Download trainer_state.json from Pathstudy/copath_router: direct link, hf CLI and curl.
- Browser
- Download file 14.9 kB
-
https://huggingface.co/Pathstudy/copath_router/resolve/main/trainer_state.json
- Command line
-
hf download hf://Pathstudy/copath_router/trainer_state.json
-
curl -L -o trainer_state.json https://huggingface.co/Pathstudy/copath_router/resolve/main/trainer_state.json
14.9 kB
| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 82, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.012232415902140673, | |
| "grad_norm": 615.5435180664062, | |
| "learning_rate": 0.0, | |
| "loss": 4.4577, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.024464831804281346, | |
| "grad_norm": 605.0523071289062, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 4.3946, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.03669724770642202, | |
| "grad_norm": 551.2265014648438, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 4.2964, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.04892966360856269, | |
| "grad_norm": 380.5257263183594, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 3.8621, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.06116207951070336, | |
| "grad_norm": 351.9261779785156, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 3.2941, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.07339449541284404, | |
| "grad_norm": 138.64157104492188, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 1.825, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0856269113149847, | |
| "grad_norm": 80.24703979492188, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.7192, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.09785932721712538, | |
| "grad_norm": 72.78546905517578, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.3413, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.11009174311926606, | |
| "grad_norm": 17.034770965576172, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.0888, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.12232415902140673, | |
| "grad_norm": 78.42652130126953, | |
| "learning_rate": 1e-05, | |
| "loss": 0.0485, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.1345565749235474, | |
| "grad_norm": 4.733565330505371, | |
| "learning_rate": 9.995370575511151e-06, | |
| "loss": 0.0272, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.14678899082568808, | |
| "grad_norm": 1.7306946516036987, | |
| "learning_rate": 9.98149087467304e-06, | |
| "loss": 0.007, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.15902140672782875, | |
| "grad_norm": 314.4232482910156, | |
| "learning_rate": 9.95838659949645e-06, | |
| "loss": 0.1905, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.1712538226299694, | |
| "grad_norm": 859.4495239257812, | |
| "learning_rate": 9.926100533780304e-06, | |
| "loss": 0.5829, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.1834862385321101, | |
| "grad_norm": 6.539334297180176, | |
| "learning_rate": 9.88469246388591e-06, | |
| "loss": 0.011, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.19571865443425077, | |
| "grad_norm": 0.41706421971321106, | |
| "learning_rate": 9.834239068026388e-06, | |
| "loss": 0.0009, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.20795107033639143, | |
| "grad_norm": 5.274829864501953, | |
| "learning_rate": 9.774833774276278e-06, | |
| "loss": 0.0216, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.22018348623853212, | |
| "grad_norm": 291.6263732910156, | |
| "learning_rate": 9.706586587564236e-06, | |
| "loss": 0.1885, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.2324159021406728, | |
| "grad_norm": 605.3851318359375, | |
| "learning_rate": 9.62962388596925e-06, | |
| "loss": 0.3885, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.24464831804281345, | |
| "grad_norm": 6.560539245605469, | |
| "learning_rate": 9.544088186697515e-06, | |
| "loss": 0.0048, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.25688073394495414, | |
| "grad_norm": 0.42615190148353577, | |
| "learning_rate": 9.450137882173385e-06, | |
| "loss": 0.0006, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.2691131498470948, | |
| "grad_norm": 0.41888362169265747, | |
| "learning_rate": 9.347946946733055e-06, | |
| "loss": 0.0002, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.28134556574923547, | |
| "grad_norm": 0.3983878195285797, | |
| "learning_rate": 9.237704614464157e-06, | |
| "loss": 0.0018, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.29357798165137616, | |
| "grad_norm": 0.020351959392428398, | |
| "learning_rate": 9.119615028787771e-06, | |
| "loss": 0.0001, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.3058103975535168, | |
| "grad_norm": 0.026613201946020126, | |
| "learning_rate": 8.993896864431825e-06, | |
| "loss": 0.0001, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.3180428134556575, | |
| "grad_norm": 1.70807683467865, | |
| "learning_rate": 8.860782922495821e-06, | |
| "loss": 0.0066, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.3302752293577982, | |
