348729792147806, | |
| "grad_norm": 2.0798065662384033, | |
| "learning_rate": 1.4400000000000001e-05, | |
| "loss": 0.1749, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0013672055427251732, | |
| "grad_norm": 6.253926753997803, | |
| "learning_rate": 1.46e-05, | |
| "loss": 0.3956, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0013856812933025404, | |
| "grad_norm": 5.404087543487549, | |
| "learning_rate": 1.48e-05, | |
| "loss": 0.4205, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0014041570438799076, | |
| "grad_norm": 4.7791056632995605, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 0.3566, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0014226327944572748, | |
| "grad_norm": 5.373254299163818, | |
| "learning_rate": 1.5200000000000002e-05, | |
| "loss": 0.3713, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.001441108545034642, | |
| "grad_norm": 5.821455001831055, | |
| "learning_rate": 1.54e-05, | |
| "loss": 0.5614, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.0014595842956120091, | |
| "grad_norm": 4.620726585388184, | |
| "learning_rate": 1.5600000000000003e-05, | |
| "loss": 0.3735, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.0014780600461893765, | |
| "grad_norm": 6.009064674377441, | |
| "learning_rate": 1.58e-05, | |
| "loss": 0.3939, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0014965357967667437, | |
| "grad_norm": 9.253092765808105, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 0.4909, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.001515011547344111, | |
| "grad_norm": 1.2246159315109253, | |
| "learning_rate": 1.62e-05, | |
| "loss": 0.1898, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0015334872979214781, | |
| "grad_norm": 6.891019344329834, | |
| "learning_rate": 1.64e-05, | |
| "loss": 0.5467, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.0015519630484988453, | |
| "grad_norm": 1.9832881689071655, | |
| "learning_rate": 1.66e-05, | |
| "loss": 0.1615, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.0015704387990762125, | |
| "grad_norm": 7.6134138107299805, | |
| "learning_rate": 1.6800000000000002e-05, | |
| "loss": 0.3582, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.0015889145496535797, | |
| "grad_norm": 1.0416377782821655, | |
| "learning_rate": 1.7e-05, | |
| "loss": 0.1523, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.0016073903002309469, | |
| "grad_norm": 4.736614227294922, | |
| "learning_rate": 1.72e-05, | |
| "loss": 0.3068, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.001625866050808314, | |
| "grad_norm": 3.606045961380005, | |
| "learning_rate": 1.7400000000000003e-05, | |
| "loss": 0.2582, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.0016443418013856812, | |
| "grad_norm": 1.6798365116119385, | |
| "learning_rate": 1.76e-05, | |
| "loss": 0.1624, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0016628175519630484, | |
| "grad_norm": 6.613982200622559, | |
| "learning_rate": 1.7800000000000002e-05, | |
| "loss": 0.3216, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0016812933025404156, | |
| "grad_norm": 9.24609375, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.5028, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0016997690531177828, | |
| "grad_norm": 1.40059494972229, | |
| "learning_rate": 1.8200000000000002e-05, | |
| "loss": 0.1339, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0017182448036951502, | |
| "grad_norm": 5.237722396850586, | |
| "learning_rate": 1.8400000000000003e-05, | |
| "loss": 0.3446, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0017367205542725174, | |
| "grad_norm": 9.918924331665039, | |
| "learning_rate": 1.86e-05, | |
| "loss": 0.6905, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.0017551963048498846, | |
| "grad_norm": 1.7404794692993164, | |
| "learning_rate": 1.88e-05, | |
| "loss": 0.1692, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.0017736720554272518, | |
| "grad_norm": 3.403000593185425, | |
| "learning_rate": 1.9e-05, | |
| "loss": 0.2389, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.001792147806004619, | |
| "grad_norm": 3.996196985244751, | |
| "learning_rate": 1.9200000000000003e-05, | |
| "loss": 0.3229, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.0018106235565819861, | |
| "grad_norm": 3.08899188041687, | |
| "learning_rate": 1.94e-05, | |
| "loss": 0.288, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.0018290993071593533, | |
| "grad_norm": 4.761425018310547, | |
| "learning_rate": 1.9600000000000002e-05, | |
| "loss": 0.2725, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.0018475750577367205, | |
| "grad_norm": 11.36471176147461, | |
| "learning_rate": 1.98e-05, | |
| "loss": 0.4134, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0018660508083140877, | |
| "grad_norm": 4.157029151916504, | |
| "learning_rate": 2e-05, | |
| "loss": 0.3237, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.001884526558891455, | |
| "grad_norm": 4.969934940338135, | |
| "learning_rate": 1.9999939076577906e-05, | |
| "loss": 0.3539, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.001903002309468822, | |
| "grad_norm": 2.8749794960021973, | |
| "learning_rate": 1.9999756307053947e-05, | |
| "loss": 0.2316, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.0019214780600461893, | |
| "grad_norm": 4.532221794128418, | |
| "learning_rate": 1.9999451693655125e-05, | |
| "loss": 0.3129, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.0019399538106235567, | |
| "grad_norm": 4.383476257324219, | |
| "learning_rate": 1.9999025240093045e-05, | |
| "loss": 0.2416, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.0019584295612009237, | |
| "grad_norm": 10.787030220031738, | |
| "learning_rate": 1.9998476951563914e-05, | |
| "loss": 0.5814, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.001976905311778291, | |
| "grad_norm": 0.481875479221344, | |
| "learning_rate": 1.9997806834748455e-05, | |
| "loss": 0.1093, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.001995381062355658, | |
| "grad_norm": 2.0356526374816895, | |
| "learning_rate": 1.9997014897811834e-05, | |
| "loss": 0.1621, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.002013856812933025, | |
| "grad_norm": 3.2280030250549316, | |
| "learning_rate": 1.9996101150403543e-05, | |
| "loss": 0.2875, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.0020323325635103924, | |
| "grad_norm": 3.201484441757202, | |
| "learning_rate": 1.9995065603657317e-05, | |
| "loss": 0.2406, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.00205080831408776, | |
| "grad_norm": 6.806441783905029, | |
| "learning_rate": 1.999390827019096e-05, | |
| "loss": 0.3433, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.002069284064665127, | |
| "grad_norm": 2.1120567321777344, | |
| "learning_rate": 1.999262916410621e-05, | |
| "loss": 0.1857, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.0020877598152424944, | |
| "grad_norm": 3.2112979888916016, | |
| "learning_rate": 1.9991228300988586e-05, | |
| "loss": 0.2638, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.0021062355658198616, | |
| "grad_norm": 0.6795028448104858, | |
| "learning_rate": 1.998970569790715e-05, | |
| "loss": 0.1188, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.0021247113163972288, | |
| "grad_norm": 2.4822771549224854, | |
| "learning_rate": 1.9988061373414342e-05, | |
| "loss": 0.1741, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.002143187066974596, | |
| "grad_norm": 1.8658642768859863, | |
| "learning_rate": 1.9986295347545738e-05, | |
| "loss": 0.1472, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.002161662817551963, | |
| "grad_norm": 0.7293748259544373, | |
| "learning_rate": 1.9984407641819812e-05, | |
| "loss": 0.1112, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.0021801385681293303, | |
| "grad_norm": 1.8126376867294312, | |
| "learning_rate": 1.9982398279237657e-05, | |
| "loss": 0.1312, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.0021986143187066975, | |
| "grad_norm": 3.559358596801758, | |
| "learning_rate": 1.9980267284282718e-05, | |
| "loss": 0.2045, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.0022170900692840647, | |
| "grad_norm": 4.099513530731201, | |
| "learning_rate": 1.9978014682920503e-05, | |
| "loss": 0.2042, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.002235565819861432, | |
| "grad_norm": 5.324329853057861, | |
| "learning_rate": 1.9975640502598243e-05, | |
| "loss": 0.2095, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.002254041570438799, | |
| "grad_norm": 6.035201072692871, | |
| "learning_rate": 1.997314477224458e-05, | |
| "loss": 0.3251, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.0022725173210161663, | |
| "grad_norm": 1.535263180732727, | |
| "learning_rate": 1.9970527522269204e-05, | |
| "loss": 0.1234, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.0022909930715935335, | |
| "grad_norm": 1.5387904644012451, | |
| "learning_rate": 1.9967788784562474e-05, | |
| "loss": 0.112, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.0023094688221709007, | |
| "grad_norm": 8.320265769958496, | |
| "learning_rate": 1.9964928592495046e-05, | |
| "loss": 0.2823, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.002327944572748268, | |
| "grad_norm": 3.680867910385132, | |
| "learning_rate": 1.9961946980917457e-05, | |
| "loss": 0.187, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.002346420323325635, | |
| "grad_norm": 9.67679500579834, | |
| "learning_rate": 1.9958843986159705e-05, | |
| "loss": 0.4469, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.0023648960739030022, | |
| "grad_norm": 0.5361282825469971, | |
| "learning_rate": 1.99556196460308e-05, | |
| "loss": 0.0976, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.0023833718244803694, | |
| "grad_norm": 2.377816915512085, | |
| "learning_rate": 1.9952273999818312e-05, | |
| "loss": 0.2091, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.0024018475750577366, | |
| "grad_norm": 3.169578790664673, | |
| "learning_rate": 1.9948807088287884e-05, | |
| "loss": 0.191, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.002420323325635104, | |
| "grad_norm": 6.146152973175049, | |
| "learning_rate": 1.9945218953682736e-05, | |
| "loss": 0.2768, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.002438799076212471, | |
| "grad_norm": 2.137664794921875, | |
| "learning_rate": 1.9941509639723155e-05, | |
| "loss": 0.1385, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.002457274826789838, | |
| "grad_norm": 0.8729813098907471, | |
| "learning_rate": 1.9937679191605964e-05, | |
| "loss": 0.1136, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.0024757505773672053, | |
| "grad_norm": 1.049378752708435, | |
| "learning_rate": 1.9933727656003964e-05, | |
| "loss": 0.1324, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.0024942263279445725, | |
| "grad_norm": 1.994313359260559, | |
| "learning_rate": 1.992965508106537e-05, | |
| "loss": 0.1472, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.00251270207852194, | |
| "grad_norm": 0.8398813009262085, | |
| "learning_rate": 1.9925461516413224e-05, | |
| "loss": 0.1153, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.0025311778290993073, | |
| "grad_norm": 12.569574356079102, | |
| "learning_rate": 1.9921147013144782e-05, | |
| "loss": 0.8833, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.0025496535796766745, | |
| "grad_norm": 6.19824743270874, | |
| "learning_rate": 1.9916711623830904e-05, | |
| "loss": 0.7949, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.0025681293302540417, | |
| "grad_norm": 10.820578575134277, | |
| "learning_rate": 1.991215540251542e-05, | |
| "loss": 0.4554, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.002586605080831409, | |
| "grad_norm": 0.8254913091659546, | |
| "learning_rate": 1.9907478404714438e-05, | |
| "loss": 0.1007, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.002605080831408776, | |
| "grad_norm": 1.0913249254226685, | |
| "learning_rate": 1.9902680687415704e-05, | |
| "loss": 0.1376, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.0026235565819861433, | |
| "grad_norm": 2.495514392852783, | |
| "learning_rate": 1.989776230907789e-05, | |
| "loss": 0.1877, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.0026420323325635105, | |
| "grad_norm": 5.1712164878845215, | |
| "learning_rate": 1.9892723329629885e-05, | |
| "loss": 0.2565, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.0026605080831408777, | |
| "grad_norm": 1.013431429862976, | |
| "learning_rate": 1.988756381047006e-05, | |
| "loss": 0.1165, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.002678983833718245, | |
| "grad_norm": 11.041007041931152, | |
| "learning_rate": 1.988228381446553e-05, | |
| "loss": 0.2602, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.002697459584295612, | |
| "grad_norm": 0.6723874807357788, | |
| "learning_rate": 1.9876883405951378e-05, | |
| "loss": 0.0996, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.0027159353348729792, | |
| "grad_norm": 3.5139999389648438, | |
| "learning_rate": 1.987136265072988e-05, | |
| "loss": 0.1612, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.0027344110854503464, | |
| "grad_norm": 2.811295747756958, | |
| "learning_rate": 1.9865721616069695e-05, | |
| "loss": 0.1644, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.0027528868360277136, | |
| "grad_norm": 6.439018249511719, | |
| "learning_rate": 1.985996037070505e-05, | |
| "loss": 0.2701, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.002771362586605081, | |
| "grad_norm": 8.684128761291504, | |
| "learning_rate": 1.9854078984834904e-05, | |
| "loss": 0.3185, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.002789838337182448, | |
| "grad_norm": 9.836875915527344, | |
| "learning_rate": 1.9848077530122083e-05, | |
| "loss": 0.6175, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.002808314087759815, | |
| "grad_norm": 2.435478925704956, | |
| "learning_rate": 1.984195607969242e-05, | |
| "loss": 0.1349, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.0028267898383371824, | |
| "grad_norm": 5.893956661224365, | |
| "learning_rate": 1.983571470813386e-05, | |
| "loss": 0.2348, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.0028452655889145495, | |
| "grad_norm": 6.817545413970947, | |
| "learning_rate": 1.9829353491495545e-05, | |
| "loss": 0.2202, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.0028637413394919167, | |
| "grad_norm": 9.590438842773438, | |
| "learning_rate": 1.982287250728689e-05, | |
| "loss": 0.6335, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.002882217090069284, | |
| "grad_norm": 0.6016367077827454, | |
| "learning_rate": 1.9816271834476642e-05, | |
| "loss": 0.1068, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.002900692840646651, | |
| "grad_norm": 2.6583564281463623, | |
| "learning_rate": 1.9809551553491918e-05, | |
| "loss": 0.1483, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.0029191685912240183, | |
| "grad_norm": 3.617194414138794, | |
| "learning_rate": 1.9802711746217222e-05, | |
| "loss": 0.1858, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.0029376443418013855, | |
| "grad_norm": 1.4362705945968628, | |
| "learning_rate": 1.979575249599344e-05, | |
| "loss": 0.141, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.002956120092378753, | |
| "grad_norm": 0.7956472039222717, | |
| "learning_rate": 1.9788673887616852e-05, | |
| "loss": 0.1051, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.0029745958429561203, | |
| "grad_norm": 12.44459056854248, | |
| "learning_rate": 1.9781476007338058e-05, | |
| "loss": 0.7184, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.0029930715935334875, | |
| "grad_norm": 8.499845504760742, | |
| "learning_rate": 1.9774158942860962e-05, | |
| "loss": 0.4093, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.0030115473441108547, | |
| "grad_norm": 4.637056350708008, | |
| "learning_rate": 1.9766722783341682e-05, | |
| "loss": 0.2223, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.003030023094688222, | |
| "grad_norm": 5.749958038330078, | |
| "learning_rate": 1.9759167619387474e-05, | |
| "loss": 0.3341, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.003048498845265589, | |
| "grad_norm": 6.744971752166748, | |
| "learning_rate": 1.9751493543055634e-05, | |
| "loss": 0.3504, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.0030669745958429562, | |
| "grad_norm": 7.550142765045166, | |
| "learning_rate": 1.9743700647852356e-05, | |
| "loss": 0.3722, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.0030854503464203234, | |
| "grad_norm": 2.827252149581909, | |
| "learning_rate": 1.9735789028731603e-05, | |
| "loss": 0.176, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.0031039260969976906, | |
| "grad_norm": 11.453800201416016, | |
| "learning_rate": 1.972775878209397e-05, | |
| "loss": 0.5393, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.003122401847575058, | |
| "grad_norm": 1.4012537002563477, | |
| "learning_rate": 1.9719610005785466e-05, | |
| "loss": 0.1216, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.003140877598152425, | |
| "grad_norm": 3.6133031845092773, | |
| "learning_rate": 1.971134279909636e-05, | |
| "loss": 0.2281, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.003159353348729792, | |
| "grad_norm": 2.48596453666687, | |
| "learning_rate": 1.9702957262759964e-05, | |
| "loss": 0.129, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.0031778290993071594, | |
| "grad_norm": 4.8542280197143555, | |
| "learning_rate": 1.9694453498951392e-05, | |
| "loss": 0.2945, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.0031963048498845265, | |
| "grad_norm": 7.3076252937316895, | |
| "learning_rate": 1.9685831611286312e-05, | |
| "loss": 0.4099, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.0032147806004618937, | |
| "grad_norm": 2.754129648208618, | |
| "learning_rate": 1.9677091704819714e-05, | |
| "loss": 0.1379, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.003233256351039261, | |
| "grad_norm": 13.845019340515137, | |
| "learning_rate": 1.9668233886044597e-05, | |
| "loss": 0.5544, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.003251732101616628, | |
| "grad_norm": 1.5796810388565063, | |
| "learning_rate": 1.9659258262890683e-05, | |
| "loss": 0.1171, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.0032702078521939953, | |
| "grad_norm": 1.0137169361114502, | |
| "learning_rate": 1.9650164944723116e-05, | |
| "loss": 0.1101, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.0032886836027713625, | |
| "grad_norm": 1.3974528312683105, | |
| "learning_rate": 1.96409540423411e-05, | |
| "loss": 0.115, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.0033071593533487297, | |
| "grad_norm": 7.187819004058838, | |
| "learning_rate": 1.9631625667976584e-05, | |
| "loss": 0.2381, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.003325635103926097, | |
| "grad_norm": 2.6284053325653076, | |
| "learning_rate": 1.9622179935292855e-05, | |
| "loss": 0.1477, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.003344110854503464, | |
| "grad_norm": 1.2598968744277954, | |
| "learning_rate": 1.961261695938319e-05, | |
| "loss": 0.1292, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.0033625866050808312, | |
| "grad_norm": 8.733541488647461, | |
| "learning_rate": 1.9602936856769432e-05, | |
| "loss": 0.6258, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.0033810623556581984, | |
| "grad_norm": 3.4297659397125244, | |
| "learning_rate": 1.9593139745400575e-05, | |
| "loss": 0.1499, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.0033995381062355656, | |
| "grad_norm": 1.8799363374710083, | |
| "learning_rate": 1.9583225744651334e-05, | |
| "loss": 0.1277, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.0034180138568129332, | |
| "grad_norm": 7.182145118713379, | |
| "learning_rate": 1.9573194975320672e-05, | |
| "loss": 0.3813, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.0034364896073903004, | |
| "grad_norm": 1.1324434280395508, | |
| "learning_rate": 1.9563047559630356e-05, | |
| "loss": 0.1183, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.0034549653579676676, | |
| "grad_norm": 7.767648220062256, | |
| "learning_rate": 1.9552783621223437e-05, | |
| "loss": 0.1979, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.003473441108545035, | |
| "grad_norm": 1.4186517000198364, | |
| "learning_rate": 1.954240328516277e-05, | |
| "loss": 0.1442, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.003491916859122402, | |
| "grad_norm": 5.253149032592773, | |
| "learning_rate": 1.9531906677929472e-05, | |
| "loss": 0.1732, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.003510392609699769, | |
| "grad_norm": 10.433268547058105, | |
| "learning_rate": 1.9521293927421388e-05, | |
| "loss": 0.2958, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.0035288683602771364, | |
| "grad_norm": 0.7180191874504089, | |
| "learning_rate": 1.9510565162951538e-05, | |
| "loss": 0.0961, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.0035473441108545035, | |
| "grad_norm": 4.97097110748291, | |
| "learning_rate": 1.9499720515246524e-05, | |
| "loss": 0.2368, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.0035658198614318707, | |
| "grad_norm": 6.852952003479004, | |
| "learning_rate": 1.9488760116444966e-05, | |
| "loss": 0.4742, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.003584295612009238, | |
| "grad_norm": 4.5728068351745605, | |
| "learning_rate": 1.947768410009586e-05, | |
| "loss": 0.2806, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.003602771362586605, | |
| "grad_norm": 3.9592857360839844, | |
| "learning_rate": 1.9466492601156964e-05, | |
| "loss": 0.1958, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.0036212471131639723, | |
| "grad_norm": 2.2949304580688477, | |
| "learning_rate": 1.945518575599317e-05, | |
| "loss": 0.1325, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.0036397228637413395, | |
| "grad_norm": 1.1706838607788086, | |
| "learning_rate": 1.944376370237481e-05, | |
| "loss": 0.1231, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.0036581986143187067, | |
| "grad_norm": 2.7541165351867676, | |
| "learning_rate": 1.943222657947601e-05, | |
| "loss": 0.1576, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.003676674364896074, | |
| "grad_norm": 6.214320659637451, | |
| "learning_rate": 1.942057452787297e-05, | |
| "loss": 0.7367, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.003695150115473441, | |
| "grad_norm": 0.5719465613365173, | |
| "learning_rate": 1.9408807689542257e-05, | |
| "loss": 0.0951, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.0037136258660508082, | |
| "grad_norm": 2.071755886077881, | |
| "learning_rate": 1.9396926207859085e-05, | |
| "loss": 0.1626, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.0037321016166281754, | |
| "grad_norm": 6.556413650512695, | |
| "learning_rate": 1.938493022759556e-05, | |
| "loss": 0.3211, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.0037505773672055426, | |
| "grad_norm": 6.717318534851074, | |
| "learning_rate": 1.937281989491892e-05, | |
| "loss": 0.6473, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.00376905311778291, | |
| "grad_norm": 4.546645641326904, | |
| "learning_rate": 1.9360595357389735e-05, | |
| "loss": 0.2684, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.003787528868360277, | |
| "grad_norm": 2.214829206466675, | |
| "learning_rate": 1.9348256763960146e-05, | |
| "loss": 0.1473, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.003806004618937644, | |
| "grad_norm": 8.836477279663086, | |
| "learning_rate": 1.9335804264972018e-05, | |
| "loss": 0.286, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.0038244803695150114, | |
| "grad_norm": 1.4433702230453491, | |
| "learning_rate": 1.9323238012155125e-05, | |
| "loss": 0.1327, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.0038429561200923786, | |
| "grad_norm": 3.100055456161499, | |
| "learning_rate": 1.9310558158625286e-05, | |
| "loss": 0.208, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.003861431870669746, | |
| "grad_norm": 3.310422897338867, | |
| "learning_rate": 1.9297764858882516e-05, | |
| "loss": 0.2025, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.0038799076212471134, | |
| "grad_norm": 5.193705081939697, | |
| "learning_rate": 1.9284858268809135e-05, | |
| "loss": 0.1897, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.0038983833718244805, | |
| "grad_norm": 0.7226227521896362, | |
| "learning_rate": 1.9271838545667876e-05, | |
| "loss": 0.1147, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.003916859122401847, | |
| "grad_norm": 1.7036936283111572, | |
| "learning_rate": 1.925870584809995e-05, | |
| "loss": 0.1572, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.0039353348729792145, | |
| "grad_norm": 0.9940951466560364, | |
| "learning_rate": 1.9245460336123136e-05, | |
| "loss": 0.1101, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.003953810623556582, | |
| "grad_norm": 4.127312183380127, | |
| "learning_rate": 1.923210217112981e-05, | |
| "loss": 0.2239, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.003972286374133949, | |
| "grad_norm": 5.69522762298584, | |
| "learning_rate": 1.9218631515885007e-05, | |
| "loss": 0.2143, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.003990762124711316, | |
| "grad_norm": 2.805601119995117, | |
| "learning_rate": 1.9205048534524405e-05, | |
| "loss": 0.16, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.004009237875288683, | |
| "grad_norm": 2.847810983657837, | |
| "learning_rate": 1.9191353392552346e-05, | |
| "loss": 0.1683, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.00402771362586605, | |
| "grad_norm": 1.9425185918807983, | |
| "learning_rate": 1.9177546256839814e-05, | |
| "loss": 0.1562, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.004046189376443418, | |
| "grad_norm": 3.480607748031616, | |
| "learning_rate": 1.9163627295622397e-05, | |
| "loss": 0.1981, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.004064665127020785, | |
| "grad_norm": 9.408126831054688, | |
| "learning_rate": 1.914959667849825e-05, | |
| "loss": 0.328, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.004083140877598152, | |
| "grad_norm": 6.1869282722473145, | |
| "learning_rate": 1.913545457642601e-05, | |
| "loss": 0.3474, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.00410161662817552, | |
| "grad_norm": 0.5331553816795349, | |
| "learning_rate": 1.9121201161722732e-05, | |
| "loss": 0.0991, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.004120092378752887, | |
| "grad_norm": 0.6851425766944885, | |
| "learning_rate": 1.910683660806177e-05, | |
| "loss": 0.1062, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.004138568129330254, | |
| "grad_norm": 4.599374771118164, | |
| "learning_rate": 1.9092361090470688e-05, | |
| "loss": 0.191, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.004157043879907622, | |
| "grad_norm": 3.847731828689575, | |
| "learning_rate": 1.907777478532909e-05, | |
| "loss": 0.173, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.004175519630484989, | |
| "grad_norm": 2.8963265419006348, | |
| "learning_rate": 1.9063077870366504e-05, | |
| "loss": 0.2124, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.004193995381062356, | |
| "grad_norm": 6.79649019241333, | |
| "learning_rate": 1.9048270524660197e-05, | |
| "loss": 0.4054, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.004212471131639723, | |
| "grad_norm": 14.4025239944458, | |
| "learning_rate": 1.903335292863301e-05, | |
| "loss": 0.5937, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.00423094688221709, | |
| "grad_norm": 11.668035507202148, | |
| "learning_rate": 1.901832526405114e-05, | |
| "loss": 0.6448, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.0042494226327944576, | |
| "grad_norm": 8.959061622619629, | |
| "learning_rate": 1.9003187714021936e-05, | |
| "loss": 0.1986, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.004267898383371825, | |
| "grad_norm": 2.872007369995117, | |
| "learning_rate": 1.8987940462991673e-05, | |
| "loss": 0.2371, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.004286374133949192, | |
| "grad_norm": 1.2436494827270508, | |
| "learning_rate": 1.8972583696743284e-05, | |
| "loss": 0.1191, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.004304849884526559, | |
| "grad_norm": 1.8871135711669922, | |
| "learning_rate": 1.895711760239413e-05, | |
| "loss": 0.1138, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.004323325635103926, | |
| "grad_norm": 4.414292335510254, | |
| "learning_rate": 1.8941542368393683e-05, | |
| "loss": 0.1862, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.0043418013856812935, | |
| "grad_norm": 11.602113723754883, | |
| "learning_rate": 1.892585818452126e-05, | |
| "loss": 0.3518, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.004360277136258661, | |
| "grad_norm": 1.1844524145126343, | |
| "learning_rate": 1.891006524188368e-05, | |
| "loss": 0.1173, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.004378752886836028, | |
| "grad_norm": 0.7867116928100586, | |
| "learning_rate": 1.889416373291298e-05, | |
| "loss": 0.1032, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.004397228637413395, | |
| "grad_norm": 4.562929153442383, | |
| "learning_rate": 1.8878153851364013e-05, | |
| "loss": 0.2363, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.004415704387990762, | |
| "grad_norm": 1.203452467918396, | |
| "learning_rate": 1.8862035792312148e-05, | |
| "loss": 0.1159, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.004434180138568129, | |
| "grad_norm": 2.852126359939575, | |
| "learning_rate": 1.884580975215084e-05, | |
| "loss": 0.1581, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.004452655889145497, | |
| "grad_norm": 2.060908079147339, | |
| "learning_rate": 1.8829475928589272e-05, | |
| "loss": 0.1281, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.004471131639722864, | |
| "grad_norm": 3.143167734146118, | |
| "learning_rate": 1.8813034520649923e-05, | |
| "loss": 0.1263, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.004489607390300231, | |
| "grad_norm": 7.002387046813965, | |
| "learning_rate": 1.879648572866617e-05, | |
| "loss": 0.3517, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.004508083140877598, | |
| "grad_norm": 1.2260137796401978, | |
| "learning_rate": 1.8779829754279806e-05, | |
| "loss": 0.1122, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.004526558891454965, | |
| "grad_norm": 1.0295634269714355, | |
| "learning_rate": 1.8763066800438638e-05, | |
| "loss": 0.1159, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.0045450346420323326, | |
| "grad_norm": 5.545650959014893, | |
| "learning_rate": 1.874619707139396e-05, | |
| "loss": 0.2099, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.0045635103926097, | |
| "grad_norm": 4.304337024688721, | |
| "learning_rate": 1.8729220772698096e-05, | |
| "loss": 0.1848, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.004581986143187067, | |
| "grad_norm": 17.612524032592773, | |
| "learning_rate": 1.8712138111201898e-05, | |
| "loss": 0.5794, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.004600461893764434, | |
| "grad_norm": 2.4110946655273438, | |
| "learning_rate": 1.869494929505219e-05, | |
| "loss": 0.1659, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.004618937644341801, | |
| "grad_norm": 8.169907569885254, | |
| "learning_rate": 1.8677654533689287e-05, | |
| "loss": 0.2315, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.0046374133949191685, | |
| "grad_norm": 7.461135387420654, | |
| "learning_rate": 1.866025403784439e-05, | |
| "loss": 0.5243, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.004655889145496536, | |
| "grad_norm": 1.049268364906311, | |
| "learning_rate": 1.864274801953705e-05, | |
| "loss": 0.1097, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.004674364896073903, | |
| "grad_norm": 0.7230748534202576, | |
| "learning_rate": 1.8625136692072577e-05, | |
| "loss": 0.0895, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.00469284064665127, | |
| "grad_norm": 2.38818359375, | |
| "learning_rate": 1.860742027003944e-05, | |
| "loss": 0.1854, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.004711316397228637, | |
| "grad_norm": 8.430672645568848, | |
| "learning_rate": 1.8589598969306646e-05, | |
| "loss": 0.3956, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.0047297921478060044, | |
| "grad_norm": 0.6229478716850281, | |
| "learning_rate": 1.8571673007021124e-05, | |
| "loss": 0.1082, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.004748267898383372, | |
| "grad_norm": 0.9911284446716309, | |
| "learning_rate": 1.855364260160507e-05, | |
| "loss": 0.1084, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.004766743648960739, | |
| "grad_norm": 1.2117606401443481, | |
| "learning_rate": 1.8535507972753275e-05, | |
| "loss": 0.1226, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.004785219399538106, | |
| "grad_norm": 0.48902785778045654, | |
| "learning_rate": 1.851726934143048e-05, | |
| "loss": 0.1008, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.004803695150115473, | |
| "grad_norm": 4.6468634605407715, | |
| "learning_rate": 1.849892692986864e-05, | |
| "loss": 0.1876, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.00482217090069284, | |
| "grad_norm": 1.3272570371627808, | |
| "learning_rate": 1.848048096156426e-05, | |
| "loss": 0.1131, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.004840646651270208, | |
| "grad_norm": 7.653775215148926, | |
| "learning_rate": 1.8461931661275642e-05, | |
| "loss": 0.2721, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.004859122401847575, | |
| "grad_norm": 0.4216022789478302, | |
| "learning_rate": 1.8443279255020153e-05, | |
| "loss": 0.0827, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.004877598152424942, | |
| "grad_norm": 1.9425665140151978, | |
| "learning_rate": 1.842452397007148e-05, | |
| "loss": 0.1404, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.004896073903002309, | |
| "grad_norm": 0.6309486627578735, | |
| "learning_rate": 1.8405666034956842e-05, | |
| "loss": 0.1155, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.004914549653579676, | |
| "grad_norm": 10.29332160949707, | |
| "learning_rate": 1.8386705679454243e-05, | |
| "loss": 0.2914, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.0049330254041570435, | |
| "grad_norm": 1.5594719648361206, | |
| "learning_rate": 1.836764313458962e-05, | |
| "loss": 0.1007, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.004951501154734411, | |
| "grad_norm": 6.527675628662109, | |
| "learning_rate": 1.8348478632634067e-05, | |
| "loss": 0.6297, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.004969976905311778, | |
| "grad_norm": 10.980656623840332, | |
| "learning_rate": 1.8329212407100996e-05, | |
| "loss": 0.7336, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.004988452655889145, | |
| "grad_norm": 6.41602087020874, | |
| "learning_rate": 1.8309844692743283e-05, | |
| "loss": 0.2192, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.005006928406466513, | |
| "grad_norm": 0.7219845652580261, | |
| "learning_rate": 1.8290375725550417e-05, | |
| "loss": 0.0958, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.00502540415704388, | |
| "grad_norm": 5.545040130615234, | |
| "learning_rate": 1.827080574274562e-05, | |
| "loss": 0.3424, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.0050438799076212475, | |
| "grad_norm": 1.2542800903320312, | |
| "learning_rate": 1.8251134982782952e-05, | |
| "loss": 0.1187, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.005062355658198615, | |
| "grad_norm": 2.3587517738342285, | |
| "learning_rate": 1.8231363685344422e-05, | |
| "loss": 0.1393, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.005080831408775982, | |
| "grad_norm": 2.3231513500213623, | |
| "learning_rate": 1.821149209133704e-05, | |
| "loss": 0.1624, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.005099307159353349, | |
| "grad_norm": 7.323733806610107, | |
| "learning_rate": 1.819152044288992e-05, | |
| "loss": 0.5742, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.005117782909930716, | |
| "grad_norm": 8.194135665893555, | |
| "learning_rate": 1.8171448983351284e-05, | |
| "loss": 0.3908, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.0051362586605080834, | |
| "grad_norm": 3.8306798934936523, | |
| "learning_rate": 1.815127795728554e-05, | |
| "loss": 0.1965, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.005154734411085451, | |
| "grad_norm": 3.301453113555908, | |
| "learning_rate": 1.8131007610470278e-05, | |
| "loss": 0.162, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.005173210161662818, | |
| "grad_norm": 4.991317272186279, | |
| "learning_rate": 1.8110638189893267e-05, | |
| "loss": 0.3039, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.005191685912240185, | |
| "grad_norm": 1.6207114458084106, | |
| "learning_rate": 1.8090169943749477e-05, | |
| "loss": 0.1342, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.005210161662817552, | |
| "grad_norm": 1.194091558456421, | |
| "learning_rate": 1.806960312143802e-05, | |
| "loss": 0.119, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.005228637413394919, | |
| "grad_norm": 0.6452730298042297, | |
| "learning_rate": 1.804893797355914e-05, | |
| "loss": 0.0929, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.005247113163972287, | |
| "grad_norm": 1.9252386093139648, | |
| "learning_rate": 1.8028174751911147e-05, | |
| "loss": 0.1484, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.005265588914549654, | |
| "grad_norm": 1.1159170866012573, | |
| "learning_rate": 1.8007313709487334e-05, | |
| "loss": 0.1011, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.005284064665127021, | |
| "grad_norm": 1.3085510730743408, | |
| "learning_rate": 1.798635510047293e-05, | |
| "loss": 0.1254, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.005302540415704388, | |
| "grad_norm": 1.387012004852295, | |
| "learning_rate": 1.7965299180241963e-05, | |
| "loss": 0.0913, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.005321016166281755, | |
| "grad_norm": 5.165996074676514, | |
| "learning_rate": 1.7944146205354182e-05, | |
| "loss": 0.2532, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.0053394919168591225, | |
| "grad_norm": 8.985640525817871, | |
| "learning_rate": 1.792289643355191e-05, | |
| "loss": 0.3112, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.00535796766743649, | |
| "grad_norm": 2.128986120223999, | |
| "learning_rate": 1.7901550123756906e-05, | |
| "loss": 0.1192, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.005376443418013857, | |
| "grad_norm": 3.902437686920166, | |
| "learning_rate": 1.788010753606722e-05, | |
| "loss": 0.1604, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.005394919168591224, | |
| "grad_norm": 11.507761001586914, | |
| "learning_rate": 1.785856893175402e-05, | |
| "loss": 0.4229, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.005413394919168591, | |
| "grad_norm": 11.484999656677246, | |
| "learning_rate": 1.78369345732584e-05, | |
| "loss": 0.5436, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.0054318706697459584, | |
| "grad_norm": 1.0835720300674438, | |
| "learning_rate": 1.781520472418819e-05, | |
| "loss": 0.1023, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.005450346420323326, | |
| "grad_norm": 6.267183780670166, | |
| "learning_rate": 1.7793379649314743e-05, | |
| "loss": 0.3118, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.005468822170900693, | |
| "grad_norm": 2.6327121257781982, | |
| "learning_rate": 1.777145961456971e-05, | |
| "loss": 0.1529, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.00548729792147806, | |
| "grad_norm": 0.7303285598754883, | |
| "learning_rate": 1.7749444887041797e-05, | |
| "loss": 0.1089, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.005505773672055427, | |
| "grad_norm": 2.059903621673584, | |
| "learning_rate": 1.7727335734973512e-05, | |
| "loss": 0.1386, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.005524249422632794, | |
| "grad_norm": 1.108911395072937, | |
| "learning_rate": 1.7705132427757895e-05, | |
| "loss": 0.1061, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.005542725173210162, | |
| "grad_norm": 19.801851272583008, | |
| "learning_rate": 1.7682835235935236e-05, | |
| "loss": 0.4851, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.005561200923787529, | |
| "grad_norm": 1.7808235883712769, | |
| "learning_rate": 1.766044443118978e-05, | |
| "loss": 0.1242, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.005579676674364896, | |
| "grad_norm": 3.045708417892456, | |
| "learning_rate": 1.7637960286346423e-05, | |
| "loss": 0.2039, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.005598152424942263, | |
| "grad_norm": 2.138157606124878, | |
| "learning_rate": 1.761538307536737e-05, | |
| "loss": 0.1586, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.00561662817551963, | |
| "grad_norm": 5.394024848937988, | |
| "learning_rate": 1.759271307334881e-05, | |
| "loss": 0.3096, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.0056351039260969975, | |
| "grad_norm": 0.7137441635131836, | |
| "learning_rate": 1.7569950556517566e-05, | |
| "loss": 0.0834, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.005653579676674365, | |
| "grad_norm": 2.353285312652588, | |
| "learning_rate": 1.7547095802227723e-05, | |
| "loss": 0.1524, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.005672055427251732, | |
| "grad_norm": 3.8065831661224365, | |
| "learning_rate": 1.7524149088957244e-05, | |
| "loss": 0.1238, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.005690531177829099, | |
| "grad_norm": 8.202301979064941, | |
| "learning_rate": 1.7501110696304598e-05, | |
| "loss": 0.2229, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.005709006928406466, | |
| "grad_norm": 0.9660981297492981, | |
| "learning_rate": 1.747798090498532e-05, | |
| "loss": 0.1039, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.0057274826789838335, | |
| "grad_norm": 6.513055324554443, | |
| "learning_rate": 1.7454759996828622e-05, | |
| "loss": 0.3981, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.005745958429561201, | |
| "grad_norm": 0.712873637676239, | |
| "learning_rate": 1.7431448254773943e-05, | |
| "loss": 0.1074, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.005764434180138568, | |
| "grad_norm": 2.2026023864746094, | |
| "learning_rate": 1.74080459628675e-05, | |
| "loss": 0.124, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.005782909930715935, | |
| "grad_norm": 3.3080198764801025, | |
| "learning_rate": 1.7384553406258842e-05, | |
| "loss": 0.1579, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.005801385681293302, | |
| "grad_norm": 8.671109199523926, | |
| "learning_rate": 1.7360970871197347e-05, | |
| "loss": 0.2322, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.005819861431870669, | |
| "grad_norm": 1.1543076038360596, | |
| "learning_rate": 1.7337298645028764e-05, | |
| "loss": 0.1191, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.005838337182448037, | |
| "grad_norm": 0.8305059671401978, | |
| "learning_rate": 1.7313537016191706e-05, | |
| "loss": 0.1111, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.005856812933025404, | |
| "grad_norm": 0.7246889472007751, | |
| "learning_rate": 1.7289686274214116e-05, | |
| "loss": 0.1283, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.005875288683602771, | |
| "grad_norm": 3.9882657527923584, | |
| "learning_rate": 1.7265746709709762e-05, | |
| "loss": 0.1549, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.005893764434180138, | |
| "grad_norm": 1.6284881830215454, | |
| "learning_rate": 1.7241718614374678e-05, | |
| "loss": 0.1072, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.005912240184757506, | |
| "grad_norm": 8.891485214233398, | |
| "learning_rate": 1.7217602280983622e-05, | |
| "loss": 0.1899, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.005930715935334873, | |
| "grad_norm": 2.554180860519409, | |
| "learning_rate": 1.7193398003386514e-05, | |
| "loss": 0.1429, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.005949191685912241, | |
| "grad_norm": 0.7529696822166443, | |
| "learning_rate": 1.716910607650483e-05, | |
| "loss": 0.0992, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.005967667436489608, | |
| "grad_norm": 0.782574474811554, | |
| "learning_rate": 1.7144726796328034e-05, | |
| "loss": 0.1417, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.005986143187066975, | |
| "grad_norm": 7.975888252258301, | |
| "learning_rate": 1.712026045990997e-05, | |
| "loss": 0.3964, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.006004618937644342, | |
| "grad_norm": 0.980945885181427, | |
| "learning_rate": 1.709570736536521e-05, | |
| "loss": 0.1195, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.006023094688221709, | |
| "grad_norm": 3.4679291248321533, | |
| "learning_rate": 1.7071067811865477e-05, | |
| "loss": 0.1935, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.0060415704387990765, | |
| "grad_norm": 7.2458720207214355, | |
| "learning_rate": 1.7046342099635948e-05, | |
| "loss": 0.8158, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.006060046189376444, | |
| "grad_norm": 7.2671895027160645, | |
| "learning_rate": 1.7021530529951627e-05, | |
| "loss": 0.225, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.006078521939953811, | |
| "grad_norm": 0.7106726169586182, | |
| "learning_rate": 1.6996633405133656e-05, | |
| "loss": 0.0939, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.006096997690531178, | |
| "grad_norm": 0.7158485651016235, | |
| "learning_rate": 1.697165102854565e-05, | |
| "loss": 0.0982, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.006115473441108545, | |
| "grad_norm": 8.160673141479492, | |
| "learning_rate": 1.6946583704589973e-05, | |
| "loss": 0.4307, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.0061339491916859125, | |
| "grad_norm": 4.940622329711914, | |
| "learning_rate": 1.692143173870407e-05, | |
| "loss": 0.2084, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.00615242494226328, | |
| "grad_norm": 9.330638885498047, | |
| "learning_rate": 1.68961954373567e-05, | |
| "loss": 0.4745, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.006170900692840647, | |
| "grad_norm": 17.111072540283203, | |
| "learning_rate": 1.6870875108044233e-05, | |
| "loss": 0.6827, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.006189376443418014, | |
| "grad_norm": 3.640718460083008, | |
| "learning_rate": 1.684547105928689e-05, | |
| "loss": 0.2285, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.006207852193995381, | |
| "grad_norm": 1.2472453117370605, | |
| "learning_rate": 1.6819983600624986e-05, | |
| "loss": 0.1043, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.006226327944572748, | |
| "grad_norm": 1.2750074863433838, | |
| "learning_rate": 1.6794413042615168e-05, | |
| "loss": 0.1007, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.006244803695150116, | |
| "grad_norm": 3.740368127822876, | |
| "learning_rate": 1.6768759696826608e-05, | |
| "loss": 0.1659, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.006263279445727483, | |
| "grad_norm": 5.449174404144287, | |
| "learning_rate": 1.6743023875837233e-05, | |
| "loss": 0.4581, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.00628175519630485, | |
| "grad_norm": 5.260900974273682, | |
| "learning_rate": 1.6717205893229904e-05, | |
| "loss": 0.2102, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.006300230946882217, | |
| "grad_norm": 4.941977024078369, | |
| "learning_rate": 1.6691306063588583e-05, | |
| "loss": 0.2179, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.006318706697459584, | |
| "grad_norm": 7.90458869934082, | |
| "learning_rate": 1.6665324702494524e-05, | |
| "loss": 0.3013, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.0063371824480369515, | |
| "grad_norm": 9.964789390563965, | |
| "learning_rate": 1.6639262126522417e-05, | |
| "loss": 0.2579, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.006355658198614319, | |
| "grad_norm": 4.498533725738525, | |
| "learning_rate": 1.661311865323652e-05, | |
| "loss": 0.2532, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.006374133949191686, | |
| "grad_norm": 2.2392873764038086, | |
| "learning_rate": 1.6586894601186804e-05, | |
| "loss": 0.1666, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.006392609699769053, | |
| "grad_norm": 1.348547101020813, | |
| "learning_rate": 1.6560590289905074e-05, | |
| "loss": 0.1117, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.00641108545034642, | |
| "grad_norm": 3.4107110500335693, | |
| "learning_rate": 1.6534206039901057e-05, | |
| "loss": 0.1692, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.0064295612009237875, | |
| "grad_norm": 6.425887107849121, | |
| "learning_rate": 1.650774217265851e-05, | |
| "loss": 0.421, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.006448036951501155, | |
| "grad_norm": 3.297750234603882, | |
| "learning_rate": 1.6481199010631312e-05, | |
| "loss": 0.1761, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.006466512702078522, | |
| "grad_norm": 5.199448108673096, | |
| "learning_rate": 1.645457687723951e-05, | |
| "loss": 0.2639, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.006484988452655889, | |
| "grad_norm": 3.288149833679199, | |
| "learning_rate": 1.6427876096865394e-05, | |
| "loss": 0.1937, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.006503464203233256, | |
| "grad_norm": 0.9052378535270691, | |
| "learning_rate": 1.6401096994849558e-05, | |
| "loss": 0.0999, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.006521939953810623, | |
| "grad_norm": 4.848758697509766, | |
| "learning_rate": 1.63742398974869e-05, | |
| "loss": 0.2093, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.006540415704387991, | |
| "grad_norm": 1.5989115238189697, | |
| "learning_rate": 1.6347305132022677e-05, | |
| "loss": 0.0909, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.006558891454965358, | |
| "grad_norm": 5.258222579956055, | |
| "learning_rate": 1.632029302664851e-05, | |
| "loss": 0.1665, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.006577367205542725, | |
| "grad_norm": 2.0112485885620117, | |
| "learning_rate": 1.6293203910498375e-05, | |
| "loss": 0.132, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.006595842956120092, | |
| "grad_norm": 0.9058921337127686, | |
| "learning_rate": 1.6266038113644605e-05, | |
| "loss": 0.0973, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.006614318706697459, | |
| "grad_norm": 1.4918032884597778, | |
| "learning_rate": 1.6238795967093865e-05, | |
| "loss": 0.1061, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.0066327944572748265, | |
| "grad_norm": 1.350417137145996, | |
| "learning_rate": 1.6211477802783105e-05, | |
| "loss": 0.0994, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.006651270207852194, | |
| "grad_norm": 0.7838285565376282, | |
| "learning_rate": 1.6184083953575543e-05, | |
| "loss": 0.1058, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.006669745958429561, | |
| "grad_norm": 2.9641215801239014, | |
| "learning_rate": 1.6156614753256583e-05, | |
| "loss": 0.1525, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.006688221709006928, | |
| "grad_norm": 1.702981948852539, | |
| "learning_rate": 1.6129070536529767e-05, | |
| "loss": 0.1195, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.006706697459584295, | |
| "grad_norm": 0.7067826390266418, | |
| "learning_rate": 1.610145163901268e-05, | |
| "loss": 0.1153, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.0067251732101616625, | |
| "grad_norm": 0.5825372934341431, | |
| "learning_rate": 1.607375839723287e-05, | |
| "loss": 0.1081, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.00674364896073903, | |
| "grad_norm": 0.8679400682449341, | |
| "learning_rate": 1.6045991148623752e-05, | |
| "loss": 0.0859, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.006762124711316397, | |
| "grad_norm": 0.5362609624862671, | |
| "learning_rate": 1.6018150231520486e-05, | |
| "loss": 0.0833, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.006780600461893764, | |
| "grad_norm": 0.6094454526901245, | |
| "learning_rate": 1.599023598515586e-05, | |
| "loss": 0.0947, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.006799076212471131, | |
| "grad_norm": 0.698591947555542, | |
| "learning_rate": 1.5962248749656158e-05, | |
| "loss": 0.0992, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.006817551963048499, | |
| "grad_norm": 0.34043464064598083, | |
| "learning_rate": 1.5934188866037017e-05, | |
| "loss": 0.0647, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.0068360277136258665, | |
| "grad_norm": 11.83703327178955, | |
| "learning_rate": 1.5906056676199256e-05, | |
| "loss": 0.4663, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.006854503464203234, | |
| "grad_norm": 3.8346962928771973, | |
| "learning_rate": 1.5877852522924733e-05, | |
| "loss": 0.1536, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.006872979214780601, | |
| "grad_norm": 3.227011203765869, | |
| "learning_rate": 1.584957674987216e-05, | |
| "loss": 0.1455, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.006891454965357968, | |
| "grad_norm": 0.7077368497848511, | |
| "learning_rate": 1.5821229701572897e-05, | |
| "loss": 0.0872, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.006909930715935335, | |
| "grad_norm": 6.417845249176025, | |
| "learning_rate": 1.5792811723426787e-05, | |
| "loss": 0.18, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.006928406466512702, | |
| "grad_norm": 11.544971466064453, | |
| "learning_rate": 1.5764323161697933e-05, | |
| "loss": 0.6268, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.00694688221709007, | |
| "grad_norm": 1.5311028957366943, | |
| "learning_rate": 1.573576436351046e-05, | |
| "loss": 0.1065, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.006965357967667437, | |
| "grad_norm": 1.9619712829589844, | |
| "learning_rate": 1.570713567684432e-05, | |
| "loss": 0.1259, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.006983833718244804, | |
| "grad_norm": 9.510198593139648, | |
| "learning_rate": 1.5678437450531014e-05, | |
| "loss": 0.2707, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.007002309468822171, | |
| "grad_norm": 1.8756345510482788, | |
| "learning_rate": 1.564967003424938e-05, | |
| "loss": 0.1387, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.007020785219399538, | |
| "grad_norm": 16.97416114807129, | |
| "learning_rate": 1.5620833778521306e-05, | |
| "loss": 0.5556, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.0070392609699769055, | |
| "grad_norm": 3.3661611080169678, | |
| "learning_rate": 1.5591929034707468e-05, | |
| "loss": 0.1602, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.007057736720554273, | |
| "grad_norm": 1.2877082824707031, | |
| "learning_rate": 1.556295615500305e-05, | |
| "loss": 0.1069, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.00707621247113164, | |
| "grad_norm": 0.7841387391090393, | |
| "learning_rate": 1.553391549243344e-05, | |
| "loss": 0.1152, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.007094688221709007, | |
| "grad_norm": 1.9054657220840454, | |
| "learning_rate": 1.5504807400849957e-05, | |
| "loss": 0.1328, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.007113163972286374, | |
| "grad_norm": 6.285136699676514, | |
| "learning_rate": 1.5475632234925505e-05, | |
| "loss": 0.2663, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.0071316397228637415, | |
| "grad_norm": 1.3094302415847778, | |
| "learning_rate": 1.5446390350150272e-05, | |
| "loss": 0.093, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.007150115473441109, | |
| "grad_norm": 2.2979140281677246, | |
| "learning_rate": 1.54170821028274e-05, | |
| "loss": 0.1149, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.007168591224018476, | |
| "grad_norm": 4.052133560180664, | |
| "learning_rate": 1.5387707850068633e-05, | |
| "loss": 0.1317, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.007187066974595843, | |
| "grad_norm": 0.5509690642356873, | |
| "learning_rate": 1.5358267949789968e-05, | |
| "loss": 0.0897, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.00720554272517321, | |
| "grad_norm": 2.494608163833618, | |
| "learning_rate": 1.53287627607073e-05, | |
| "loss": 0.131, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.007224018475750577, | |
| "grad_norm": 3.893519401550293, | |
| "learning_rate": 1.529919264233205e-05, | |
| "loss": 0.1605, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.007242494226327945, | |
| "grad_norm": 0.6864449977874756, | |
| "learning_rate": 1.5269557954966777e-05, | |
| "loss": 0.0964, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.007260969976905312, | |
| "grad_norm": 0.8827430605888367, | |
| "learning_rate": 1.5239859059700794e-05, | |
| "loss": 0.0949, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.007279445727482679, | |
| "grad_norm": 10.671297073364258, | |
| "learning_rate": 1.5210096318405768e-05, | |
| "loss": 0.4285, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.007297921478060046, | |
| "grad_norm": 0.3259846866130829, | |
| "learning_rate": 1.5180270093731305e-05, | |
| "loss": 0.0776, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.007316397228637413, | |
| "grad_norm": 6.536836624145508, | |
| "learning_rate": 1.5150380749100545e-05, | |
| "loss": 0.2632, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.0073348729792147805, | |
| "grad_norm": 2.776939630508423, | |
| "learning_rate": 1.5120428648705716e-05, | |
| "loss": 0.1348, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.007353348729792148, | |
| "grad_norm": 1.3437862396240234, | |
| "learning_rate": 1.5090414157503715e-05, | |
| "loss": 0.1157, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.007371824480369515, | |
| "grad_norm": 8.23919677734375, | |
| "learning_rate": 1.5060337641211637e-05, | |
| "loss": 0.7675, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.007390300230946882, | |
| "grad_norm": 0.6749206781387329, | |
| "learning_rate": 1.5030199466302354e-05, | |
| "loss": 0.1067, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.007408775981524249, | |
| "grad_norm": 15.703801155090332, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 0.5354, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.0074272517321016165, | |
| "grad_norm": 0.32032838463783264, | |
| "learning_rate": 1.4969739610275556e-05, | |
| "loss": 0.0701, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.007445727482678984, | |
| "grad_norm": 0.8290475606918335, | |
| "learning_rate": 1.493941866584231e-05, | |
| "loss": 0.0845, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.007464203233256351, | |
| "grad_norm": 0.9650348424911499, | |
| "learning_rate": 1.490903753615141e-05, | |
| "loss": 0.1013, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.007482678983833718, | |
| "grad_norm": 1.0794436931610107, | |
| "learning_rate": 1.4878596591387329e-05, | |
| "loss": 0.0951, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.007501154734411085, | |
| "grad_norm": 1.1679524183273315, | |
| "learning_rate": 1.4848096202463373e-05, | |
| "loss": 0.1134, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.007519630484988452, | |
| "grad_norm": 0.45709890127182007, | |
| "learning_rate": 1.4817536741017153e-05, | |
| "loss": 0.0846, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.00753810623556582, | |
| "grad_norm": 10.355935096740723, | |
| "learning_rate": 1.478691857940607e-05, | |
| "loss": 0.2055, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.007556581986143187, | |
| "grad_norm": 1.6484851837158203, | |
| "learning_rate": 1.4756242090702756e-05, | |
| "loss": 0.1184, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.007575057736720554, | |
| "grad_norm": 1.2918403148651123, | |
| "learning_rate": 1.4725507648690542e-05, | |
| "loss": 0.1169, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.007593533487297921, | |
| "grad_norm": 0.49354541301727295, | |
| "learning_rate": 1.469471562785891e-05, | |
| "loss": 0.0774, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.007612009237875288, | |
| "grad_norm": 0.7458063364028931, | |
| "learning_rate": 1.4663866403398915e-05, | |
| "loss": 0.0994, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.0076304849884526555, | |
| "grad_norm": 9.849067687988281, | |
| "learning_rate": 1.463296035119862e-05, | |
| "loss": 0.4676, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.007648960739030023, | |
| "grad_norm": 12.387116432189941, | |
| "learning_rate": 1.4601997847838518e-05, | |
| "loss": 0.4014, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.00766743648960739, | |
| "grad_norm": 4.2863264083862305, | |
| "learning_rate": 1.4570979270586944e-05, | |
| "loss": 0.1361, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.007685912240184757, | |
| "grad_norm": 0.3984392583370209, | |
| "learning_rate": 1.4539904997395468e-05, | |
| "loss": 0.0924, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.007704387990762124, | |
| "grad_norm": 2.2181036472320557, | |
| "learning_rate": 1.4508775406894308e-05, | |
| "loss": 0.1226, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.007722863741339492, | |
| "grad_norm": 2.38778018951416, | |
| "learning_rate": 1.4477590878387697e-05, | |
| "loss": 0.1358, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.0077413394919168595, | |
| "grad_norm": 4.680348873138428, | |
| "learning_rate": 1.4446351791849276e-05, | |
| "loss": 0.2542, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.007759815242494227, | |
| "grad_norm": 2.908165693283081, | |
| "learning_rate": 1.4415058527917454e-05, | |
| "loss": 0.1301, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.007778290993071594, | |
| "grad_norm": 1.1746968030929565, | |
| "learning_rate": 1.4383711467890776e-05, | |
| "loss": 0.1107, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.007796766743648961, | |
| "grad_norm": 1.5787891149520874, | |
| "learning_rate": 1.4352310993723277e-05, | |
| "loss": 0.0956, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.007815242494226328, | |
| "grad_norm": 1.6053221225738525, | |
| "learning_rate": 1.4320857488019826e-05, | |
| "loss": 0.1095, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.007833718244803695, | |
| "grad_norm": 5.308477878570557, | |
| "learning_rate": 1.4289351334031461e-05, | |
| "loss": 0.216, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.007852193995381063, | |
| "grad_norm": 13.702173233032227, | |
| "learning_rate": 1.4257792915650728e-05, | |
| "loss": 0.2206, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.007870669745958429, | |
| "grad_norm": 3.6765360832214355, | |
| "learning_rate": 1.4226182617406996e-05, | |
| "loss": 0.1498, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.007889145496535797, | |
| "grad_norm": 1.3325588703155518, | |
| "learning_rate": 1.4194520824461773e-05, | |
| "loss": 0.0987, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.007907621247113163, | |
| "grad_norm": 1.9526255130767822, | |
| "learning_rate": 1.4162807922604014e-05, | |
| "loss": 0.1, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.007926096997690531, | |
| "grad_norm": 0.32403743267059326, | |
| "learning_rate": 1.413104429824542e-05, | |
| "loss": 0.0812, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.007944572748267898, | |
| "grad_norm": 0.4677944481372833, | |
| "learning_rate": 1.4099230338415728e-05, | |
| "loss": 0.0895, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.007963048498845266, | |
| "grad_norm": 7.707451343536377, | |
| "learning_rate": 1.4067366430758004e-05, | |
| "loss": 0.4918, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.007981524249422632, | |
| "grad_norm": 0.6431747674942017, | |
| "learning_rate": 1.4035452963523903e-05, | |
| "loss": 0.1029, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.008, | |
| "grad_norm": 1.304093599319458, | |
| "learning_rate": 1.4003490325568953e-05, | |
| "loss": 0.1125, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.008018475750577366, | |
| "grad_norm": 0.6481981873512268, | |
| "learning_rate": 1.3971478906347806e-05, | |
| "loss": 0.1036, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.008036951501154735, | |
| "grad_norm": 2.279773473739624, | |
| "learning_rate": 1.3939419095909513e-05, | |
| "loss": 0.1274, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.0080554272517321, | |
| "grad_norm": 3.684098243713379, | |
| "learning_rate": 1.3907311284892737e-05, | |
| "loss": 0.1866, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.008073903002309469, | |
| "grad_norm": 0.6002702116966248, | |
| "learning_rate": 1.3875155864521031e-05, | |
| "loss": 0.1043, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.008092378752886835, | |
| "grad_norm": 9.312369346618652, | |
| "learning_rate": 1.3842953226598036e-05, | |
| "loss": 0.7908, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.008110854503464203, | |
| "grad_norm": 1.6161905527114868, | |
| "learning_rate": 1.3810703763502744e-05, | |
| "loss": 0.1042, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.00812933025404157, | |
| "grad_norm": 8.988299369812012, | |
| "learning_rate": 1.3778407868184674e-05, | |
| "loss": 0.2041, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.008147806004618938, | |
| "grad_norm": 8.070473670959473, | |
| "learning_rate": 1.3746065934159123e-05, | |
| "loss": 0.4467, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.008166281755196304, | |
| "grad_norm": 0.9457348585128784, | |
| "learning_rate": 1.371367835550235e-05, | |
| "loss": 0.0938, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.008184757505773672, | |
| "grad_norm": 4.6980814933776855, | |
| "learning_rate": 1.3681245526846782e-05, | |
| "loss": 0.1515, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.00820323325635104, | |
| "grad_norm": 0.6849774122238159, | |
| "learning_rate": 1.3648767843376196e-05, | |
| "loss": 0.0875, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.008221709006928406, | |
| "grad_norm": 3.691784143447876, | |
| "learning_rate": 1.3616245700820922e-05, | |
| "loss": 0.1537, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.008240184757505774, | |
| "grad_norm": 3.544032573699951, | |
| "learning_rate": 1.3583679495453e-05, | |
| "loss": 0.129, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.00825866050808314, | |
| "grad_norm": 2.254021644592285, | |
| "learning_rate": 1.3551069624081372e-05, | |
| "loss": 0.1128, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.008277136258660509, | |
| "grad_norm": 5.0423173904418945, | |
| "learning_rate": 1.3518416484047018e-05, | |
| "loss": 0.2143, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.008295612009237875, | |
| "grad_norm": 0.44525039196014404, | |
| "learning_rate": 1.3485720473218153e-05, | |
| "loss": 0.0931, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.008314087759815243, | |
| "grad_norm": 10.2832612991333, | |
| "learning_rate": 1.3452981989985347e-05, | |
| "loss": 0.3233, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.00833256351039261, | |
| "grad_norm": 0.45719677209854126, | |
| "learning_rate": 1.342020143325669e-05, | |
| "loss": 0.0897, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.008351039260969978, | |
| "grad_norm": 7.944047451019287, | |
| "learning_rate": 1.3387379202452917e-05, | |
| "loss": 0.4925, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.008369515011547344, | |
| "grad_norm": 1.1281418800354004, | |
| "learning_rate": 1.3354515697502552e-05, | |
| "loss": 0.1093, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.008387990762124712, | |
| "grad_norm": 1.5809013843536377, | |
| "learning_rate": 1.3321611318837033e-05, | |
| "loss": 0.1361, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.008406466512702078, | |
| "grad_norm": 12.290745735168457, | |
| "learning_rate": 1.3288666467385834e-05, | |
| "loss": 0.26, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.008424942263279446, | |
| "grad_norm": 7.899825096130371, | |
| "learning_rate": 1.3255681544571568e-05, | |
| "loss": 0.4696, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.008443418013856813, | |
| "grad_norm": 2.0052428245544434, | |
| "learning_rate": 1.3222656952305113e-05, | |
| "loss": 0.1109, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.00846189376443418, | |
| "grad_norm": 0.4509698748588562, | |
| "learning_rate": 1.3189593092980701e-05, | |
| "loss": 0.074, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.008480369515011547, | |
| "grad_norm": 0.49958622455596924, | |
| "learning_rate": 1.3156490369471026e-05, | |
| "loss": 0.0931, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.008498845265588915, | |
| "grad_norm": 0.8973973989486694, | |
| "learning_rate": 1.3123349185122328e-05, | |
| "loss": 0.0997, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.008517321016166281, | |
| "grad_norm": 3.8769736289978027, | |
| "learning_rate": 1.3090169943749475e-05, | |
| "loss": 0.1911, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.00853579676674365, | |
| "grad_norm": 4.02487325668335, | |
| "learning_rate": 1.3056953049631059e-05, | |
| "loss": 0.1984, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.008554272517321016, | |
| "grad_norm": 3.804443120956421, | |
| "learning_rate": 1.3023698907504447e-05, | |
| "loss": 0.1543, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.008572748267898384, | |
| "grad_norm": 2.7212209701538086, | |
| "learning_rate": 1.2990407922560869e-05, | |
| "loss": 0.1743, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.00859122401847575, | |
| "grad_norm": 1.5701078176498413, | |
| "learning_rate": 1.2957080500440469e-05, | |
| "loss": 0.1072, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.008609699769053118, | |
| "grad_norm": 0.9316039681434631, | |
| "learning_rate": 1.2923717047227368e-05, | |
| "loss": 0.0943, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.008628175519630485, | |
| "grad_norm": 13.156824111938477, | |
| "learning_rate": 1.2890317969444716e-05, | |
| "loss": 0.6323, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.008646651270207853, | |
| "grad_norm": 3.8034515380859375, | |
| "learning_rate": 1.2856883674049736e-05, | |
| "loss": 0.1835, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.008665127020785219, | |
| "grad_norm": 2.9714114665985107, | |
| "learning_rate": 1.2823414568428767e-05, | |
| "loss": 0.159, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.008683602771362587, | |
| "grad_norm": 1.578321099281311, | |
| "learning_rate": 1.2789911060392295e-05, | |
| "loss": 0.114, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.008702078521939953, | |
| "grad_norm": 5.823441028594971, | |
| "learning_rate": 1.2756373558169992e-05, | |
| "loss": 0.226, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.008720554272517321, | |
| "grad_norm": 1.6854499578475952, | |
| "learning_rate": 1.2722802470405744e-05, | |
| "loss": 0.0995, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.008739030023094688, | |
| "grad_norm": 9.106096267700195, | |
| "learning_rate": 1.2689198206152657e-05, | |
| "loss": 0.452, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.008757505773672056, | |
| "grad_norm": 6.119338512420654, | |
| "learning_rate": 1.265556117486809e-05, | |
| "loss": 0.2219, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.008775981524249422, | |
| "grad_norm": 0.7591765522956848, | |
| "learning_rate": 1.2621891786408648e-05, | |
| "loss": 0.1049, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.00879445727482679, | |
| "grad_norm": 3.043548107147217, | |
| "learning_rate": 1.2588190451025209e-05, | |
| "loss": 0.1308, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.008812933025404156, | |
| "grad_norm": 6.669645309448242, | |
| "learning_rate": 1.2554457579357906e-05, | |
| "loss": 0.3306, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.008831408775981524, | |
| "grad_norm": 4.448661804199219, | |
| "learning_rate": 1.252069358243114e-05, | |
| "loss": 0.1436, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.00884988452655889, | |
| "grad_norm": 0.6678202152252197, | |
| "learning_rate": 1.2486898871648552e-05, | |
| "loss": 0.0871, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.008868360277136259, | |
| "grad_norm": 5.38719367980957, | |
| "learning_rate": 1.2453073858788027e-05, | |
| "loss": 0.2865, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.008886836027713625, | |
| "grad_norm": 2.346839666366577, | |
| "learning_rate": 1.2419218955996677e-05, | |
| "loss": 0.1249, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.008905311778290993, | |
| "grad_norm": 6.323116302490234, | |
| "learning_rate": 1.238533457578581e-05, | |
| "loss": 0.1331, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.00892378752886836, | |
| "grad_norm": 0.8868709802627563, | |
| "learning_rate": 1.23514211310259e-05, | |
| "loss": 0.0958, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.008942263279445728, | |
| "grad_norm": 6.789799690246582, | |
| "learning_rate": 1.2317479034941572e-05, | |
| "loss": 0.4795, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.008960739030023094, | |
| "grad_norm": 5.569411277770996, | |
| "learning_rate": 1.2283508701106559e-05, | |
| "loss": 0.1929, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.008979214780600462, | |
| "grad_norm": 1.181089997291565, | |
| "learning_rate": 1.2249510543438652e-05, | |
| "loss": 0.1146, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.008997690531177828, | |
| "grad_norm": 2.9746992588043213, | |
| "learning_rate": 1.2215484976194675e-05, | |
| "loss": 0.1367, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.009016166281755196, | |
| "grad_norm": 2.86682391166687, | |
| "learning_rate": 1.2181432413965428e-05, | |
| "loss": 0.1361, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.009034642032332563, | |
| "grad_norm": 0.6102558374404907, | |
| "learning_rate": 1.2147353271670634e-05, | |
| "loss": 0.0968, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.00905311778290993, | |
| "grad_norm": 11.491408348083496, | |
| "learning_rate": 1.211324796455389e-05, | |
| "loss": 0.2806, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.009071593533487297, | |
| "grad_norm": 1.4330095052719116, | |
| "learning_rate": 1.2079116908177592e-05, | |
| "loss": 0.1129, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.009090069284064665, | |
| "grad_norm": 2.116281270980835, | |
| "learning_rate": 1.2044960518417902e-05, | |
| "loss": 0.1513, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.009108545034642033, | |
| "grad_norm": 3.6523022651672363, | |
| "learning_rate": 1.2010779211459649e-05, | |
| "loss": 0.1662, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.0091270207852194, | |
| "grad_norm": 1.281662940979004, | |
| "learning_rate": 1.1976573403791263e-05, | |
| "loss": 0.1049, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.009145496535796768, | |
| "grad_norm": 6.462726593017578, | |
| "learning_rate": 1.194234351219972e-05, | |
| "loss": 0.1684, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.009163972286374134, | |
| "grad_norm": 11.459948539733887, | |
| "learning_rate": 1.190808995376545e-05, | |
| "loss": 0.2178, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.009182448036951502, | |
| "grad_norm": 1.3683007955551147, | |
| "learning_rate": 1.187381314585725e-05, | |
| "loss": 0.1179, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.009200923787528868, | |
| "grad_norm": 12.80216121673584, | |
| "learning_rate": 1.1839513506127202e-05, | |
| "loss": 0.4293, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.009219399538106236, | |
| "grad_norm": 5.037327289581299, | |
| "learning_rate": 1.1805191452505602e-05, | |
| "loss": 0.1729, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.009237875288683603, | |
| "grad_norm": 3.80422043800354, | |
| "learning_rate": 1.1770847403195836e-05, | |
| "loss": 0.1684, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.00925635103926097, | |
| "grad_norm": 0.3601294755935669, | |
| "learning_rate": 1.1736481776669307e-05, | |
| "loss": 0.098, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.009274826789838337, | |
| "grad_norm": 1.2909562587738037, | |
| "learning_rate": 1.1702094991660326e-05, | |
| "loss": 0.1163, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.009293302540415705, | |
| "grad_norm": 6.163942337036133, | |
| "learning_rate": 1.1667687467161025e-05, | |
| "loss": 0.1524, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.009311778290993071, | |
| "grad_norm": 0.7384135127067566, | |
| "learning_rate": 1.1633259622416224e-05, | |
| "loss": 0.0949, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.00933025404157044, | |
| "grad_norm": 8.905200958251953, | |
| "learning_rate": 1.159881187691835e-05, | |
| "loss": 0.3597, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.009348729792147806, | |
| "grad_norm": 0.6112732887268066, | |
| "learning_rate": 1.156434465040231e-05, | |
| "loss": 0.0827, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.009367205542725174, | |
| "grad_norm": 0.8789867162704468, | |
| "learning_rate": 1.1529858362840383e-05, | |
| "loss": 0.0921, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.00938568129330254, | |
| "grad_norm": 2.8823060989379883, | |
| "learning_rate": 1.1495353434437098e-05, | |
| "loss": 0.1376, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.009404157043879908, | |
| "grad_norm": 3.471586227416992, | |
| "learning_rate": 1.1460830285624119e-05, | |
| "loss": 0.1709, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.009422632794457275, | |
| "grad_norm": 1.043778896331787, | |
| "learning_rate": 1.1426289337055119e-05, | |
| "loss": 0.0822, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.009441108545034643, | |
| "grad_norm": 1.6313532590866089, | |
| "learning_rate": 1.1391731009600655e-05, | |
| "loss": 0.1232, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.009459584295612009, | |
| "grad_norm": 0.448840856552124, | |
| "learning_rate": 1.1357155724343046e-05, | |
| "loss": 0.0852, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.009478060046189377, | |
| "grad_norm": 0.6620725393295288, | |
| "learning_rate": 1.1322563902571227e-05, | |
| "loss": 0.1028, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.009496535796766743, | |
| "grad_norm": 0.613250732421875, | |
| "learning_rate": 1.128795596577563e-05, | |
| "loss": 0.0818, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.009515011547344111, | |
| "grad_norm": 0.8659865260124207, | |
| "learning_rate": 1.1253332335643043e-05, | |
| "loss": 0.0994, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.009533487297921478, | |
| "grad_norm": 17.253829956054688, | |
| "learning_rate": 1.1218693434051475e-05, | |
| "loss": 0.9983, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.009551963048498846, | |
| "grad_norm": 1.6251740455627441, | |
| "learning_rate": 1.1184039683065014e-05, | |
| "loss": 0.1193, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.009570438799076212, | |
| "grad_norm": 1.427280306816101, | |
| "learning_rate": 1.1149371504928667e-05, | |
| "loss": 0.1077, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.00958891454965358, | |
| "grad_norm": 0.9821875691413879, | |
| "learning_rate": 1.1114689322063255e-05, | |
| "loss": 0.0862, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.009607390300230946, | |
| "grad_norm": 5.466465950012207, | |
| "learning_rate": 1.1079993557060228e-05, | |
| "loss": 0.2611, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.009625866050808314, | |
| "grad_norm": 6.57417106628418, | |
| "learning_rate": 1.1045284632676535e-05, | |
| "loss": 0.3494, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.00964434180138568, | |
| "grad_norm": 0.6029220223426819, | |
| "learning_rate": 1.1010562971829464e-05, | |
| "loss": 0.0899, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.009662817551963049, | |
| "grad_norm": 0.4714032709598541, | |
| "learning_rate": 1.0975828997591496e-05, | |
| "loss": 0.0884, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.009681293302540415, | |
| "grad_norm": 0.9918659329414368, | |
| "learning_rate": 1.0941083133185146e-05, | |
| "loss": 0.0941, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.009699769053117783, | |
| "grad_norm": 6.702113151550293, | |
| "learning_rate": 1.0906325801977804e-05, | |
| "loss": 0.2365, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.00971824480369515, | |
| "grad_norm": 0.7268968820571899, | |
| "learning_rate": 1.0871557427476585e-05, | |
| "loss": 0.098, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.009736720554272518, | |
| "grad_norm": 1.0268120765686035, | |
| "learning_rate": 1.083677843332316e-05, | |
| "loss": 0.1, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.009755196304849884, | |
| "grad_norm": 0.5114659667015076, | |
| "learning_rate": 1.0801989243288588e-05, | |
| "loss": 0.0839, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.009773672055427252, | |
| "grad_norm": 9.7759370803833, | |
| "learning_rate": 1.0767190281268187e-05, | |
| "loss": 0.2115, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.009792147806004618, | |
| "grad_norm": 0.9417822360992432, | |
| "learning_rate": 1.0732381971276318e-05, | |
| "loss": 0.1081, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.009810623556581986, | |
| "grad_norm": 0.822449803352356, | |
| "learning_rate": 1.0697564737441254e-05, | |
| "loss": 0.0942, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.009829099307159353, | |
| "grad_norm": 0.6435971260070801, | |
| "learning_rate": 1.0662739004000005e-05, | |
| "loss": 0.0902, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.00984757505773672, | |
| "grad_norm": 7.9442219734191895, | |
| "learning_rate": 1.0627905195293135e-05, | |
| "loss": 0.2197, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.009866050808314087, | |
| "grad_norm": 0.9835085272789001, | |
| "learning_rate": 1.0593063735759619e-05, | |
| "loss": 0.0942, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.009884526558891455, | |
| "grad_norm": 0.8477045297622681, | |
| "learning_rate": 1.055821504993164e-05, | |
| "loss": 0.0984, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.009903002309468821, | |
| "grad_norm": 0.29623472690582275, | |
| "learning_rate": 1.0523359562429441e-05, | |
| "loss": 0.0788, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.00992147806004619, | |
| "grad_norm": 11.50942325592041, | |
| "learning_rate": 1.0488497697956134e-05, | |
| "loss": 0.3236, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.009939953810623556, | |
| "grad_norm": 10.585582733154297, | |
| "learning_rate": 1.0453629881292537e-05, | |
| "loss": 0.3335, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.009958429561200924, | |
| "grad_norm": 0.6936651468276978, | |
| "learning_rate": 1.0418756537291996e-05, | |
| "loss": 0.0979, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.00997690531177829, | |
| "grad_norm": 8.163126945495605, | |
| "learning_rate": 1.03838780908752e-05, | |
| "loss": 0.2133, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.009995381062355658, | |
| "grad_norm": 8.653374671936035, | |
| "learning_rate": 1.0348994967025012e-05, | |
| "loss": 0.5174, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.010013856812933026, | |
| "grad_norm": 1.7608228921890259, | |
| "learning_rate": 1.0314107590781284e-05, | |
| "loss": 0.0947, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.010032332563510393, | |
| "grad_norm": 1.3849074840545654, | |
| "learning_rate": 1.0279216387235691e-05, | |
| "loss": 0.1019, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.01005080831408776, | |
| "grad_norm": 3.2330820560455322, | |
| "learning_rate": 1.0244321781526533e-05, | |
| "loss": 0.1184, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.010069284064665127, | |
| "grad_norm": 10.652017593383789, | |
| "learning_rate": 1.0209424198833571e-05, | |
| "loss": 0.239, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.010087759815242495, | |
| "grad_norm": 3.7272324562072754, | |
| "learning_rate": 1.0174524064372837e-05, | |
| "loss": 0.1643, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.010106235565819861, | |
| "grad_norm": 2.1730306148529053, | |
| "learning_rate": 1.0139621803391454e-05, | |
| "loss": 0.1237, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.01012471131639723, | |
| "grad_norm": 4.45306396484375, | |
| "learning_rate": 1.010471784116246e-05, | |
| "loss": 0.1778, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.010143187066974596, | |
| "grad_norm": 1.7231372594833374, | |
| "learning_rate": 1.0069812602979617e-05, | |
| "loss": 0.1331, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.010161662817551964, | |
| "grad_norm": 0.9715551137924194, | |
| "learning_rate": 1.0034906514152239e-05, | |
| "loss": 0.0958, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.01018013856812933, | |
| "grad_norm": 3.9421510696411133, | |
| "learning_rate": 1e-05, | |
| "loss": 0.1446, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.010198614318706698, | |
| "grad_norm": 2.027198553085327, | |
| "learning_rate": 9.965093485847766e-06, | |
| "loss": 0.1317, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.010217090069284064, | |
| "grad_norm": 1.534792423248291, | |
| "learning_rate": 9.930187397020385e-06, | |
| "loss": 0.1015, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.010235565819861433, | |
| "grad_norm": 8.953399658203125, | |
| "learning_rate": 9.895282158837545e-06, | |
| "loss": 0.278, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.010254041570438799, | |
| "grad_norm": 7.772527694702148, | |
| "learning_rate": 9.860378196608549e-06, | |
| "loss": 0.6269, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.010272517321016167, | |
| "grad_norm": 0.5983942747116089, | |
| "learning_rate": 9.825475935627165e-06, | |
| "loss": 0.0908, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.010290993071593533, | |
| "grad_norm": 1.4128079414367676, | |
| "learning_rate": 9.790575801166432e-06, | |
| "loss": 0.1183, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.010309468822170901, | |
| "grad_norm": 1.1565468311309814, | |
| "learning_rate": 9.75567821847347e-06, | |
| "loss": 0.1058, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.010327944572748268, | |
| "grad_norm": 1.2901358604431152, | |
| "learning_rate": 9.720783612764314e-06, | |
| "loss": 0.1056, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.010346420323325636, | |
| "grad_norm": 9.584195137023926, | |
| "learning_rate": 9.685892409218718e-06, | |
| "loss": 0.3054, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.010364896073903002, | |
| "grad_norm": 4.921411991119385, | |
| "learning_rate": 9.651005032974994e-06, | |
| "loss": 0.1596, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.01038337182448037, | |
| "grad_norm": 6.909355640411377, | |
| "learning_rate": 9.616121909124801e-06, | |
| "loss": 0.1858, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.010401847575057736, | |
| "grad_norm": 0.5534290075302124, | |
| "learning_rate": 9.581243462708007e-06, | |
| "loss": 0.0885, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.010420323325635104, | |
| "grad_norm": 11.534533500671387, | |
| "learning_rate": 9.546370118707463e-06, | |
| "loss": 0.4811, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.01043879907621247, | |
| "grad_norm": 1.6030110120773315, | |
| "learning_rate": 9.511502302043867e-06, | |
| "loss": 0.1079, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.010457274826789839, | |
| "grad_norm": 1.1885261535644531, | |
| "learning_rate": 9.476640437570562e-06, | |
| "loss": 0.0864, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.010475750577367205, | |
| "grad_norm": 9.166067123413086, | |
| "learning_rate": 9.441784950068362e-06, | |
| "loss": 0.4275, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.010494226327944573, | |
| "grad_norm": 0.6943501234054565, | |
| "learning_rate": 9.406936264240386e-06, | |
| "loss": 0.0885, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.01051270207852194, | |
| "grad_norm": 1.8906638622283936, | |
| "learning_rate": 9.372094804706867e-06, | |
| "loss": 0.1083, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.010531177829099308, | |
| "grad_norm": 0.9647622108459473, | |
| "learning_rate": 9.337260996000002e-06, | |
| "loss": 0.1019, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.010549653579676674, | |
| "grad_norm": 6.013514995574951, | |
| "learning_rate": 9.302435262558748e-06, | |
| "loss": 0.2284, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.010568129330254042, | |
| "grad_norm": 6.332157611846924, | |
| "learning_rate": 9.267618028723687e-06, | |
| "loss": 0.2941, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.010586605080831408, | |
| "grad_norm": 2.301401376724243, | |
| "learning_rate": 9.232809718731815e-06, | |
| "loss": 0.1255, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.010605080831408776, | |
| "grad_norm": 2.704345464706421, | |
| "learning_rate": 9.198010756711413e-06, | |
| "loss": 0.1445, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.010623556581986143, | |
| "grad_norm": 1.5195242166519165, | |
| "learning_rate": 9.163221566676847e-06, | |
| "loss": 0.1178, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.01064203233256351, | |
| "grad_norm": 1.2829123735427856, | |
| "learning_rate": 9.128442572523418e-06, | |
| "loss": 0.1073, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.010660508083140877, | |
| "grad_norm": 4.345551490783691, | |
| "learning_rate": 9.093674198022201e-06, | |
| "loss": 0.1879, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.010678983833718245, | |
| "grad_norm": 1.8225771188735962, | |
| "learning_rate": 9.058916866814857e-06, | |
| "loss": 0.1192, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.010697459584295611, | |
| "grad_norm": 4.983187675476074, | |
| "learning_rate": 9.024171002408507e-06, | |
| "loss": 0.2073, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.01071593533487298, | |
| "grad_norm": 5.874568939208984, | |
| "learning_rate": 8.989437028170537e-06, | |
| "loss": 0.1937, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.010734411085450346, | |
| "grad_norm": 1.2565988302230835, | |
| "learning_rate": 8.954715367323468e-06, | |
| "loss": 0.0909, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.010752886836027714, | |
| "grad_norm": 0.9957895874977112, | |
| "learning_rate": 8.920006442939772e-06, | |
| "loss": 0.0833, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.01077136258660508, | |
| "grad_norm": 0.40146133303642273, | |
| "learning_rate": 8.885310677936746e-06, | |
| "loss": 0.0719, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.010789838337182448, | |
| "grad_norm": 12.554910659790039, | |
| "learning_rate": 8.850628495071336e-06, | |
| "loss": 0.1697, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.010808314087759814, | |
| "grad_norm": 0.5177097916603088, | |
| "learning_rate": 8.815960316934991e-06, | |
| "loss": 0.0902, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.010826789838337183, | |
| "grad_norm": 3.0518546104431152, | |
| "learning_rate": 8.781306565948528e-06, | |
| "loss": 0.1316, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.010845265588914549, | |
| "grad_norm": 8.815834045410156, | |
| "learning_rate": 8.746667664356957e-06, | |
| "loss": 0.3372, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.010863741339491917, | |
| "grad_norm": 0.6400339007377625, | |
| "learning_rate": 8.712044034224374e-06, | |
| "loss": 0.0783, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.010882217090069283, | |
| "grad_norm": 0.2895418107509613, | |
| "learning_rate": 8.677436097428775e-06, | |
| "loss": 0.0757, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.010900692840646651, | |
| "grad_norm": 0.572730541229248, | |
| "learning_rate": 8.642844275656957e-06, | |
| "loss": 0.086, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.01091916859122402, | |
| "grad_norm": 1.938887119293213, | |
| "learning_rate": 8.60826899039935e-06, | |
| "loss": 0.1022, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.010937644341801386, | |
| "grad_norm": 0.760094404220581, | |
| "learning_rate": 8.573710662944884e-06, | |
| "loss": 0.0829, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.010956120092378754, | |
| "grad_norm": 2.0691940784454346, | |
| "learning_rate": 8.539169714375885e-06, | |
| "loss": 0.1147, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.01097459584295612, | |
| "grad_norm": 1.0790815353393555, | |
| "learning_rate": 8.504646565562907e-06, | |
| "loss": 0.0866, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.010993071593533488, | |
| "grad_norm": 7.380266189575195, | |
| "learning_rate": 8.47014163715962e-06, | |
| "loss": 0.1926, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.011011547344110854, | |
| "grad_norm": 14.631983757019043, | |
| "learning_rate": 8.43565534959769e-06, | |
| "loss": 0.2003, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.011030023094688222, | |
| "grad_norm": 7.973918437957764, | |
| "learning_rate": 8.401188123081653e-06, | |
| "loss": 0.2192, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.011048498845265589, | |
| "grad_norm": 0.7718592286109924, | |
| "learning_rate": 8.366740377583781e-06, | |
| "loss": 0.0861, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.011066974595842957, | |
| "grad_norm": 2.02675199508667, | |
| "learning_rate": 8.332312532838978e-06, | |
| "loss": 0.1166, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.011085450346420323, | |
| "grad_norm": 1.6027897596359253, | |
| "learning_rate": 8.297905008339677e-06, | |
| "loss": 0.1036, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.011103926096997691, | |
| "grad_norm": 0.9336921572685242, | |
| "learning_rate": 8.263518223330698e-06, | |
| "loss": 0.0879, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.011122401847575058, | |
| "grad_norm": 2.2634849548339844, | |
| "learning_rate": 8.22915259680417e-06, | |
| "loss": 0.1321, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.011140877598152426, | |
| "grad_norm": 0.44651880860328674, | |
| "learning_rate": 8.194808547494401e-06, | |
| "loss": 0.0752, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.011159353348729792, | |
| "grad_norm": 0.3159261643886566, | |
| "learning_rate": 8.1604864938728e-06, | |
| "loss": 0.0741, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.01117782909930716, | |
| "grad_norm": 3.665895700454712, | |
| "learning_rate": 8.126186854142752e-06, | |
| "loss": 0.1609, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.011196304849884526, | |
| "grad_norm": 8.490496635437012, | |
| "learning_rate": 8.091910046234552e-06, | |
| "loss": 0.4034, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.011214780600461894, | |
| "grad_norm": 2.5031774044036865, | |
| "learning_rate": 8.057656487800283e-06, | |
| "loss": 0.0967, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.01123325635103926, | |
| "grad_norm": 5.984310150146484, | |
| "learning_rate": 8.023426596208739e-06, | |
| "loss": 0.2662, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.011251732101616629, | |
| "grad_norm": 8.820603370666504, | |
| "learning_rate": 7.989220788540356e-06, | |
| "loss": 0.2424, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.011270207852193995, | |
| "grad_norm": 6.528566360473633, | |
| "learning_rate": 7.955039481582098e-06, | |
| "loss": 0.1791, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.011288683602771363, | |
| "grad_norm": 2.134223461151123, | |
| "learning_rate": 7.92088309182241e-06, | |
| "loss": 0.1031, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.01130715935334873, | |
| "grad_norm": 9.080251693725586, | |
| "learning_rate": 7.886752035446116e-06, | |
| "loss": 0.2293, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.011325635103926097, | |
| "grad_norm": 0.5130392909049988, | |
| "learning_rate": 7.852646728329368e-06, | |
| "loss": 0.0783, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.011344110854503464, | |
| "grad_norm": 0.8193851113319397, | |
| "learning_rate": 7.818567586034578e-06, | |
| "loss": 0.0875, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.011362586605080832, | |
| "grad_norm": 0.707890510559082, | |
| "learning_rate": 7.784515023805328e-06, | |
| "loss": 0.0877, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.011381062355658198, | |
| "grad_norm": 1.2535287141799927, | |
| "learning_rate": 7.750489456561351e-06, | |
| "loss": 0.1405, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.011399538106235566, | |
| "grad_norm": 0.479712575674057, | |
| "learning_rate": 7.716491298893443e-06, | |
| "loss": 0.0761, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.011418013856812933, | |
| "grad_norm": 16.281028747558594, | |
| "learning_rate": 7.68252096505843e-06, | |
| "loss": 0.3992, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.0114364896073903, | |
| "grad_norm": 0.8890343308448792, | |
| "learning_rate": 7.6485788689741e-06, | |
| "loss": 0.0958, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.011454965357967667, | |
| "grad_norm": 15.309280395507812, | |
| "learning_rate": 7.6146654242141935e-06, | |
| "loss": 0.3415, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.011473441108545035, | |
| "grad_norm": 10.811676979064941, | |
| "learning_rate": 7.580781044003324e-06, | |
| "loss": 0.3394, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.011491916859122401, | |
| "grad_norm": 2.8846611976623535, | |
| "learning_rate": 7.546926141211975e-06, | |
| "loss": 0.1482, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.01151039260969977, | |
| "grad_norm": 16.99440574645996, | |
| "learning_rate": 7.513101128351454e-06, | |
| "loss": 0.6543, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.011528868360277136, | |
| "grad_norm": 3.191331624984741, | |
| "learning_rate": 7.4793064175688635e-06, | |
| "loss": 0.1311, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.011547344110854504, | |
| "grad_norm": 2.9095370769500732, | |
| "learning_rate": 7.445542420642097e-06, | |
| "loss": 0.1495, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.01156581986143187, | |
| "grad_norm": 9.540557861328125, | |
| "learning_rate": 7.411809548974792e-06, | |
| "loss": 0.3095, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.011584295612009238, | |
| "grad_norm": 4.8060407638549805, | |
| "learning_rate": 7.378108213591355e-06, | |
| "loss": 0.1513, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.011602771362586604, | |
| "grad_norm": 1.2817132472991943, | |
| "learning_rate": 7.344438825131912e-06, | |
| "loss": 0.1095, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.011621247113163972, | |
| "grad_norm": 1.6703814268112183, | |
| "learning_rate": 7.310801793847344e-06, | |
| "loss": 0.1133, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.011639722863741339, | |
| "grad_norm": 11.130012512207031, | |
| "learning_rate": 7.277197529594257e-06, | |
| "loss": 0.2714, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.011658198614318707, | |
| "grad_norm": 3.7128093242645264, | |
| "learning_rate": 7.243626441830009e-06, | |
| "loss": 0.1928, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.011676674364896073, | |
| "grad_norm": 7.224024772644043, | |
| "learning_rate": 7.210088939607709e-06, | |
| "loss": 0.2897, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.011695150115473441, | |
| "grad_norm": 11.834304809570312, | |
| "learning_rate": 7.176585431571235e-06, | |
| "loss": 0.1282, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.011713625866050808, | |
| "grad_norm": 8.384239196777344, | |
| "learning_rate": 7.143116325950266e-06, | |
| "loss": 0.2317, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.011732101616628176, | |
| "grad_norm": 13.947283744812012, | |
| "learning_rate": 7.109682030555283e-06, | |
| "loss": 0.4589, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.011750577367205542, | |
| "grad_norm": 0.6878511905670166, | |
| "learning_rate": 7.076282952772634e-06, | |
| "loss": 0.0887, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.01176905311778291, | |
| "grad_norm": 6.7383713722229, | |
| "learning_rate": 7.042919499559538e-06, | |
| "loss": 0.694, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.011787528868360276, | |
| "grad_norm": 0.42635056376457214, | |
| "learning_rate": 7.009592077439135e-06, | |
| "loss": 0.084, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.011806004618937644, | |
| "grad_norm": 3.8634238243103027, | |
| "learning_rate": 6.976301092495556e-06, | |
| "loss": 0.1589, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.011824480369515012, | |
| "grad_norm": 0.5742267966270447, | |
| "learning_rate": 6.943046950368944e-06, | |
| "loss": 0.0829, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.011842956120092379, | |
| "grad_norm": 0.9647181034088135, | |
| "learning_rate": 6.909830056250527e-06, | |
| "loss": 0.1078, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.011861431870669747, | |
| "grad_norm": 16.41577911376953, | |
| "learning_rate": 6.876650814877675e-06, | |
| "loss": 0.5631, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.011879907621247113, | |
| "grad_norm": 5.8285322189331055, | |
| "learning_rate": 6.843509630528977e-06, | |
| "loss": 0.2127, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.011898383371824481, | |
| "grad_norm": 0.34135380387306213, | |
| "learning_rate": 6.8104069070193e-06, | |
| "loss": 0.0773, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.011916859122401847, | |
| "grad_norm": 2.6711978912353516, | |
| "learning_rate": 6.777343047694891e-06, | |
| "loss": 0.1053, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.011935334872979216, | |
| "grad_norm": 1.0284961462020874, | |
| "learning_rate": 6.744318455428436e-06, | |
| "loss": 0.0888, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.011953810623556582, | |
| "grad_norm": 0.6753608584403992, | |
| "learning_rate": 6.711333532614168e-06, | |
| "loss": 0.0959, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.01197228637413395, | |
| "grad_norm": 0.40900006890296936, | |
| "learning_rate": 6.67838868116297e-06, | |
| "loss": 0.0796, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.011990762124711316, | |
| "grad_norm": 5.518448352813721, | |
| "learning_rate": 6.645484302497452e-06, | |
| "loss": 0.1358, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.012009237875288684, | |
| "grad_norm": 0.5325848460197449, | |
| "learning_rate": 6.612620797547087e-06, | |
| "loss": 0.0722, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.01202771362586605, | |
| "grad_norm": 0.9748715758323669, | |
| "learning_rate": 6.579798566743314e-06, | |
| "loss": 0.0979, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.012046189376443419, | |
| "grad_norm": 1.7126754522323608, | |
| "learning_rate": 6.547018010014654e-06, | |
| "loss": 0.1143, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.012064665127020785, | |
| "grad_norm": 1.0465246438980103, | |
| "learning_rate": 6.5142795267818505e-06, | |
| "loss": 0.0917, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.012083140877598153, | |
| "grad_norm": 12.524375915527344, | |
| "learning_rate": 6.481583515952983e-06, | |
| "loss": 0.2337, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.01210161662817552, | |
| "grad_norm": 1.395440697669983, | |
| "learning_rate": 6.448930375918632e-06, | |
| "loss": 0.0888, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.012120092378752887, | |
| "grad_norm": 0.7288289666175842, | |
| "learning_rate": 6.4163205045469975e-06, | |
| "loss": 0.0936, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.012138568129330254, | |
| "grad_norm": 0.4729667007923126, | |
| "learning_rate": 6.383754299179079e-06, | |
| "loss": 0.0774, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.012157043879907622, | |
| "grad_norm": 1.5001037120819092, | |
| "learning_rate": 6.351232156623803e-06, | |
| "loss": 0.1056, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.012175519630484988, | |
| "grad_norm": 21.4486083984375, | |
| "learning_rate": 6.318754473153221e-06, | |
| "loss": 0.4898, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.012193995381062356, | |
| "grad_norm": 0.7856937050819397, | |
| "learning_rate": 6.286321644497655e-06, | |
| "loss": 0.0867, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.012212471131639722, | |
| "grad_norm": 0.6097049117088318, | |
| "learning_rate": 6.25393406584088e-06, | |
| "loss": 0.0904, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.01223094688221709, | |
| "grad_norm": 1.0945930480957031, | |
| "learning_rate": 6.22159213181533e-06, | |
| "loss": 0.0937, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.012249422632794457, | |
| "grad_norm": 0.6195545792579651, | |
| "learning_rate": 6.18929623649726e-06, | |
| "loss": 0.0861, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.012267898383371825, | |
| "grad_norm": 8.916356086730957, | |
| "learning_rate": 6.157046773401964e-06, | |
| "loss": 0.178, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.012286374133949191, | |
| "grad_norm": 1.3142259120941162, | |
| "learning_rate": 6.124844135478971e-06, | |
| "loss": 0.0881, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.01230484988452656, | |
| "grad_norm": 8.279459953308105, | |
| "learning_rate": 6.092688715107265e-06, | |
| "loss": 0.3659, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.012323325635103926, | |
| "grad_norm": 1.7175309658050537, | |
| "learning_rate": 6.06058090409049e-06, | |
| "loss": 0.1066, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.012341801385681294, | |
| "grad_norm": 9.963860511779785, | |
| "learning_rate": 6.028521093652195e-06, | |
| "loss": 0.2864, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.01236027713625866, | |
| "grad_norm": 2.892608165740967, | |
| "learning_rate": 5.996509674431053e-06, | |
| "loss": 0.1223, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.012378752886836028, | |
| "grad_norm": 10.879988670349121, | |
| "learning_rate": 5.9645470364761e-06, | |
| "loss": 0.3824, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.012397228637413394, | |
| "grad_norm": 9.321985244750977, | |
| "learning_rate": 5.932633569242e-06, | |
| "loss": 0.5043, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.012415704387990762, | |
| "grad_norm": 2.3833630084991455, | |
| "learning_rate": 5.900769661584273e-06, | |
| "loss": 0.1277, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.012434180138568129, | |
| "grad_norm": 10.050370216369629, | |
| "learning_rate": 5.868955701754584e-06, | |
| "loss": 0.2046, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.012452655889145497, | |
| "grad_norm": 1.4900848865509033, | |
| "learning_rate": 5.83719207739599e-06, | |
| "loss": 0.1096, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.012471131639722863, | |
| "grad_norm": 0.642914891242981, | |
| "learning_rate": 5.8054791755382286e-06, | |
| "loss": 0.0967, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.012489607390300231, | |
| "grad_norm": 1.1943944692611694, | |
| "learning_rate": 5.773817382593008e-06, | |
| "loss": 0.1016, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.012508083140877597, | |
| "grad_norm": 0.7675429582595825, | |
| "learning_rate": 5.742207084349274e-06, | |
| "loss": 0.0943, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.012526558891454966, | |
| "grad_norm": 1.9565619230270386, | |
| "learning_rate": 5.710648665968543e-06, | |
| "loss": 0.1023, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.012545034642032332, | |
| "grad_norm": 0.4500929117202759, | |
| "learning_rate": 5.679142511980176e-06, | |
| "loss": 0.0872, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.0125635103926097, | |
| "grad_norm": 3.7682559490203857, | |
| "learning_rate": 5.647689006276727e-06, | |
| "loss": 0.0988, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.012581986143187066, | |
| "grad_norm": 1.6648107767105103, | |
| "learning_rate": 5.616288532109225e-06, | |
| "loss": 0.103, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.012600461893764434, | |
| "grad_norm": 0.41996586322784424, | |
| "learning_rate": 5.584941472082549e-06, | |
| "loss": 0.0801, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.0126189376443418, | |
| "grad_norm": 0.7385615706443787, | |
| "learning_rate": 5.553648208150728e-06, | |
| "loss": 0.0868, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.012637413394919169, | |
| "grad_norm": 5.990048408508301, | |
| "learning_rate": 5.522409121612304e-06, | |
| "loss": 0.1866, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.012655889145496535, | |
| "grad_norm": 2.873253345489502, | |
| "learning_rate": 5.491224593105695e-06, | |
| "loss": 0.1006, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.012674364896073903, | |
| "grad_norm": 8.026762962341309, | |
| "learning_rate": 5.460095002604533e-06, | |
| "loss": 0.2686, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.01269284064665127, | |
| "grad_norm": 10.281550407409668, | |
| "learning_rate": 5.429020729413062e-06, | |
| "loss": 0.4579, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.012711316397228637, | |
| "grad_norm": 4.544410705566406, | |
| "learning_rate": 5.398002152161484e-06, | |
| "loss": 0.1284, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.012729792147806005, | |
| "grad_norm": 0.4521482586860657, | |
| "learning_rate": 5.367039648801386e-06, | |
| "loss": 0.0689, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.012748267898383372, | |
| "grad_norm": 2.0848639011383057, | |
| "learning_rate": 5.336133596601089e-06, | |
| "loss": 0.1283, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.01276674364896074, | |
| "grad_norm": 0.5293313264846802, | |
| "learning_rate": 5.305284372141095e-06, | |
| "loss": 0.0793, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.012785219399538106, | |
| "grad_norm": 9.645563125610352, | |
| "learning_rate": 5.274492351309462e-06, | |
| "loss": 0.1992, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.012803695150115474, | |
| "grad_norm": 1.6959121227264404, | |
| "learning_rate": 5.243757909297247e-06, | |
| "loss": 0.1281, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.01282217090069284, | |
| "grad_norm": 5.802919864654541, | |
| "learning_rate": 5.213081420593933e-06, | |
| "loss": 0.1692, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.012840646651270209, | |
| "grad_norm": 0.36798590421676636, | |
| "learning_rate": 5.1824632589828465e-06, | |
| "loss": 0.0798, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.012859122401847575, | |
| "grad_norm": 0.759617030620575, | |
| "learning_rate": 5.151903797536631e-06, | |
| "loss": 0.0789, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.012877598152424943, | |
| "grad_norm": 0.5548503994941711, | |
| "learning_rate": 5.121403408612672e-06, | |
| "loss": 0.0917, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.01289607390300231, | |
| "grad_norm": 7.482494831085205, | |
| "learning_rate": 5.090962463848592e-06, | |
| "loss": 0.1796, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.012914549653579677, | |
| "grad_norm": 2.071514368057251, | |
| "learning_rate": 5.060581334157693e-06, | |
| "loss": 0.1122, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.012933025404157044, | |
| "grad_norm": 2.0072550773620605, | |
| "learning_rate": 5.030260389724447e-06, | |
| "loss": 0.1266, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.012951501154734412, | |
| "grad_norm": 13.8541841506958, | |
| "learning_rate": 5.000000000000003e-06, | |
| "loss": 0.352, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.012969976905311778, | |
| "grad_norm": 6.458776950836182, | |
| "learning_rate": 4.96980053369765e-06, | |
| "loss": 0.1756, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.012988452655889146, | |
| "grad_norm": 0.46050941944122314, | |
| "learning_rate": 4.939662358788364e-06, | |
| "loss": 0.0788, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.013006928406466512, | |
| "grad_norm": 2.7678890228271484, | |
| "learning_rate": 4.909585842496287e-06, | |
| "loss": 0.1282, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.01302540415704388, | |
| "grad_norm": 0.29941099882125854, | |
| "learning_rate": 4.879571351294287e-06, | |
| "loss": 0.078, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.013043879907621247, | |
| "grad_norm": 1.6738927364349365, | |
| "learning_rate": 4.849619250899458e-06, | |
| "loss": 0.0958, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.013062355658198615, | |
| "grad_norm": 11.580183029174805, | |
| "learning_rate": 4.8197299062687e-06, | |
| "loss": 0.1625, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.013080831408775981, | |
| "grad_norm": 2.496164321899414, | |
| "learning_rate": 4.78990368159424e-06, | |
| "loss": 0.1214, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.01309930715935335, | |
| "grad_norm": 0.9565783143043518, | |
| "learning_rate": 4.76014094029921e-06, | |
| "loss": 0.0918, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.013117782909930716, | |
| "grad_norm": 2.017683982849121, | |
| "learning_rate": 4.7304420450332244e-06, | |
| "loss": 0.1035, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.013136258660508084, | |
| "grad_norm": 6.559094429016113, | |
| "learning_rate": 4.700807357667953e-06, | |
| "loss": 0.1821, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.01315473441108545, | |
| "grad_norm": 2.4273195266723633, | |
| "learning_rate": 4.671237239292699e-06, | |
| "loss": 0.122, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.013173210161662818, | |
| "grad_norm": 10.825758934020996, | |
| "learning_rate": 4.641732050210032e-06, | |
| "loss": 0.4601, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.013191685912240184, | |
| "grad_norm": 1.1227682828903198, | |
| "learning_rate": 4.612292149931369e-06, | |
| "loss": 0.0855, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.013210161662817552, | |
| "grad_norm": 0.5358126163482666, | |
| "learning_rate": 4.582917897172603e-06, | |
| "loss": 0.0882, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.013228637413394919, | |
| "grad_norm": 1.6175380945205688, | |
| "learning_rate": 4.5536096498497295e-06, | |
| "loss": 0.1108, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.013247113163972287, | |
| "grad_norm": 0.41255176067352295, | |
| "learning_rate": 4.524367765074499e-06, | |
| "loss": 0.0829, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.013265588914549653, | |
| "grad_norm": 9.501791954040527, | |
| "learning_rate": 4.495192599150045e-06, | |
| "loss": 0.2043, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.013284064665127021, | |
| "grad_norm": 5.367221355438232, | |
| "learning_rate": 4.46608450756656e-06, | |
| "loss": 0.1379, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.013302540415704387, | |
| "grad_norm": 0.407569944858551, | |
| "learning_rate": 4.437043844996952e-06, | |
| "loss": 0.0871, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.013321016166281755, | |
| "grad_norm": 0.3360728621482849, | |
| "learning_rate": 4.408070965292534e-06, | |
| "loss": 0.0786, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.013339491916859122, | |
| "grad_norm": 0.6986740231513977, | |
| "learning_rate": 4.379166221478697e-06, | |
| "loss": 0.0812, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.01335796766743649, | |
| "grad_norm": 0.5860750675201416, | |
| "learning_rate": 4.350329965750622e-06, | |
| "loss": 0.093, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.013376443418013856, | |
| "grad_norm": 0.45176026225090027, | |
| "learning_rate": 4.321562549468991e-06, | |
| "loss": 0.0717, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.013394919168591224, | |
| "grad_norm": 1.6555050611495972, | |
| "learning_rate": 4.292864323155684e-06, | |
| "loss": 0.0848, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.01341339491916859, | |
| "grad_norm": 3.6132729053497314, | |
| "learning_rate": 4.264235636489542e-06, | |
| "loss": 0.1455, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.013431870669745959, | |
| "grad_norm": 1.2024433612823486, | |
| "learning_rate": 4.235676838302069e-06, | |
| "loss": 0.0887, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.013450346420323325, | |
| "grad_norm": 10.77360725402832, | |
| "learning_rate": 4.207188276573214e-06, | |
| "loss": 0.3715, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.013468822170900693, | |
| "grad_norm": 4.56793737411499, | |
| "learning_rate": 4.178770298427107e-06, | |
| "loss": 0.1556, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.01348729792147806, | |
| "grad_norm": 0.8813223242759705, | |
| "learning_rate": 4.150423250127846e-06, | |
| "loss": 0.0945, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.013505773672055427, | |
| "grad_norm": 0.5578274130821228, | |
| "learning_rate": 4.12214747707527e-06, | |
| "loss": 0.0906, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.013524249422632794, | |
| "grad_norm": 1.011737585067749, | |
| "learning_rate": 4.093943323800746e-06, | |
| "loss": 0.0859, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.013542725173210162, | |
| "grad_norm": 0.6939584612846375, | |
| "learning_rate": 4.065811133962987e-06, | |
| "loss": 0.1084, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.013561200923787528, | |
| "grad_norm": 0.6531679630279541, | |
| "learning_rate": 4.037751250343841e-06, | |
| "loss": 0.0836, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.013579676674364896, | |
| "grad_norm": 22.63693618774414, | |
| "learning_rate": 4.009764014844143e-06, | |
| "loss": 0.8514, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.013598152424942262, | |
| "grad_norm": 0.3238315284252167, | |
| "learning_rate": 3.981849768479516e-06, | |
| "loss": 0.0729, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.01361662817551963, | |
| "grad_norm": 0.3579612970352173, | |
| "learning_rate": 3.954008851376252e-06, | |
| "loss": 0.0804, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.013635103926096999, | |
| "grad_norm": 0.531743049621582, | |
| "learning_rate": 3.9262416027671354e-06, | |
| "loss": 0.0937, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.013653579676674365, | |
| "grad_norm": 1.2231431007385254, | |
| "learning_rate": 3.898548360987325e-06, | |
| "loss": 0.1195, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.013672055427251733, | |
| "grad_norm": 9.986356735229492, | |
| "learning_rate": 3.8709294634702374e-06, | |
| "loss": 0.1768, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.0136905311778291, | |
| "grad_norm": 7.637360095977783, | |
| "learning_rate": 3.8433852467434175e-06, | |
| "loss": 0.1729, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.013709006928406467, | |
| "grad_norm": 0.7323989868164062, | |
| "learning_rate": 3.81591604642446e-06, | |
| "loss": 0.0891, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.013727482678983834, | |
| "grad_norm": 15.846328735351562, | |
| "learning_rate": 3.7885221972168974e-06, | |
| "loss": 0.3992, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.013745958429561202, | |
| "grad_norm": 1.9192430973052979, | |
| "learning_rate": 3.7612040329061405e-06, | |
| "loss": 0.0975, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.013764434180138568, | |
| "grad_norm": 1.4047856330871582, | |
| "learning_rate": 3.7339618863553983e-06, | |
| "loss": 0.0973, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.013782909930715936, | |
| "grad_norm": 0.745542049407959, | |
| "learning_rate": 3.7067960895016277e-06, | |
| "loss": 0.0867, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.013801385681293302, | |
| "grad_norm": 3.5601065158843994, | |
| "learning_rate": 3.679706973351491e-06, | |
| "loss": 0.1498, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.01381986143187067, | |
| "grad_norm": 5.545723915100098, | |
| "learning_rate": 3.6526948679773256e-06, | |
| "loss": 0.2211, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.013838337182448037, | |
| "grad_norm": 2.100994110107422, | |
| "learning_rate": 3.625760102513103e-06, | |
| "loss": 0.101, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.013856812933025405, | |
| "grad_norm": 0.7321968078613281, | |
| "learning_rate": 3.598903005150444e-06, | |
| "loss": 0.0941, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.013875288683602771, | |
| "grad_norm": 16.122934341430664, | |
| "learning_rate": 3.5721239031346067e-06, | |
| "loss": 1.0905, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.01389376443418014, | |
| "grad_norm": 0.8299577236175537, | |
| "learning_rate": 3.545423122760493e-06, | |
| "loss": 0.0847, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.013912240184757505, | |
| "grad_norm": 15.437261581420898, | |
| "learning_rate": 3.5188009893686916e-06, | |
| "loss": 0.2216, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.013930715935334874, | |
| "grad_norm": 7.0192131996154785, | |
| "learning_rate": 3.492257827341492e-06, | |
| "loss": 0.2409, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.01394919168591224, | |
| "grad_norm": 2.5094802379608154, | |
| "learning_rate": 3.4657939600989453e-06, | |
| "loss": 0.0991, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.013967667436489608, | |
| "grad_norm": 1.0215903520584106, | |
| "learning_rate": 3.4394097100949286e-06, | |
| "loss": 0.1046, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.013986143187066974, | |
| "grad_norm": 4.129768371582031, | |
| "learning_rate": 3.4131053988131947e-06, | |
| "loss": 0.1561, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.014004618937644342, | |
| "grad_norm": 0.7603411078453064, | |
| "learning_rate": 3.3868813467634833e-06, | |
| "loss": 0.1091, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.014023094688221709, | |
| "grad_norm": 3.7008919715881348, | |
| "learning_rate": 3.360737873477584e-06, | |
| "loss": 0.1846, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.014041570438799077, | |
| "grad_norm": 2.1636464595794678, | |
| "learning_rate": 3.3346752975054763e-06, | |
| "loss": 0.1248, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.014060046189376443, | |
| "grad_norm": 0.5979150533676147, | |
| "learning_rate": 3.308693936411421e-06, | |
| "loss": 0.086, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.014078521939953811, | |
| "grad_norm": 4.374721527099609, | |
| "learning_rate": 3.2827941067700996e-06, | |
| "loss": 0.1363, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.014096997690531177, | |
| "grad_norm": 2.692699432373047, | |
| "learning_rate": 3.2569761241627694e-06, | |
| "loss": 0.0987, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.014115473441108545, | |
| "grad_norm": 0.885154128074646, | |
| "learning_rate": 3.2312403031733943e-06, | |
| "loss": 0.0879, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.014133949191685912, | |
| "grad_norm": 1.592898964881897, | |
| "learning_rate": 3.2055869573848374e-06, | |
| "loss": 0.1046, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.01415242494226328, | |
| "grad_norm": 5.411794185638428, | |
| "learning_rate": 3.1800163993750166e-06, | |
| "loss": 0.1176, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.014170900692840646, | |
| "grad_norm": 5.728789806365967, | |
| "learning_rate": 3.1545289407131128e-06, | |
| "loss": 0.155, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.014189376443418014, | |
| "grad_norm": 0.4029484689235687, | |
| "learning_rate": 3.1291248919557717e-06, | |
| "loss": 0.074, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.01420785219399538, | |
| "grad_norm": 0.6320480108261108, | |
| "learning_rate": 3.103804562643302e-06, | |
| "loss": 0.0941, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.014226327944572749, | |
| "grad_norm": 1.162419319152832, | |
| "learning_rate": 3.0785682612959334e-06, | |
| "loss": 0.0967, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.014244803695150115, | |
| "grad_norm": 0.48519471287727356, | |
| "learning_rate": 3.0534162954100264e-06, | |
| "loss": 0.0796, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.014263279445727483, | |
| "grad_norm": 0.7229211926460266, | |
| "learning_rate": 3.028348971454356e-06, | |
| "loss": 0.0752, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.01428175519630485, | |
| "grad_norm": 1.0490971803665161, | |
| "learning_rate": 3.003366594866345e-06, | |
| "loss": 0.0891, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.014300230946882217, | |
| "grad_norm": 6.083411693572998, | |
| "learning_rate": 2.978469470048376e-06, | |
| "loss": 0.1687, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.014318706697459584, | |
| "grad_norm": 0.8749793767929077, | |
| "learning_rate": 2.953657900364053e-06, | |
| "loss": 0.0934, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.014337182448036952, | |
| "grad_norm": 0.7029964327812195, | |
| "learning_rate": 2.9289321881345257e-06, | |
| "loss": 0.0764, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.014355658198614318, | |
| "grad_norm": 0.34574633836746216, | |
| "learning_rate": 2.9042926346347932e-06, | |
| "loss": 0.0742, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.014374133949191686, | |
| "grad_norm": 18.775291442871094, | |
| "learning_rate": 2.8797395400900362e-06, | |
| "loss": 0.3891, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.014392609699769052, | |
| "grad_norm": 0.43610191345214844, | |
| "learning_rate": 2.855273203671969e-06, | |
| "loss": 0.0823, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.01441108545034642, | |
| "grad_norm": 0.4868527948856354, | |
| "learning_rate": 2.830893923495173e-06, | |
| "loss": 0.0755, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.014429561200923787, | |
| "grad_norm": 11.030800819396973, | |
| "learning_rate": 2.8066019966134907e-06, | |
| "loss": 0.4278, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.014448036951501155, | |
| "grad_norm": 0.4643419682979584, | |
| "learning_rate": 2.7823977190163788e-06, | |
| "loss": 0.0768, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.014466512702078521, | |
| "grad_norm": 0.9213546514511108, | |
| "learning_rate": 2.7582813856253276e-06, | |
| "loss": 0.0807, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.01448498845265589, | |
| "grad_norm": 11.641874313354492, | |
| "learning_rate": 2.7342532902902418e-06, | |
| "loss": 0.6559, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.014503464203233256, | |
| "grad_norm": 0.32688185572624207, | |
| "learning_rate": 2.7103137257858867e-06, | |
| "loss": 0.0784, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.014521939953810624, | |
| "grad_norm": 12.124221801757812, | |
| "learning_rate": 2.6864629838082957e-06, | |
| "loss": 0.2114, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.014540415704387992, | |
| "grad_norm": 3.694218873977661, | |
| "learning_rate": 2.6627013549712355e-06, | |
| "loss": 0.1185, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.014558891454965358, | |
| "grad_norm": 1.008851408958435, | |
| "learning_rate": 2.639029128802657e-06, | |
| "loss": 0.1627, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.014577367205542726, | |
| "grad_norm": 10.219413757324219, | |
| "learning_rate": 2.615446593741161e-06, | |
| "loss": 0.2502, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.014595842956120092, | |
| "grad_norm": 0.5060427188873291, | |
| "learning_rate": 2.5919540371325005e-06, | |
| "loss": 0.0888, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.01461431870669746, | |
| "grad_norm": 4.4916229248046875, | |
| "learning_rate": 2.5685517452260566e-06, | |
| "loss": 0.1232, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.014632794457274827, | |
| "grad_norm": 0.5018702745437622, | |
| "learning_rate": 2.5452400031713786e-06, | |
| "loss": 0.0822, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.014651270207852195, | |
| "grad_norm": 6.717902183532715, | |
| "learning_rate": 2.522019095014683e-06, | |
| "loss": 0.1391, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.014669745958429561, | |
| "grad_norm": 0.6607494950294495, | |
| "learning_rate": 2.4988893036954045e-06, | |
| "loss": 0.0826, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.014688221709006929, | |
| "grad_norm": 1.6110329627990723, | |
| "learning_rate": 2.4758509110427576e-06, | |
| "loss": 0.1042, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.014706697459584295, | |
| "grad_norm": 4.945409774780273, | |
| "learning_rate": 2.45290419777228e-06, | |
| "loss": 0.1202, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.014725173210161664, | |
| "grad_norm": 1.9938654899597168, | |
| "learning_rate": 2.4300494434824373e-06, | |
| "loss": 0.0914, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.01474364896073903, | |
| "grad_norm": 1.2510606050491333, | |
| "learning_rate": 2.407286926651192e-06, | |
| "loss": 0.08, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.014762124711316398, | |
| "grad_norm": 0.9995988011360168, | |
| "learning_rate": 2.3846169246326345e-06, | |
| "loss": 0.0996, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.014780600461893764, | |
| "grad_norm": 0.9600606560707092, | |
| "learning_rate": 2.362039713653581e-06, | |
| "loss": 0.0884, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.014799076212471132, | |
| "grad_norm": 0.7279072403907776, | |
| "learning_rate": 2.339555568810221e-06, | |
| "loss": 0.086, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.014817551963048499, | |
| "grad_norm": 0.6564614772796631, | |
| "learning_rate": 2.317164764064769e-06, | |
| "loss": 0.0857, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.014836027713625867, | |
| "grad_norm": 0.45701906085014343, | |
| "learning_rate": 2.2948675722421086e-06, | |
| "loss": 0.0743, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.014854503464203233, | |
| "grad_norm": 13.51742935180664, | |
| "learning_rate": 2.27266426502649e-06, | |
| "loss": 0.2679, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.014872979214780601, | |
| "grad_norm": 6.072559833526611, | |
| "learning_rate": 2.2505551129582047e-06, | |
| "loss": 0.133, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.014891454965357967, | |
| "grad_norm": 0.7837164998054504, | |
| "learning_rate": 2.2285403854302912e-06, | |
| "loss": 0.0901, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.014909930715935335, | |
| "grad_norm": 4.972738742828369, | |
| "learning_rate": 2.206620350685257e-06, | |
| "loss": 0.135, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.014928406466512702, | |
| "grad_norm": 8.200922966003418, | |
| "learning_rate": 2.1847952758118118e-06, | |
| "loss": 0.264, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.01494688221709007, | |
| "grad_norm": 3.3197109699249268, | |
| "learning_rate": 2.163065426741603e-06, | |
| "loss": 0.1297, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.014965357967667436, | |
| "grad_norm": 15.642394065856934, | |
| "learning_rate": 2.1414310682459805e-06, | |
| "loss": 0.7439, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.014983833718244804, | |
| "grad_norm": 2.877195119857788, | |
| "learning_rate": 2.119892463932781e-06, | |
| "loss": 0.1172, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.01500230946882217, | |
| "grad_norm": 1.013956904411316, | |
| "learning_rate": 2.098449876243096e-06, | |
| "loss": 0.0908, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.015020785219399539, | |
| "grad_norm": 0.6518961787223816, | |
| "learning_rate": 2.0771035664480944e-06, | |
| "loss": 0.0871, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.015039260969976905, | |
| "grad_norm": 0.6690924763679504, | |
| "learning_rate": 2.0558537946458177e-06, | |
| "loss": 0.0923, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.015057736720554273, | |
| "grad_norm": 10.403573989868164, | |
| "learning_rate": 2.0347008197580376e-06, | |
| "loss": 0.2172, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.01507621247113164, | |
| "grad_norm": 0.704076886177063, | |
| "learning_rate": 2.013644899527074e-06, | |
| "loss": 0.0877, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.015094688221709007, | |
| "grad_norm": 3.530696153640747, | |
| "learning_rate": 1.9926862905126663e-06, | |
| "loss": 0.1202, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.015113163972286374, | |
| "grad_norm": 1.5041415691375732, | |
| "learning_rate": 1.9718252480888567e-06, | |
| "loss": 0.085, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.015131639722863742, | |
| "grad_norm": 2.162738800048828, | |
| "learning_rate": 1.95106202644086e-06, | |
| "loss": 0.0982, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.015150115473441108, | |
| "grad_norm": 9.021785736083984, | |
| "learning_rate": 1.930396878561983e-06, | |
| "loss": 0.1669, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.015168591224018476, | |
| "grad_norm": 1.1271936893463135, | |
| "learning_rate": 1.9098300562505266e-06, | |
| "loss": 0.0884, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.015187066974595842, | |
| "grad_norm": 1.4151726961135864, | |
| "learning_rate": 1.8893618101067357e-06, | |
| "loss": 0.0997, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.01520554272517321, | |
| "grad_norm": 4.075433731079102, | |
| "learning_rate": 1.8689923895297247e-06, | |
| "loss": 0.1662, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.015224018475750577, | |
| "grad_norm": 0.8632172346115112, | |
| "learning_rate": 1.848722042714457e-06, | |
| "loss": 0.0812, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.015242494226327945, | |
| "grad_norm": 13.40947151184082, | |
| "learning_rate": 1.8285510166487154e-06, | |
| "loss": 0.3629, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.015260969976905311, | |
| "grad_norm": 0.8392366766929626, | |
| "learning_rate": 1.808479557110081e-06, | |
| "loss": 0.0729, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.01527944572748268, | |
| "grad_norm": 2.4267046451568604, | |
| "learning_rate": 1.7885079086629598e-06, | |
| "loss": 0.1325, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.015297921478060045, | |
| "grad_norm": 0.9064236879348755, | |
| "learning_rate": 1.7686363146555807e-06, | |
| "loss": 0.0917, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.015316397228637414, | |
| "grad_norm": 2.2571587562561035, | |
| "learning_rate": 1.7488650172170496e-06, | |
| "loss": 0.121, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.01533487297921478, | |
| "grad_norm": 1.8526854515075684, | |
| "learning_rate": 1.7291942572543806e-06, | |
| "loss": 0.0982, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.015353348729792148, | |
| "grad_norm": 0.41169992089271545, | |
| "learning_rate": 1.709624274449584e-06, | |
| "loss": 0.0699, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.015371824480369514, | |
| "grad_norm": 0.5750823616981506, | |
| "learning_rate": 1.6901553072567189e-06, | |
| "loss": 0.0833, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.015390300230946882, | |
| "grad_norm": 8.618647575378418, | |
| "learning_rate": 1.6707875928990059e-06, | |
| "loss": 0.1925, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.015408775981524249, | |
| "grad_norm": 0.5770317912101746, | |
| "learning_rate": 1.651521367365936e-06, | |
| "loss": 0.094, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.015427251732101617, | |
| "grad_norm": 1.3221720457077026, | |
| "learning_rate": 1.6323568654103838e-06, | |
| "loss": 0.0939, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.015445727482678985, | |
| "grad_norm": 1.7528797388076782, | |
| "learning_rate": 1.6132943205457607e-06, | |
| "loss": 0.0982, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.015464203233256351, | |
| "grad_norm": 0.3195449113845825, | |
| "learning_rate": 1.5943339650431578e-06, | |
| "loss": 0.0618, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.015482678983833719, | |
| "grad_norm": 0.9524803161621094, | |
| "learning_rate": 1.5754760299285255e-06, | |
| "loss": 0.0795, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.015501154734411085, | |
| "grad_norm": 1.1368300914764404, | |
| "learning_rate": 1.5567207449798517e-06, | |
| "loss": 0.0957, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.015519630484988453, | |
| "grad_norm": 0.482770174741745, | |
| "learning_rate": 1.538068338724361e-06, | |
| "loss": 0.0674, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.01553810623556582, | |
| "grad_norm": 0.5334994792938232, | |
| "learning_rate": 1.5195190384357405e-06, | |
| "loss": 0.0776, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.015556581986143188, | |
| "grad_norm": 0.5965861082077026, | |
| "learning_rate": 1.5010730701313626e-06, | |
| "loss": 0.0929, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.015575057736720554, | |
| "grad_norm": 10.337512969970703, | |
| "learning_rate": 1.4827306585695234e-06, | |
| "loss": 0.462, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.015593533487297922, | |
| "grad_norm": 0.46454837918281555, | |
| "learning_rate": 1.4644920272467245e-06, | |
| "loss": 0.0724, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.015612009237875289, | |
| "grad_norm": 0.4935779869556427, | |
| "learning_rate": 1.446357398394934e-06, | |
| "loss": 0.0708, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.015630484988452657, | |
| "grad_norm": 3.0126607418060303, | |
| "learning_rate": 1.4283269929788779e-06, | |
| "loss": 0.1239, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.015648960739030025, | |
| "grad_norm": 1.0828951597213745, | |
| "learning_rate": 1.4104010306933558e-06, | |
| "loss": 0.0813, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.01566743648960739, | |
| "grad_norm": 0.6123186945915222, | |
| "learning_rate": 1.3925797299605649e-06, | |
| "loss": 0.0744, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.015685912240184757, | |
| "grad_norm": 0.3049459457397461, | |
| "learning_rate": 1.3748633079274254e-06, | |
| "loss": 0.061, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.015704387990762125, | |
| "grad_norm": 2.8126583099365234, | |
| "learning_rate": 1.3572519804629537e-06, | |
| "loss": 0.1198, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.015722863741339493, | |
| "grad_norm": 3.7487950325012207, | |
| "learning_rate": 1.339745962155613e-06, | |
| "loss": 0.1572, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.015741339491916858, | |
| "grad_norm": 1.1969902515411377, | |
| "learning_rate": 1.322345466310717e-06, | |
| "loss": 0.0872, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.015759815242494226, | |
| "grad_norm": 0.5315749645233154, | |
| "learning_rate": 1.30505070494781e-06, | |
| "loss": 0.0801, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.015778290993071594, | |
| "grad_norm": 4.150108814239502, | |
| "learning_rate": 1.2878618887981064e-06, | |
| "loss": 0.1082, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.015796766743648962, | |
| "grad_norm": 18.863311767578125, | |
| "learning_rate": 1.2707792273019049e-06, | |
| "loss": 0.2975, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.015815242494226327, | |
| "grad_norm": 1.677560806274414, | |
| "learning_rate": 1.2538029286060428e-06, | |
| "loss": 0.0893, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.015833718244803695, | |
| "grad_norm": 0.7258471250534058, | |
| "learning_rate": 1.2369331995613664e-06, | |
| "loss": 0.0946, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.015852193995381063, | |
| "grad_norm": 3.0752053260803223, | |
| "learning_rate": 1.2201702457201948e-06, | |
| "loss": 0.1533, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.01587066974595843, | |
| "grad_norm": 0.5181306004524231, | |
| "learning_rate": 1.2035142713338366e-06, | |
| "loss": 0.0825, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.015889145496535795, | |
| "grad_norm": 0.40795764327049255, | |
| "learning_rate": 1.1869654793500784e-06, | |
| "loss": 0.0741, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.015907621247113164, | |
| "grad_norm": 0.6895514726638794, | |
| "learning_rate": 1.1705240714107301e-06, | |
| "loss": 0.0895, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.01592609699769053, | |
| "grad_norm": 0.5122327208518982, | |
| "learning_rate": 1.1541902478491607e-06, | |
| "loss": 0.0818, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.0159445727482679, | |
| "grad_norm": 0.4552116096019745, | |
| "learning_rate": 1.1379642076878528e-06, | |
| "loss": 0.0808, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.015963048498845264, | |
| "grad_norm": 1.596439003944397, | |
| "learning_rate": 1.1218461486359878e-06, | |
| "loss": 0.1172, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.015981524249422632, | |
| "grad_norm": 0.5557100772857666, | |
| "learning_rate": 1.1058362670870248e-06, | |
| "loss": 0.0813, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.016, | |
| "grad_norm": 4.133650779724121, | |
| "learning_rate": 1.0899347581163222e-06, | |
| "loss": 0.1238, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.01601847575057737, | |
| "grad_norm": 1.7508395910263062, | |
| "learning_rate": 1.0741418154787443e-06, | |
| "loss": 0.1152, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.016036951501154733, | |
| "grad_norm": 11.699679374694824, | |
| "learning_rate": 1.058457631606319e-06, | |
| "loss": 0.4958, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.0160554272517321, | |
| "grad_norm": 4.510467529296875, | |
| "learning_rate": 1.042882397605871e-06, | |
| "loss": 0.1335, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.01607390300230947, | |
| "grad_norm": 0.6048879027366638, | |
| "learning_rate": 1.0274163032567165e-06, | |
| "loss": 0.0696, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.016092378752886837, | |
| "grad_norm": 11.806937217712402, | |
| "learning_rate": 1.012059537008332e-06, | |
| "loss": 0.3041, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.0161108545034642, | |
| "grad_norm": 0.4878392219543457, | |
| "learning_rate": 9.968122859780648e-07, | |
| "loss": 0.0773, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.01612933025404157, | |
| "grad_norm": 8.298155784606934, | |
| "learning_rate": 9.816747359488632e-07, | |
| "loss": 0.3257, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.016147806004618938, | |
| "grad_norm": 27.34232521057129, | |
| "learning_rate": 9.666470713669918e-07, | |
| "loss": 0.4234, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.016166281755196306, | |
| "grad_norm": 11.97795581817627, | |
| "learning_rate": 9.517294753398066e-07, | |
| "loss": 0.3336, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.01618475750577367, | |
| "grad_norm": 4.471753120422363, | |
| "learning_rate": 9.369221296335007e-07, | |
| "loss": 0.1599, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.01620323325635104, | |
| "grad_norm": 5.755516529083252, | |
| "learning_rate": 9.222252146709143e-07, | |
| "loss": 0.1858, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.016221709006928407, | |
| "grad_norm": 0.6894332766532898, | |
| "learning_rate": 9.076389095293148e-07, | |
| "loss": 0.0879, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.016240184757505775, | |
| "grad_norm": 5.352380752563477, | |
| "learning_rate": 8.931633919382299e-07, | |
| "loss": 0.14, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.01625866050808314, | |
| "grad_norm": 8.2031888961792, | |
| "learning_rate": 8.787988382772705e-07, | |
| "loss": 0.1732, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.016277136258660507, | |
| "grad_norm": 0.8718963265419006, | |
| "learning_rate": 8.645454235739903e-07, | |
| "loss": 0.0884, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.016295612009237875, | |
| "grad_norm": 2.092503786087036, | |
| "learning_rate": 8.504033215017527e-07, | |
| "loss": 0.1106, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.016314087759815243, | |
| "grad_norm": 7.39290189743042, | |
| "learning_rate": 8.363727043776037e-07, | |
| "loss": 0.1781, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.016332563510392608, | |
| "grad_norm": 0.3521372675895691, | |
| "learning_rate": 8.224537431601886e-07, | |
| "loss": 0.0794, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.016351039260969976, | |
| "grad_norm": 2.420149326324463, | |
| "learning_rate": 8.086466074476562e-07, | |
| "loss": 0.107, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.016369515011547344, | |
| "grad_norm": 0.7112013697624207, | |
| "learning_rate": 7.949514654755963e-07, | |
| "loss": 0.074, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.016387990762124712, | |
| "grad_norm": 17.587255477905273, | |
| "learning_rate": 7.81368484114996e-07, | |
| "loss": 0.4649, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.01640646651270208, | |
| "grad_norm": 0.8972907662391663, | |
| "learning_rate": 7.678978288701911e-07, | |
| "loss": 0.0787, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.016424942263279445, | |
| "grad_norm": 14.27460765838623, | |
| "learning_rate": 7.545396638768698e-07, | |
| "loss": 0.4648, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.016443418013856813, | |
| "grad_norm": 3.213332414627075, | |
| "learning_rate": 7.412941519000527e-07, | |
| "loss": 0.1255, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.01646189376443418, | |
| "grad_norm": 0.40093788504600525, | |
| "learning_rate": 7.281614543321269e-07, | |
| "loss": 0.0867, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.01648036951501155, | |
| "grad_norm": 0.31747710704803467, | |
| "learning_rate": 7.151417311908648e-07, | |
| "loss": 0.0688, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.016498845265588914, | |
| "grad_norm": 5.613777160644531, | |
| "learning_rate": 7.022351411174866e-07, | |
| "loss": 0.1747, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.01651732101616628, | |
| "grad_norm": 11.971619606018066, | |
| "learning_rate": 6.894418413747183e-07, | |
| "loss": 0.4888, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.01653579676674365, | |
| "grad_norm": 0.5404453873634338, | |
| "learning_rate": 6.767619878448783e-07, | |
| "loss": 0.0742, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.016554272517321018, | |
| "grad_norm": 2.8522422313690186, | |
| "learning_rate": 6.641957350279838e-07, | |
| "loss": 0.1075, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.016572748267898382, | |
| "grad_norm": 1.9566073417663574, | |
| "learning_rate": 6.517432360398556e-07, | |
| "loss": 0.1125, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.01659122401847575, | |
| "grad_norm": 0.3344014585018158, | |
| "learning_rate": 6.394046426102673e-07, | |
| "loss": 0.0783, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.01660969976905312, | |
| "grad_norm": 4.855746269226074, | |
| "learning_rate": 6.271801050810856e-07, | |
| "loss": 0.1677, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.016628175519630486, | |
| "grad_norm": 0.48509731888771057, | |
| "learning_rate": 6.150697724044407e-07, | |
| "loss": 0.0868, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.01664665127020785, | |
| "grad_norm": 17.23429298400879, | |
| "learning_rate": 6.030737921409169e-07, | |
| "loss": 0.2653, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.01666512702078522, | |
| "grad_norm": 0.5404847860336304, | |
| "learning_rate": 5.911923104577455e-07, | |
| "loss": 0.0709, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.016683602771362587, | |
| "grad_norm": 22.001861572265625, | |
| "learning_rate": 5.794254721270331e-07, | |
| "loss": 0.2953, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.016702078521939955, | |
| "grad_norm": 2.4089291095733643, | |
| "learning_rate": 5.677734205239904e-07, | |
| "loss": 0.1104, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.01672055427251732, | |
| "grad_norm": 8.30461311340332, | |
| "learning_rate": 5.562362976251901e-07, | |
| "loss": 0.1453, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.016739030023094688, | |
| "grad_norm": 15.267427444458008, | |
| "learning_rate": 5.448142440068316e-07, | |
| "loss": 0.437, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.016757505773672056, | |
| "grad_norm": 1.4987224340438843, | |
| "learning_rate": 5.335073988430373e-07, | |
| "loss": 0.0908, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.016775981524249424, | |
| "grad_norm": 0.3668877184391022, | |
| "learning_rate": 5.223158999041444e-07, | |
| "loss": 0.0715, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.01679445727482679, | |
| "grad_norm": 0.8721979260444641, | |
| "learning_rate": 5.112398835550348e-07, | |
| "loss": 0.0841, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.016812933025404157, | |
| "grad_norm": 0.8178991079330444, | |
| "learning_rate": 5.002794847534765e-07, | |
| "loss": 0.0888, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.016831408775981525, | |
| "grad_norm": 9.230072975158691, | |
| "learning_rate": 4.894348370484648e-07, | |
| "loss": 0.1871, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.016849884526558893, | |
| "grad_norm": 4.41793966293335, | |
| "learning_rate": 4.787060725786141e-07, | |
| "loss": 0.1735, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.016868360277136257, | |
| "grad_norm": 2.220468759536743, | |
| "learning_rate": 4.6809332207053083e-07, | |
| "loss": 0.1118, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.016886836027713625, | |
| "grad_norm": 0.8025916218757629, | |
| "learning_rate": 4.575967148372318e-07, | |
| "loss": 0.0945, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.016905311778290993, | |
| "grad_norm": 0.5296105742454529, | |
| "learning_rate": 4.4721637877656377e-07, | |
| "loss": 0.0888, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.01692378752886836, | |
| "grad_norm": 5.663786888122559, | |
| "learning_rate": 4.3695244036964567e-07, | |
| "loss": 0.1268, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.016942263279445726, | |
| "grad_norm": 0.3008377254009247, | |
| "learning_rate": 4.268050246793276e-07, | |
| "loss": 0.0692, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.016960739030023094, | |
| "grad_norm": 0.4833626449108124, | |
| "learning_rate": 4.167742553486676e-07, | |
| "loss": 0.0829, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.016979214780600462, | |
| "grad_norm": 1.6660363674163818, | |
| "learning_rate": 4.068602545994249e-07, | |
| "loss": 0.0824, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.01699769053117783, | |
| "grad_norm": 0.6289246678352356, | |
| "learning_rate": 3.9706314323056936e-07, | |
| "loss": 0.0796, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.017016166281755195, | |
| "grad_norm": 0.7974089980125427, | |
| "learning_rate": 3.8738304061681107e-07, | |
| "loss": 0.0981, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.017034642032332563, | |
| "grad_norm": 0.5418569445610046, | |
| "learning_rate": 3.7782006470714614e-07, | |
| "loss": 0.0807, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.01705311778290993, | |
| "grad_norm": 14.680021286010742, | |
| "learning_rate": 3.68374332023419e-07, | |
| "loss": 0.276, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.0170715935334873, | |
| "grad_norm": 1.6710370779037476, | |
| "learning_rate": 3.590459576589e-07, | |
| "loss": 0.0922, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.017090069284064664, | |
| "grad_norm": 0.3066285252571106, | |
| "learning_rate": 3.498350552768859e-07, | |
| "loss": 0.0686, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.01710854503464203, | |
| "grad_norm": 1.7458354234695435, | |
| "learning_rate": 3.4074173710931804e-07, | |
| "loss": 0.1225, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.0171270207852194, | |
| "grad_norm": 1.7867599725723267, | |
| "learning_rate": 3.3176611395540625e-07, | |
| "loss": 0.101, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.017145496535796768, | |
| "grad_norm": 0.5294427275657654, | |
| "learning_rate": 3.2290829518028867e-07, | |
| "loss": 0.0894, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.017163972286374132, | |
| "grad_norm": 0.9586027264595032, | |
| "learning_rate": 3.1416838871368925e-07, | |
| "loss": 0.0841, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.0171824480369515, | |
| "grad_norm": 10.964192390441895, | |
| "learning_rate": 3.0554650104861137e-07, | |
| "loss": 0.4954, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.01720092378752887, | |
| "grad_norm": 1.5284767150878906, | |
| "learning_rate": 2.970427372400353e-07, | |
| "loss": 0.0992, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.017219399538106236, | |
| "grad_norm": 1.291322112083435, | |
| "learning_rate": 2.8865720090364037e-07, | |
| "loss": 0.0899, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.0172378752886836, | |
| "grad_norm": 11.606953620910645, | |
| "learning_rate": 2.8038999421453827e-07, | |
| "loss": 0.3002, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.01725635103926097, | |
| "grad_norm": 3.919572353363037, | |
| "learning_rate": 2.7224121790603517e-07, | |
| "loss": 0.1225, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.017274826789838337, | |
| "grad_norm": 2.1454436779022217, | |
| "learning_rate": 2.6421097126839714e-07, | |
| "loss": 0.1059, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.017293302540415705, | |
| "grad_norm": 1.2310255765914917, | |
| "learning_rate": 2.5629935214764866e-07, | |
| "loss": 0.0928, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.017311778290993073, | |
| "grad_norm": 15.258710861206055, | |
| "learning_rate": 2.4850645694436736e-07, | |
| "loss": 0.5669, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.017330254041570438, | |
| "grad_norm": 1.8618184328079224, | |
| "learning_rate": 2.4083238061252565e-07, | |
| "loss": 0.0996, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.017348729792147806, | |
| "grad_norm": 0.8714627027511597, | |
| "learning_rate": 2.332772166583208e-07, | |
| "loss": 0.0896, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.017367205542725174, | |
| "grad_norm": 1.522209882736206, | |
| "learning_rate": 2.2584105713904126e-07, | |
| "loss": 0.0857, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.017385681293302542, | |
| "grad_norm": 5.695614337921143, | |
| "learning_rate": 2.1852399266194312e-07, | |
| "loss": 0.1367, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.017404157043879907, | |
| "grad_norm": 0.8899729251861572, | |
| "learning_rate": 2.1132611238315004e-07, | |
| "loss": 0.0882, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.017422632794457275, | |
| "grad_norm": 9.237454414367676, | |
| "learning_rate": 2.0424750400655947e-07, | |
| "loss": 0.3788, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.017441108545034643, | |
| "grad_norm": 0.4933810532093048, | |
| "learning_rate": 1.9728825378278248e-07, | |
| "loss": 0.0731, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.01745958429561201, | |
| "grad_norm": 0.5540867447853088, | |
| "learning_rate": 1.9044844650808468e-07, | |
| "loss": 0.0922, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.017478060046189375, | |
| "grad_norm": 0.467784583568573, | |
| "learning_rate": 1.8372816552336025e-07, | |
| "loss": 0.093, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.017496535796766743, | |
| "grad_norm": 3.6386563777923584, | |
| "learning_rate": 1.7712749271311392e-07, | |
| "loss": 0.1149, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.01751501154734411, | |
| "grad_norm": 5.295081615447998, | |
| "learning_rate": 1.706465085044584e-07, | |
| "loss": 0.1344, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.01753348729792148, | |
| "grad_norm": 0.4963618516921997, | |
| "learning_rate": 1.6428529186614195e-07, | |
| "loss": 0.077, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.017551963048498844, | |
| "grad_norm": 1.4411686658859253, | |
| "learning_rate": 1.580439203075812e-07, | |
| "loss": 0.1181, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.017570438799076212, | |
| "grad_norm": 0.9312250018119812, | |
| "learning_rate": 1.519224698779198e-07, | |
| "loss": 0.0944, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.01758891454965358, | |
| "grad_norm": 0.3995867669582367, | |
| "learning_rate": 1.4592101516509916e-07, | |
| "loss": 0.0862, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.01760739030023095, | |
| "grad_norm": 0.4238080084323883, | |
| "learning_rate": 1.400396292949513e-07, | |
| "loss": 0.0803, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.017625866050808313, | |
| "grad_norm": 11.962114334106445, | |
| "learning_rate": 1.3427838393030634e-07, | |
| "loss": 0.4659, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.01764434180138568, | |
| "grad_norm": 0.5033251643180847, | |
| "learning_rate": 1.2863734927012094e-07, | |
| "loss": 0.0753, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.01766281755196305, | |
| "grad_norm": 1.7680706977844238, | |
| "learning_rate": 1.231165940486234e-07, | |
| "loss": 0.1006, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.017681293302540417, | |
| "grad_norm": 5.311273574829102, | |
| "learning_rate": 1.1771618553447217e-07, | |
| "loss": 0.1513, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.01769976905311778, | |
| "grad_norm": 0.6406497955322266, | |
| "learning_rate": 1.1243618952994195e-07, | |
| "loss": 0.0929, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.01771824480369515, | |
| "grad_norm": 1.3393217325210571, | |
| "learning_rate": 1.0727667037011668e-07, | |
| "loss": 0.1038, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.017736720554272518, | |
| "grad_norm": 11.503267288208008, | |
| "learning_rate": 1.0223769092211012e-07, | |
| "loss": 0.6077, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.017755196304849886, | |
| "grad_norm": 1.1504697799682617, | |
| "learning_rate": 9.731931258429638e-08, | |
| "loss": 0.0992, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.01777367205542725, | |
| "grad_norm": 0.5612615942955017, | |
| "learning_rate": 9.252159528556404e-08, | |
| "loss": 0.0789, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.01779214780600462, | |
| "grad_norm": 0.6853100061416626, | |
| "learning_rate": 8.784459748458318e-08, | |
| "loss": 0.092, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.017810623556581986, | |
| "grad_norm": 0.8947520852088928, | |
| "learning_rate": 8.328837616909612e-08, | |
| "loss": 0.0921, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.017829099307159355, | |
| "grad_norm": 1.0538597106933594, | |
| "learning_rate": 7.885298685522235e-08, | |
| "loss": 0.1221, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.01784757505773672, | |
| "grad_norm": 12.896748542785645, | |
| "learning_rate": 7.453848358678018e-08, | |
| "loss": 0.2736, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.017866050808314087, | |
| "grad_norm": 0.4337773025035858, | |
| "learning_rate": 7.034491893463059e-08, | |
| "loss": 0.0789, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.017884526558891455, | |
| "grad_norm": 10.789532661437988, | |
| "learning_rate": 6.627234399603554e-08, | |
| "loss": 0.2208, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.017903002309468823, | |
| "grad_norm": 7.355002403259277, | |
| "learning_rate": 6.232080839403631e-08, | |
| "loss": 0.2166, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.017921478060046188, | |
| "grad_norm": 1.3429850339889526, | |
| "learning_rate": 5.849036027684607e-08, | |
| "loss": 0.0886, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.017939953810623556, | |
| "grad_norm": 0.6590784788131714, | |
| "learning_rate": 5.4781046317267103e-08, | |
| "loss": 0.0783, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.017958429561200924, | |
| "grad_norm": 1.5127907991409302, | |
| "learning_rate": 5.119291171211793e-08, | |
| "loss": 0.0968, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.017976905311778292, | |
| "grad_norm": 0.9531641006469727, | |
| "learning_rate": 4.772600018168816e-08, | |
| "loss": 0.0896, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.017995381062355657, | |
| "grad_norm": 3.2513186931610107, | |
| "learning_rate": 4.438035396920004e-08, | |
| "loss": 0.1322, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.018013856812933025, | |
| "grad_norm": 0.3798637092113495, | |
| "learning_rate": 4.115601384029666e-08, | |
| "loss": 0.0755, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.018032332563510393, | |
| "grad_norm": 0.9231575727462769, | |
| "learning_rate": 3.805301908254455e-08, | |
| "loss": 0.0904, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.01805080831408776, | |
| "grad_norm": 0.39575543999671936, | |
| "learning_rate": 3.50714075049563e-08, | |
| "loss": 0.078, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.018069284064665125, | |
| "grad_norm": 1.1086126565933228, | |
| "learning_rate": 3.22112154375287e-08, | |
| "loss": 0.0967, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.018087759815242493, | |
| "grad_norm": 0.6990697979927063, | |
| "learning_rate": 2.947247773079753e-08, | |
| "loss": 0.09, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.01810623556581986, | |
| "grad_norm": 2.450481653213501, | |
| "learning_rate": 2.6855227755419046e-08, | |
| "loss": 0.1066, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.01812471131639723, | |
| "grad_norm": 1.9788362979888916, | |
| "learning_rate": 2.4359497401758026e-08, | |
| "loss": 0.1264, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.018143187066974594, | |
| "grad_norm": 2.655947208404541, | |
| "learning_rate": 2.1985317079500358e-08, | |
| "loss": 0.109, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.018161662817551962, | |
| "grad_norm": 1.3773173093795776, | |
| "learning_rate": 1.973271571728441e-08, | |
| "loss": 0.1338, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.01818013856812933, | |
| "grad_norm": 9.524226188659668, | |
| "learning_rate": 1.7601720762346895e-08, | |
| "loss": 0.2325, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.0181986143187067, | |
| "grad_norm": 11.638543128967285, | |
| "learning_rate": 1.5592358180189782e-08, | |
| "loss": 0.2914, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.018217090069284066, | |
| "grad_norm": 1.4122259616851807, | |
| "learning_rate": 1.370465245426167e-08, | |
| "loss": 0.0949, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.01823556581986143, | |
| "grad_norm": 1.3410162925720215, | |
| "learning_rate": 1.1938626585660252e-08, | |
| "loss": 0.0814, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.0182540415704388, | |
| "grad_norm": 1.1606661081314087, | |
| "learning_rate": 1.0294302092853647e-08, | |
| "loss": 0.082, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.018272517321016167, | |
| "grad_norm": 0.5039264559745789, | |
| "learning_rate": 8.771699011416169e-09, | |
| "loss": 0.0739, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.018290993071593535, | |
| "grad_norm": 4.659420013427734, | |
| "learning_rate": 7.370835893788508e-09, | |
| "loss": 0.249, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.0183094688221709, | |
| "grad_norm": 8.673989295959473, | |
| "learning_rate": 6.091729809042379e-09, | |
| "loss": 0.2017, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.018327944572748268, | |
| "grad_norm": 11.509638786315918, | |
| "learning_rate": 4.9343963426840006e-09, | |
| "loss": 0.2739, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.018346420323325636, | |
| "grad_norm": 0.5751962661743164, | |
| "learning_rate": 3.898849596456477e-09, | |
| "loss": 0.0788, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.018364896073903004, | |
| "grad_norm": 8.960288047790527, | |
| "learning_rate": 2.9851021881688314e-09, | |
| "loss": 0.2301, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.01838337182448037, | |
| "grad_norm": 2.6444497108459473, | |
| "learning_rate": 2.193165251545004e-09, | |
| "loss": 0.0944, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.018401847575057736, | |
| "grad_norm": 4.862392425537109, | |
| "learning_rate": 1.5230484360873043e-09, | |
| "loss": 0.1018, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.018420323325635105, | |
| "grad_norm": 1.3872954845428467, | |
| "learning_rate": 9.74759906957612e-10, | |
| "loss": 0.0941, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.018438799076212473, | |
| "grad_norm": 0.35692524909973145, | |
| "learning_rate": 5.483063448785686e-10, | |
| "loss": 0.0716, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.018457274826789837, | |
| "grad_norm": 3.7752761840820312, | |
| "learning_rate": 2.436929460525317e-10, | |
| "loss": 0.1053, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.018475750577367205, | |
| "grad_norm": 8.972894668579102, | |
| "learning_rate": 6.092342209607083e-11, | |
| "loss": 0.1872, | |
| "step": 1000 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 200, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": false, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |