| { |
| "best_global_step": 109, |
| "best_metric": 0.0, |
| "best_model_checkpoint": "./cysecbert-ttp-bert-base_data/checkpoint-109", |
| "epoch": 1.0, |
| "eval_steps": 500, |
| "global_step": 109, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.009174311926605505, |
| "grad_norm": 46530.15234375, |
| "learning_rate": 0.0, |
| "loss": 0.7023, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.01834862385321101, |
| "grad_norm": 49571.35546875, |
| "learning_rate": 5.000000000000001e-07, |
| "loss": 0.7006, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.027522935779816515, |
| "grad_norm": 45885.55078125, |
| "learning_rate": 1.0000000000000002e-06, |
| "loss": 0.7033, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.03669724770642202, |
| "grad_norm": 43333.73046875, |
| "learning_rate": 1.5e-06, |
| "loss": 0.7014, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.045871559633027525, |
| "grad_norm": 44340.41796875, |
| "learning_rate": 2.0000000000000003e-06, |
| "loss": 0.6992, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.05504587155963303, |
| "grad_norm": 48282.54296875, |
| "learning_rate": 2.5e-06, |
| "loss": 0.6924, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.06422018348623854, |
| "grad_norm": 48561.56640625, |
| "learning_rate": 3e-06, |
| "loss": 0.6948, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.07339449541284404, |
| "grad_norm": 43580.078125, |
| "learning_rate": 3.5000000000000004e-06, |
| "loss": 0.6916, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.08256880733944955, |
| "grad_norm": 46552.28515625, |
| "learning_rate": 4.000000000000001e-06, |
| "loss": 0.6854, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.09174311926605505, |
| "grad_norm": 49042.15234375, |
| "learning_rate": 4.5e-06, |
| "loss": 0.6822, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.10091743119266056, |
| "grad_norm": 97274.8515625, |
| "learning_rate": 5e-06, |
| "loss": 0.6788, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.11009174311926606, |
| "grad_norm": 55310.375, |
| "learning_rate": 5.500000000000001e-06, |
| "loss": 0.6781, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.11926605504587157, |
| "grad_norm": 70317.6328125, |
| "learning_rate": 6e-06, |
| "loss": 0.6754, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.12844036697247707, |
| "grad_norm": 46825.9765625, |
| "learning_rate": 6.5000000000000004e-06, |
| "loss": 0.6626, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.13761467889908258, |
| "grad_norm": 50745.15625, |
| "learning_rate": 7.000000000000001e-06, |
| "loss": 0.655, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.14678899082568808, |
| "grad_norm": 51342.453125, |
| "learning_rate": 7.5e-06, |
| "loss": 0.6491, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.1559633027522936, |
| "grad_norm": 51815.421875, |
| "learning_rate": 8.000000000000001e-06, |
| "loss": 0.6464, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.1651376146788991, |
| "grad_norm": 54610.65234375, |
| "learning_rate": 8.500000000000002e-06, |
| "loss": 0.6298, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.1743119266055046, |
| "grad_norm": 61936.31640625, |
| "learning_rate": 9e-06, |
| "loss": 0.623, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.1834862385321101, |
| "grad_norm": 56021.69921875, |
| "learning_rate": 9.5e-06, |
| "loss": 0.6156, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.1926605504587156, |
| "grad_norm": 60806.10546875, |
| "learning_rate": 1e-05, |
| "loss": 0.5981, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.2018348623853211, |
| "grad_norm": 58582.4140625, |
| "learning_rate": 1.05e-05, |
| "loss": 0.5869, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.21100917431192662, |
| "grad_norm": 59099.16015625, |
| "learning_rate": 1.1000000000000001e-05, |
| "loss": 0.5732, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.22018348623853212, |
| "grad_norm": 52971.81640625, |
| "learning_rate": 1.1500000000000002e-05, |
| "loss": 0.5587, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.22935779816513763, |
| "grad_norm": 51724.43359375, |
| "learning_rate": 1.2e-05, |
| "loss": 0.5477, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.23853211009174313, |
| "grad_norm": 58133.453125, |
| "learning_rate": 1.25e-05, |
| "loss": 0.5422, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.24770642201834864, |
| "grad_norm": 54134.3359375, |
| "learning_rate": 1.3000000000000001e-05, |
| "loss": 0.5319, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.25688073394495414, |
| "grad_norm": 78195.7890625, |
| "learning_rate": 1.3500000000000001e-05, |
| "loss": 0.5259, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.26605504587155965, |
| "grad_norm": 47713.20703125, |
| "learning_rate": 1.4000000000000001e-05, |
| "loss": 0.5105, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.27522935779816515, |
| "grad_norm": 45838.0703125, |
| "learning_rate": 1.45e-05, |
| "loss": 0.499, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.28440366972477066, |
| "grad_norm": 46735.58203125, |
| "learning_rate": 1.5e-05, |
| "loss": 0.4914, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.29357798165137616, |
| "grad_norm": 45432.8828125, |
| "learning_rate": 1.55e-05, |
| "loss": 0.4783, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.30275229357798167, |
| "grad_norm": 46758.58984375, |
| "learning_rate": 1.6000000000000003e-05, |
| "loss": 0.4758, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.3119266055045872, |
| "grad_norm": 45245.82421875, |
| "learning_rate": 1.65e-05, |
| "loss": 0.4608, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.3211009174311927, |
| "grad_norm": 43336.75390625, |
| "learning_rate": 1.7000000000000003e-05, |
| "loss": 0.4567, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.3302752293577982, |
| "grad_norm": 41445.90625, |
| "learning_rate": 1.75e-05, |
| "loss": 0.4465, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.3394495412844037, |
| "grad_norm": 42278.80078125, |
| "learning_rate": 1.8e-05, |
| "loss": 0.4395, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.3486238532110092, |
| "grad_norm": 42039.28515625, |
| "learning_rate": 1.85e-05, |
| "loss": 0.4289, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.3577981651376147, |
| "grad_norm": 41411.61328125, |
| "learning_rate": 1.9e-05, |
| "loss": 0.423, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.3669724770642202, |
| "grad_norm": 40133.84375, |
| "learning_rate": 1.9500000000000003e-05, |
| "loss": 0.4191, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.3761467889908257, |
| "grad_norm": 39758.8828125, |
| "learning_rate": 2e-05, |
| "loss": 0.4071, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.3853211009174312, |
| "grad_norm": 38760.36328125, |
| "learning_rate": 2.05e-05, |
| "loss": 0.3996, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.3944954128440367, |
| "grad_norm": 38552.80078125, |
| "learning_rate": 2.1e-05, |
| "loss": 0.3918, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.4036697247706422, |
| "grad_norm": 38561.7578125, |
| "learning_rate": 2.15e-05, |
| "loss": 0.3865, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.41284403669724773, |
| "grad_norm": 37616.85546875, |
| "learning_rate": 2.2000000000000003e-05, |
| "loss": 0.3794, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.42201834862385323, |
| "grad_norm": 37784.78515625, |
| "learning_rate": 2.25e-05, |
| "loss": 0.3745, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.43119266055045874, |
| "grad_norm": 36332.9140625, |
| "learning_rate": 2.3000000000000003e-05, |
| "loss": 0.3711, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.44036697247706424, |
| "grad_norm": 38899.73046875, |
| "learning_rate": 2.35e-05, |
| "loss": 0.3643, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.44954128440366975, |
| "grad_norm": 36174.9765625, |
| "learning_rate": 2.4e-05, |
| "loss": 0.3575, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.45871559633027525, |
| "grad_norm": 35699.83203125, |
| "learning_rate": 2.45e-05, |
| "loss": 0.347, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.46788990825688076, |
| "grad_norm": 35562.3671875, |
| "learning_rate": 2.5e-05, |
| "loss": 0.3438, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.47706422018348627, |
| "grad_norm": 35428.0625, |
| "learning_rate": 2.5500000000000003e-05, |
| "loss": 0.3325, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.48623853211009177, |
| "grad_norm": 34396.80078125, |
| "learning_rate": 2.6000000000000002e-05, |
| "loss": 0.3302, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.4954128440366973, |
| "grad_norm": 38376.41796875, |
| "learning_rate": 2.6500000000000004e-05, |
| "loss": 0.3218, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.5045871559633027, |
| "grad_norm": 33996.23046875, |
| "learning_rate": 2.7000000000000002e-05, |
| "loss": 0.3131, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.5137614678899083, |
| "grad_norm": 33070.55859375, |
| "learning_rate": 2.7500000000000004e-05, |
| "loss": 0.3129, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.5229357798165137, |
| "grad_norm": 33200.44140625, |
| "learning_rate": 2.8000000000000003e-05, |
| "loss": 0.3003, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.5321100917431193, |
| "grad_norm": 33099.1015625, |
| "learning_rate": 2.8499999999999998e-05, |
| "loss": 0.2904, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.5412844036697247, |
| "grad_norm": 31352.298828125, |
| "learning_rate": 2.9e-05, |
| "loss": 0.2911, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.5504587155963303, |
| "grad_norm": 31965.087890625, |
| "learning_rate": 2.95e-05, |
| "loss": 0.2815, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.5596330275229358, |
| "grad_norm": 30810.849609375, |
| "learning_rate": 3e-05, |
| "loss": 0.2818, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.5688073394495413, |
| "grad_norm": 30034.779296875, |
| "learning_rate": 3.05e-05, |
| "loss": 0.2731, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.5779816513761468, |
| "grad_norm": 30009.12109375, |
| "learning_rate": 3.1e-05, |
| "loss": 0.266, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.5871559633027523, |
| "grad_norm": 29305.173828125, |
| "learning_rate": 3.15e-05, |
| "loss": 0.2609, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.5963302752293578, |
| "grad_norm": 29081.853515625, |
| "learning_rate": 3.2000000000000005e-05, |
| "loss": 0.2543, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.6055045871559633, |
| "grad_norm": 28217.021484375, |
| "learning_rate": 3.2500000000000004e-05, |
| "loss": 0.252, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.6146788990825688, |
| "grad_norm": 26448.12890625, |
| "learning_rate": 3.3e-05, |
| "loss": 0.2565, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.6238532110091743, |
| "grad_norm": 27198.80859375, |
| "learning_rate": 3.35e-05, |
| "loss": 0.2342, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.6330275229357798, |
| "grad_norm": 26946.30859375, |
| "learning_rate": 3.4000000000000007e-05, |
| "loss": 0.2273, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.6422018348623854, |
| "grad_norm": 26236.7265625, |
| "learning_rate": 3.45e-05, |
| "loss": 0.2219, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.6513761467889908, |
| "grad_norm": 25535.818359375, |
| "learning_rate": 3.5e-05, |
| "loss": 0.2197, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.6605504587155964, |
| "grad_norm": 24871.8515625, |
| "learning_rate": 3.55e-05, |
| "loss": 0.2144, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.6697247706422018, |
| "grad_norm": 24484.55078125, |
| "learning_rate": 3.6e-05, |
| "loss": 0.2066, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.6788990825688074, |
| "grad_norm": 23655.677734375, |
| "learning_rate": 3.65e-05, |
| "loss": 0.206, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.6880733944954128, |
| "grad_norm": 23129.076171875, |
| "learning_rate": 3.7e-05, |
| "loss": 0.1979, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.6972477064220184, |
| "grad_norm": 21871.90625, |
| "learning_rate": 3.7500000000000003e-05, |
| "loss": 0.1973, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.7064220183486238, |
| "grad_norm": 21561.58984375, |
| "learning_rate": 3.8e-05, |
| "loss": 0.1935, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.7155963302752294, |
| "grad_norm": 22210.29296875, |
| "learning_rate": 3.85e-05, |
| "loss": 0.1848, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.7247706422018348, |
| "grad_norm": 20555.775390625, |
| "learning_rate": 3.9000000000000006e-05, |
| "loss": 0.1801, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.7339449541284404, |
| "grad_norm": 20406.75, |
| "learning_rate": 3.9500000000000005e-05, |
| "loss": 0.1746, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.7431192660550459, |
| "grad_norm": 19150.931640625, |
| "learning_rate": 4e-05, |
| "loss": 0.1688, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.7522935779816514, |
| "grad_norm": 18422.65625, |
| "learning_rate": 4.05e-05, |
| "loss": 0.1661, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.7614678899082569, |
| "grad_norm": 16746.20703125, |
| "learning_rate": 4.1e-05, |
| "loss": 0.1711, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.7706422018348624, |
| "grad_norm": 18229.41015625, |
| "learning_rate": 4.15e-05, |
| "loss": 0.1646, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.7798165137614679, |
| "grad_norm": 17156.267578125, |
| "learning_rate": 4.2e-05, |
| "loss": 0.1588, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.7889908256880734, |
| "grad_norm": 15970.611328125, |
| "learning_rate": 4.25e-05, |
| "loss": 0.1547, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.7981651376146789, |
| "grad_norm": 16997.103515625, |
| "learning_rate": 4.3e-05, |
| "loss": 0.1493, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.8073394495412844, |
| "grad_norm": 15529.58984375, |
| "learning_rate": 4.35e-05, |
| "loss": 0.1439, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.8165137614678899, |
| "grad_norm": 15099.9052734375, |
| "learning_rate": 4.4000000000000006e-05, |
| "loss": 0.1432, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.8256880733944955, |
| "grad_norm": 14261.3232421875, |
| "learning_rate": 4.4500000000000004e-05, |
| "loss": 0.1396, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.8348623853211009, |
| "grad_norm": 13915.87890625, |
| "learning_rate": 4.5e-05, |
| "loss": 0.1317, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.8440366972477065, |
| "grad_norm": 13518.7568359375, |
| "learning_rate": 4.55e-05, |
| "loss": 0.1289, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.8532110091743119, |
| "grad_norm": 12170.1865234375, |
| "learning_rate": 4.600000000000001e-05, |
| "loss": 0.1371, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.8623853211009175, |
| "grad_norm": 11745.9775390625, |
| "learning_rate": 4.6500000000000005e-05, |
| "loss": 0.1434, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.8715596330275229, |
| "grad_norm": 12265.5556640625, |
| "learning_rate": 4.7e-05, |
| "loss": 0.1343, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.8807339449541285, |
| "grad_norm": 11386.3291015625, |
| "learning_rate": 4.75e-05, |
| "loss": 0.1317, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.8899082568807339, |
| "grad_norm": 10869.4599609375, |
| "learning_rate": 4.8e-05, |
| "loss": 0.1295, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.8990825688073395, |
| "grad_norm": 10752.25, |
| "learning_rate": 4.85e-05, |
| "loss": 0.124, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.908256880733945, |
| "grad_norm": 9687.6376953125, |
| "learning_rate": 4.9e-05, |
| "loss": 0.1271, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.9174311926605505, |
| "grad_norm": 9697.15234375, |
| "learning_rate": 4.9500000000000004e-05, |
| "loss": 0.1194, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.926605504587156, |
| "grad_norm": 9423.21875, |
| "learning_rate": 5e-05, |
| "loss": 0.1176, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.9357798165137615, |
| "grad_norm": 10253.8564453125, |
| "learning_rate": 4.994949494949495e-05, |
| "loss": 0.121, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.944954128440367, |
| "grad_norm": 9756.7646484375, |
| "learning_rate": 4.98989898989899e-05, |
| "loss": 0.1173, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.9541284403669725, |
| "grad_norm": 8482.177734375, |
| "learning_rate": 4.984848484848485e-05, |
| "loss": 0.1233, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.963302752293578, |
| "grad_norm": 7755.4892578125, |
| "learning_rate": 4.97979797979798e-05, |
| "loss": 0.1251, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.9724770642201835, |
| "grad_norm": 8553.6689453125, |
| "learning_rate": 4.974747474747475e-05, |
| "loss": 0.1287, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.981651376146789, |
| "grad_norm": 8794.58984375, |
| "learning_rate": 4.9696969696969694e-05, |
| "loss": 0.1178, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.9908256880733946, |
| "grad_norm": 8048.97509765625, |
| "learning_rate": 4.964646464646465e-05, |
| "loss": 0.1112, |
| "step": 108 |
| }, |
| { |
| "epoch": 1.0, |
| "grad_norm": 24349.96484375, |
| "learning_rate": 4.9595959595959594e-05, |
| "loss": 0.112, |
| "step": 109 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_f1_macro": 0.0, |
| "eval_f1_micro": 0.0, |
| "eval_loss": 0.1133684441447258, |
| "eval_precision": 0.0, |
| "eval_recall": 0.0, |
| "eval_runtime": 0.8578, |
| "eval_samples_per_second": 336.892, |
| "eval_steps_per_second": 15.154, |
| "step": 109 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 1090, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 10, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 683067445217280.0, |
| "train_batch_size": 24, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