| "grad_norm": 30.285245895385742, | |
| "learning_rate": 8.720519699356804e-06, | |
| "loss": 0.0294, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.3425076452599388, | |
| "grad_norm": 0.31794869899749756, | |
| "learning_rate": 8.573366930214807e-06, | |
| "loss": 0.0004, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.3547400611620795, | |
| "grad_norm": 0.03447237238287926, | |
| "learning_rate": 8.419597108123054e-06, | |
| "loss": 0.0001, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.3669724770642202, | |
| "grad_norm": 0.9660813808441162, | |
| "learning_rate": 8.259494979393563e-06, | |
| "loss": 0.0045, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.37920489296636084, | |
| "grad_norm": 0.10378360003232956, | |
| "learning_rate": 8.093357016312518e-06, | |
| "loss": 0.0003, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.39143730886850153, | |
| "grad_norm": 0.05631450191140175, | |
| "learning_rate": 7.921490868141843e-06, | |
| "loss": 0.0001, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.4036697247706422, | |
| "grad_norm": 0.459423691034317, | |
| "learning_rate": 7.744214791423597e-06, | |
| "loss": 0.0041, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.41590214067278286, | |
| "grad_norm": 0.24352239072322845, | |
| "learning_rate": 7.56185706064212e-06, | |
| "loss": 0.0002, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.42813455657492355, | |
| "grad_norm": 0.04098845645785332, | |
| "learning_rate": 7.374755360335253e-06, | |
| "loss": 0.0001, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.44036697247706424, | |
| "grad_norm": 0.35482168197631836, | |
| "learning_rate": 7.183256159780321e-06, | |
| "loss": 0.0024, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.4525993883792049, | |
| "grad_norm": 0.006163951940834522, | |
| "learning_rate": 6.987714071412781e-06, | |
| "loss": 0.0, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.4648318042813456, | |
| "grad_norm": 0.860553503036499, | |
| "learning_rate": 6.788491194165629e-06, | |
| "loss": 0.0018, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.47706422018348627, | |
| "grad_norm": 0.004722528159618378, | |
| "learning_rate": 6.585956442945531e-06, | |
| "loss": 0.0, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.4892966360856269, | |
| "grad_norm": 0.03440281003713608, | |
| "learning_rate": 6.380484865487346e-06, | |
| "loss": 0.0, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.5015290519877675, | |
| "grad_norm": 0.01446403656154871, | |
| "learning_rate": 6.1724569478520495e-06, | |
| "loss": 0.0, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.5137614678899083, | |
| "grad_norm": 0.11344193667173386, | |
| "learning_rate": 5.96225790985415e-06, | |
| "loss": 0.0004, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.5259938837920489, | |
| "grad_norm": 0.23301349580287933, | |
| "learning_rate": 5.7502769917232635e-06, | |
| "loss": 0.0008, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.5382262996941896, | |
| "grad_norm": 0.2916724681854248, | |
| "learning_rate": 5.536906733320816e-06, | |
| "loss": 0.0005, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.5504587155963303, | |
| "grad_norm": 0.12926562130451202, | |
| "learning_rate": 5.322542247246583e-06, | |
| "loss": 0.0004, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.5626911314984709, | |
| "grad_norm": 0.043027572333812714, | |
| "learning_rate": 5.107580487181112e-06, | |
| "loss": 0.0001, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.5749235474006116, | |
| "grad_norm": 0.006686689797788858, | |
| "learning_rate": 4.89241951281889e-06, | |
| "loss": 0.0, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.5871559633027523, | |
| "grad_norm": 0.7427200675010681, | |
| "learning_rate": 4.6774577527534195e-06, | |
| "loss": 0.0023, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.599388379204893, | |
| "grad_norm": 0.9017903208732605, | |
| "learning_rate": 4.463093266679185e-06, | |
| "loss": 0.0057, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.6116207951070336, | |
| "grad_norm": 0.027725042775273323, | |
| "learning_rate": 4.249723008276737e-06, | |
| "loss": 0.0001, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.6238532110091743, | |
| "grad_norm": 0.0027795429341495037, | |
| "learning_rate": 4.037742090145851e-06, | |
| "loss": 0.0, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.636085626911315, | |
| "grad_norm": 0.002522357041016221, | |
| "learning_rate": 3.827543052147952e-06, | |
| "loss": 0.0, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.6483180428134556, | |
| "grad_norm": 0.04067041724920273, | |
| "learning_rate": 3.6195151345126556e-06, | |
| "loss": 0.0001, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.6605504587155964, | |
| "grad_norm": 0.047884393483400345, | |
| "learning_rate": 3.4140435570544708e-06, | |
| "loss": 0.0001, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.672782874617737, | |
| "grad_norm": 0.5345789194107056, | |
| "learning_rate": 3.2115088058343725e-06, | |
| "loss": 0.0026, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.6850152905198776, | |
| "grad_norm": 0.0018959270091727376, | |
| "learning_rate": 3.0122859285872214e-06, | |
| "loss": 0.0, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.6972477064220184, | |
| "grad_norm": 0.07477860152721405, | |
| "learning_rate": 2.816743840219681e-06, | |
| "loss": 0.0002, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.709480122324159, | |
| "grad_norm": 0.0016865769866853952, | |
| "learning_rate": 2.6252446396647503e-06, | |
| "loss": 0.0, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.7217125382262997, | |
| "grad_norm": 0.4168635904788971, | |
| "learning_rate": 2.438142939357882e-06, | |
| "loss": 0.0014, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.7339449541284404, | |
| "grad_norm": 0.06997565925121307, | |
| "learning_rate": 2.2557852085764053e-06, | |
| "loss": 0.0001, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.746177370030581, | |
| "grad_norm": 0.0015667227562516928, | |
| "learning_rate": 2.0785091318581577e-06, | |
| "loss": 0.0, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.7584097859327217, | |
| "grad_norm": 0.18549410998821259, | |
| "learning_rate": 1.9066429836874844e-06, | |
| "loss": 0.0008, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.7706422018348624, | |
| "grad_norm": 0.12379996478557587, | |
| "learning_rate": 1.7405050206064372e-06, | |
| "loss": 0.0004, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.7828746177370031, | |
| "grad_norm": 0.20588764548301697, | |
| "learning_rate": 1.5804028918769488e-06, | |
| "loss": 0.0009, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.7951070336391437, | |
| "grad_norm": 0.06082570180296898, | |
| "learning_rate": 1.4266330697851955e-06, | |
| "loss": 0.0001, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.8073394495412844, | |
| "grad_norm": 0.07836022228002548, | |
| "learning_rate": 1.2794803006431984e-06, | |
| "loss": 0.0002, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.8195718654434251, | |
| "grad_norm": 0.001425681752152741, | |
| "learning_rate": 1.1392170775041788e-06, | |
| "loss": 0.0, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.8318042813455657, | |
| "grad_norm": 0.02862914465367794, | |
| "learning_rate": 1.0061031355681766e-06, | |
| "loss": 0.0001, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.8440366972477065, | |
| "grad_norm": 0.12509922683238983, | |
| "learning_rate": 8.803849712122292e-07, | |
| "loss": 0.0004, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.8562691131498471, | |
| "grad_norm": 0.0013649528846144676, | |
| "learning_rate": 7.622953855358456e-07, | |
| "loss": 0.0, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.8685015290519877, | |
| "grad_norm": 0.0014051853213459253, | |
| "learning_rate": 6.52053053266945e-07, | |
| "loss": 0.0, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.8807339449541285, | |
| "grad_norm": 0.09148039668798447, | |
| "learning_rate": 5.498621178266167e-07, | |
| "loss": 0.0002, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.8929663608562691, | |
| "grad_norm": 0.0013811088865622878, | |
| "learning_rate": 4.5591181330248534e-07, | |
| "loss": 0.0, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.9051987767584098, | |
| "grad_norm": 0.001428584917448461, | |
| "learning_rate": 3.70376114030751e-07, | |
| "loss": 0.0, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.9174311926605505, | |
| "grad_norm": 0.02665664628148079, | |
| "learning_rate": 2.934134124357646e-07, | |
| "loss": 0.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.9296636085626911, | |
| "grad_norm": 0.041259054094552994, | |
| "learning_rate": 2.2516622572372416e-07, | |
| "loss": 0.0001, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.9418960244648318, | |
| "grad_norm": 0.0013576567871496081, | |
| "learning_rate": 1.6576093197361253e-07, | |
| "loss": 0.0, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.9541284403669725, | |
| "grad_norm": 0.1450626254081726, | |
| "learning_rate": 1.1530753611409151e-07, | |
| "loss": 0.0003, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.9663608562691132, | |
| "grad_norm": 0.12293399125337601, | |
| "learning_rate": 7.389946621969679e-08, | |
| "loss": 0.0002, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.9785932721712538, | |
| "grad_norm": 0.0014255427522584796, | |
| "learning_rate": 4.1613400503550114e-08, | |
| "loss": 0.0, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.9908256880733946, | |
| "grad_norm": 0.0013442779891192913, | |
| "learning_rate": 1.850912532696092e-08, | |
| "loss": 0.0, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.0013442779891192913, | |
| "learning_rate": 4.629424488850065e-09, | |
| "loss": 0.001, | |
| "step": 82 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 82, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.2207070589681664e+16, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |