| { | |
| "best_metric": 1.1512540578842163, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-100", | |
| "epoch": 2.989247311827957, | |
| "eval_steps": 50, | |
| "global_step": 139, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.021505376344086023, | |
| "grad_norm": 0.3019464313983917, | |
| "learning_rate": 1.16e-05, | |
| "loss": 1.4259, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.021505376344086023, | |
| "eval_loss": 1.4605212211608887, | |
| "eval_runtime": 3.307, | |
| "eval_samples_per_second": 188.69, | |
| "eval_steps_per_second": 6.048, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.043010752688172046, | |
| "grad_norm": 0.3766098916530609, | |
| "learning_rate": 2.32e-05, | |
| "loss": 1.3061, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.06451612903225806, | |
| "grad_norm": 0.4208398163318634, | |
| "learning_rate": 3.48e-05, | |
| "loss": 1.2716, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.08602150537634409, | |
| "grad_norm": 0.4907033145427704, | |
| "learning_rate": 4.64e-05, | |
| "loss": 1.3698, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.10752688172043011, | |
| "grad_norm": 0.5819088816642761, | |
| "learning_rate": 5.8e-05, | |
| "loss": 1.5208, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.12903225806451613, | |
| "grad_norm": 0.8909784555435181, | |
| "learning_rate": 6.96e-05, | |
| "loss": 1.6946, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.15053763440860216, | |
| "grad_norm": 0.1904004067182541, | |
| "learning_rate": 8.12e-05, | |
| "loss": 1.3806, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.17204301075268819, | |
| "grad_norm": 0.2766229510307312, | |
| "learning_rate": 9.28e-05, | |
| "loss": 1.2944, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.1935483870967742, | |
| "grad_norm": 0.3707273602485657, | |
| "learning_rate": 0.0001044, | |
| "loss": 1.1893, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.21505376344086022, | |
| "grad_norm": 0.4749780595302582, | |
| "learning_rate": 0.000116, | |
| "loss": 1.2153, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.23655913978494625, | |
| "grad_norm": 0.5096077919006348, | |
| "learning_rate": 0.00011598280125101809, | |
| "loss": 1.3548, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.25806451612903225, | |
| "grad_norm": 0.46667513251304626, | |
| "learning_rate": 0.00011593121520396772, | |
| "loss": 1.4838, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.27956989247311825, | |
| "grad_norm": 0.2550894618034363, | |
| "learning_rate": 0.000115845272452486, | |
| "loss": 1.3795, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.3010752688172043, | |
| "grad_norm": 0.2862931787967682, | |
| "learning_rate": 0.00011572502396580767, | |
| "loss": 1.2467, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.3225806451612903, | |
| "grad_norm": 0.2682102620601654, | |
| "learning_rate": 0.00011557054105853753, | |
| "loss": 1.1907, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.34408602150537637, | |
| "grad_norm": 0.27467212080955505, | |
| "learning_rate": 0.0001153819153483564, | |
| "loss": 1.1003, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.3655913978494624, | |
| "grad_norm": 0.3005955219268799, | |
| "learning_rate": 0.00011515925870168636, | |
| "loss": 1.2234, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.3870967741935484, | |
| "grad_norm": 0.3794748783111572, | |
| "learning_rate": 0.00011490270316734726, | |
| "loss": 1.4082, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.40860215053763443, | |
| "grad_norm": 0.21310873329639435, | |
| "learning_rate": 0.00011461240089824378, | |
| "loss": 1.328, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.43010752688172044, | |
| "grad_norm": 0.20409537851810455, | |
| "learning_rate": 0.0001142885240611295, | |
| "loss": 1.3031, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.45161290322580644, | |
| "grad_norm": 0.22202569246292114, | |
| "learning_rate": 0.0001139312647345018, | |
| "loss": 1.1874, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.4731182795698925, | |
| "grad_norm": 0.2386716902256012, | |
| "learning_rate": 0.00011354083479468755, | |
| "loss": 1.1251, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.4946236559139785, | |
| "grad_norm": 0.26145026087760925, | |
| "learning_rate": 0.00011311746579018779, | |
| "loss": 1.1777, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.5161290322580645, | |
| "grad_norm": 0.30247944593429565, | |
| "learning_rate": 0.00011266140880435544, | |
| "loss": 1.3137, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.5376344086021505, | |
| "grad_norm": 0.587131917476654, | |
| "learning_rate": 0.00011217293430648779, | |
| "loss": 1.4008, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.5591397849462365, | |
| "grad_norm": 0.14670686423778534, | |
| "learning_rate": 0.00011165233199142182, | |
| "loss": 1.2933, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.5806451612903226, | |
| "grad_norm": 0.16923627257347107, | |
| "learning_rate": 0.00011109991060772776, | |
| "loss": 1.1914, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.6021505376344086, | |
| "grad_norm": 0.19446203112602234, | |
| "learning_rate": 0.0001105159977746025, | |
| "loss": 1.1251, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.6236559139784946, | |
| "grad_norm": 0.22473880648612976, | |
| "learning_rate": 0.00010990093978757173, | |
| "loss": 1.1065, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.6451612903225806, | |
| "grad_norm": 0.2743987441062927, | |
| "learning_rate": 0.00010925510141311572, | |
| "loss": 1.2497, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.3620847761631012, | |
| "learning_rate": 0.00010857886567234085, | |
| "loss": 1.3353, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.6881720430107527, | |
| "grad_norm": 0.14775493741035461, | |
| "learning_rate": 0.00010787263361382498, | |
| "loss": 1.2885, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.7096774193548387, | |
| "grad_norm": 0.1633865386247635, | |
| "learning_rate": 0.00010713682407577149, | |
| "loss": 1.2385, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.7311827956989247, | |
| "grad_norm": 0.1605488508939743, | |
| "learning_rate": 0.00010637187343761291, | |
| "loss": 1.0806, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.7526881720430108, | |
| "grad_norm": 0.19970649480819702, | |
| "learning_rate": 0.00010557823536121162, | |
| "loss": 1.1132, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.7741935483870968, | |
| "grad_norm": 0.24861781299114227, | |
| "learning_rate": 0.00010475638052181104, | |
| "loss": 1.1757, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.7956989247311828, | |
| "grad_norm": 0.37152165174484253, | |
| "learning_rate": 0.00010390679632889674, | |
| "loss": 1.3386, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.8172043010752689, | |
| "grad_norm": 0.16140355169773102, | |
| "learning_rate": 0.00010302998663713333, | |
| "loss": 1.3232, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.8387096774193549, | |
| "grad_norm": 0.15119719505310059, | |
| "learning_rate": 0.00010212647144754812, | |
| "loss": 1.2435, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.8602150537634409, | |
| "grad_norm": 0.15187421441078186, | |
| "learning_rate": 0.00010119678659913935, | |
| "loss": 1.0749, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.8817204301075269, | |
| "grad_norm": 0.1807960420846939, | |
| "learning_rate": 0.00010024148345109112, | |
| "loss": 1.0696, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.9032258064516129, | |
| "grad_norm": 0.2298922836780548, | |
| "learning_rate": 9.926112855578431e-05, | |
| "loss": 1.1653, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.9247311827956989, | |
| "grad_norm": 0.29605555534362793, | |
| "learning_rate": 9.825630332279677e-05, | |
| "loss": 1.235, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.946236559139785, | |
| "grad_norm": 0.2054792195558548, | |
| "learning_rate": 9.722760367409236e-05, | |
| "loss": 1.2058, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.967741935483871, | |
| "grad_norm": 0.1517435610294342, | |
| "learning_rate": 9.617563969060338e-05, | |
| "loss": 1.1643, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.989247311827957, | |
| "grad_norm": 0.2478746473789215, | |
| "learning_rate": 9.51010352504157e-05, | |
| "loss": 1.1552, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 1.010752688172043, | |
| "grad_norm": 0.2573375105857849, | |
| "learning_rate": 9.400442765877141e-05, | |
| "loss": 2.141, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 1.032258064516129, | |
| "grad_norm": 0.12741310894489288, | |
| "learning_rate": 9.288646727010848e-05, | |
| "loss": 1.0358, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 1.053763440860215, | |
| "grad_norm": 0.1643674373626709, | |
| "learning_rate": 9.174781710236128e-05, | |
| "loss": 1.1719, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 1.075268817204301, | |
| "grad_norm": 0.19035093486309052, | |
| "learning_rate": 9.058915244375091e-05, | |
| "loss": 1.0132, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.075268817204301, | |
| "eval_loss": 1.1688854694366455, | |
| "eval_runtime": 3.3032, | |
| "eval_samples_per_second": 188.91, | |
| "eval_steps_per_second": 6.055, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.096774193548387, | |
| "grad_norm": 0.24456371366977692, | |
| "learning_rate": 8.94111604522987e-05, | |
| "loss": 1.1068, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 1.118279569892473, | |
| "grad_norm": 0.34046271443367004, | |
| "learning_rate": 8.821453974829996e-05, | |
| "loss": 1.2428, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 1.139784946236559, | |
| "grad_norm": 0.247173473238945, | |
| "learning_rate": 8.7e-05, | |
| "loss": 0.9179, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 1.1612903225806452, | |
| "grad_norm": 0.17861323058605194, | |
| "learning_rate": 8.576826150271813e-05, | |
| "loss": 1.3754, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 1.1827956989247312, | |
| "grad_norm": 0.20689214766025543, | |
| "learning_rate": 8.452005475166903e-05, | |
| "loss": 1.2233, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 1.2043010752688172, | |
| "grad_norm": 0.2131056934595108, | |
| "learning_rate": 8.325612000873509e-05, | |
| "loss": 1.0103, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 1.2258064516129032, | |
| "grad_norm": 0.26632529497146606, | |
| "learning_rate": 8.197720686344642e-05, | |
| "loss": 1.0388, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.2473118279569892, | |
| "grad_norm": 0.29926690459251404, | |
| "learning_rate": 8.068407378842904e-05, | |
| "loss": 1.1619, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 1.2688172043010753, | |
| "grad_norm": 0.21377074718475342, | |
| "learning_rate": 7.937748768958499e-05, | |
| "loss": 0.548, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 1.2903225806451613, | |
| "grad_norm": 0.215475931763649, | |
| "learning_rate": 7.805822345127066e-05, | |
| "loss": 1.9897, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.3118279569892473, | |
| "grad_norm": 0.1549675166606903, | |
| "learning_rate": 7.672706347674388e-05, | |
| "loss": 0.9913, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 0.21127307415008545, | |
| "learning_rate": 7.53847972241514e-05, | |
| "loss": 1.0833, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 1.3548387096774195, | |
| "grad_norm": 0.24489474296569824, | |
| "learning_rate": 7.403222073833276e-05, | |
| "loss": 1.0426, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.3763440860215055, | |
| "grad_norm": 0.3033023178577423, | |
| "learning_rate": 7.267013617871748e-05, | |
| "loss": 1.1059, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 1.3978494623655915, | |
| "grad_norm": 0.22320985794067383, | |
| "learning_rate": 7.129935134359642e-05, | |
| "loss": 0.6576, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.4193548387096775, | |
| "grad_norm": 0.23948603868484497, | |
| "learning_rate": 6.992067919104844e-05, | |
| "loss": 1.8893, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.4408602150537635, | |
| "grad_norm": 0.1648971438407898, | |
| "learning_rate": 6.85349373568073e-05, | |
| "loss": 0.9948, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.4623655913978495, | |
| "grad_norm": 0.20279648900032043, | |
| "learning_rate": 6.714294766935446e-05, | |
| "loss": 1.0688, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.4838709677419355, | |
| "grad_norm": 0.25218087434768677, | |
| "learning_rate": 6.574553566252508e-05, | |
| "loss": 1.0871, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.5053763440860215, | |
| "grad_norm": 0.2963137626647949, | |
| "learning_rate": 6.434353008591673e-05, | |
| "loss": 1.0764, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.5268817204301075, | |
| "grad_norm": 0.41593724489212036, | |
| "learning_rate": 6.293776241339087e-05, | |
| "loss": 1.2876, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.5483870967741935, | |
| "grad_norm": 0.3219810426235199, | |
| "learning_rate": 6.152906634995881e-05, | |
| "loss": 1.1358, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.5698924731182795, | |
| "grad_norm": 0.14590708911418915, | |
| "learning_rate": 6.011827733734423e-05, | |
| "loss": 1.0, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.5913978494623655, | |
| "grad_norm": 0.18871888518333435, | |
| "learning_rate": 5.870623205851586e-05, | |
| "loss": 1.2011, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.6129032258064515, | |
| "grad_norm": 0.20429526269435883, | |
| "learning_rate": 5.729376794148415e-05, | |
| "loss": 1.0139, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.6344086021505375, | |
| "grad_norm": 0.2530740201473236, | |
| "learning_rate": 5.588172266265578e-05, | |
| "loss": 1.053, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.6559139784946235, | |
| "grad_norm": 0.3459080755710602, | |
| "learning_rate": 5.4470933650041196e-05, | |
| "loss": 1.1392, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.6774193548387095, | |
| "grad_norm": 0.2594045102596283, | |
| "learning_rate": 5.3062237586609127e-05, | |
| "loss": 0.8978, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.6989247311827957, | |
| "grad_norm": 0.1727474331855774, | |
| "learning_rate": 5.16564699140833e-05, | |
| "loss": 1.4071, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.7204301075268817, | |
| "grad_norm": 0.17758332192897797, | |
| "learning_rate": 5.025446433747493e-05, | |
| "loss": 1.1052, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.7419354838709677, | |
| "grad_norm": 0.19625383615493774, | |
| "learning_rate": 4.885705233064554e-05, | |
| "loss": 1.012, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.7634408602150538, | |
| "grad_norm": 0.25267520546913147, | |
| "learning_rate": 4.746506264319269e-05, | |
| "loss": 1.0759, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.7849462365591398, | |
| "grad_norm": 0.31696927547454834, | |
| "learning_rate": 4.6079320808951565e-05, | |
| "loss": 1.1305, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.8064516129032258, | |
| "grad_norm": 0.22307011485099792, | |
| "learning_rate": 4.470064865640358e-05, | |
| "loss": 0.5376, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.827956989247312, | |
| "grad_norm": 0.2187909334897995, | |
| "learning_rate": 4.3329863821282514e-05, | |
| "loss": 1.8596, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.849462365591398, | |
| "grad_norm": 0.16456525027751923, | |
| "learning_rate": 4.1967779261667245e-05, | |
| "loss": 1.0951, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.870967741935484, | |
| "grad_norm": 0.19617106020450592, | |
| "learning_rate": 4.06152027758486e-05, | |
| "loss": 1.0345, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.89247311827957, | |
| "grad_norm": 0.23399649560451508, | |
| "learning_rate": 3.9272936523256134e-05, | |
| "loss": 1.0037, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.913978494623656, | |
| "grad_norm": 0.3092529773712158, | |
| "learning_rate": 3.794177654872934e-05, | |
| "loss": 1.1699, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.935483870967742, | |
| "grad_norm": 0.23798301815986633, | |
| "learning_rate": 3.662251231041502e-05, | |
| "loss": 0.5823, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.956989247311828, | |
| "grad_norm": 0.250249445438385, | |
| "learning_rate": 3.531592621157096e-05, | |
| "loss": 1.6714, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.978494623655914, | |
| "grad_norm": 0.2460961937904358, | |
| "learning_rate": 3.402279313655359e-05, | |
| "loss": 1.2035, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.3106490969657898, | |
| "learning_rate": 3.274387999126492e-05, | |
| "loss": 1.5907, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 2.021505376344086, | |
| "grad_norm": 0.14660653471946716, | |
| "learning_rate": 3.1479945248330964e-05, | |
| "loss": 1.2345, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 2.043010752688172, | |
| "grad_norm": 0.17234809696674347, | |
| "learning_rate": 3.023173849728189e-05, | |
| "loss": 1.1059, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 2.064516129032258, | |
| "grad_norm": 0.18329447507858276, | |
| "learning_rate": 2.9000000000000014e-05, | |
| "loss": 0.9879, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 2.086021505376344, | |
| "grad_norm": 0.222031369805336, | |
| "learning_rate": 2.7785460251700053e-05, | |
| "loss": 0.9824, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 2.10752688172043, | |
| "grad_norm": 0.284976065158844, | |
| "learning_rate": 2.6588839547701294e-05, | |
| "loss": 1.0231, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 2.129032258064516, | |
| "grad_norm": 0.3977857828140259, | |
| "learning_rate": 2.541084755624909e-05, | |
| "loss": 1.0588, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 2.150537634408602, | |
| "grad_norm": 0.1437109261751175, | |
| "learning_rate": 2.4252182897638746e-05, | |
| "loss": 1.1997, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.150537634408602, | |
| "eval_loss": 1.1512540578842163, | |
| "eval_runtime": 3.6809, | |
| "eval_samples_per_second": 169.524, | |
| "eval_steps_per_second": 5.433, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.172043010752688, | |
| "grad_norm": 0.17709468305110931, | |
| "learning_rate": 2.3113532729891522e-05, | |
| "loss": 1.1533, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 2.193548387096774, | |
| "grad_norm": 0.1920265406370163, | |
| "learning_rate": 2.1995572341228588e-05, | |
| "loss": 1.0311, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 2.21505376344086, | |
| "grad_norm": 0.2193988710641861, | |
| "learning_rate": 2.089896474958432e-05, | |
| "loss": 0.9886, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 2.236559139784946, | |
| "grad_norm": 0.2702076733112335, | |
| "learning_rate": 1.9824360309396626e-05, | |
| "loss": 1.0325, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 2.258064516129032, | |
| "grad_norm": 0.36857450008392334, | |
| "learning_rate": 1.877239632590764e-05, | |
| "loss": 1.0513, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 2.279569892473118, | |
| "grad_norm": 0.1756919026374817, | |
| "learning_rate": 1.774369667720323e-05, | |
| "loss": 1.2273, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 2.3010752688172045, | |
| "grad_norm": 0.17812420427799225, | |
| "learning_rate": 1.67388714442157e-05, | |
| "loss": 1.1593, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 2.3225806451612905, | |
| "grad_norm": 0.18077994883060455, | |
| "learning_rate": 1.575851654890888e-05, | |
| "loss": 1.0257, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 2.3440860215053765, | |
| "grad_norm": 0.21289852261543274, | |
| "learning_rate": 1.4803213400860651e-05, | |
| "loss": 0.9742, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 2.3655913978494625, | |
| "grad_norm": 0.2521963119506836, | |
| "learning_rate": 1.3873528552451873e-05, | |
| "loss": 0.9653, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 2.3870967741935485, | |
| "grad_norm": 0.32557451725006104, | |
| "learning_rate": 1.2970013362866697e-05, | |
| "loss": 1.013, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 2.4086021505376345, | |
| "grad_norm": 0.2451455295085907, | |
| "learning_rate": 1.2093203671103267e-05, | |
| "loss": 1.2257, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 2.4301075268817205, | |
| "grad_norm": 0.1599583923816681, | |
| "learning_rate": 1.1243619478188961e-05, | |
| "loss": 1.1622, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 2.4516129032258065, | |
| "grad_norm": 0.17558430135250092, | |
| "learning_rate": 1.0421764638788365e-05, | |
| "loss": 1.0423, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 2.4731182795698925, | |
| "grad_norm": 0.20616887509822845, | |
| "learning_rate": 9.628126562387086e-06, | |
| "loss": 0.9993, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 2.4946236559139785, | |
| "grad_norm": 0.24216623604297638, | |
| "learning_rate": 8.863175924228501e-06, | |
| "loss": 0.9776, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 2.5161290322580645, | |
| "grad_norm": 0.3300863802433014, | |
| "learning_rate": 8.127366386175014e-06, | |
| "loss": 1.054, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 2.5376344086021505, | |
| "grad_norm": 0.5971299409866333, | |
| "learning_rate": 7.421134327659152e-06, | |
| "loss": 1.0757, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 2.5591397849462365, | |
| "grad_norm": 0.14677409827709198, | |
| "learning_rate": 6.744898586884296e-06, | |
| "loss": 1.2074, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 2.5806451612903225, | |
| "grad_norm": 0.17917855083942413, | |
| "learning_rate": 6.099060212428274e-06, | |
| "loss": 1.0735, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.6021505376344085, | |
| "grad_norm": 0.2020336389541626, | |
| "learning_rate": 5.484002225397496e-06, | |
| "loss": 0.9547, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 2.6236559139784945, | |
| "grad_norm": 0.23973648250102997, | |
| "learning_rate": 4.900089392272253e-06, | |
| "loss": 0.9674, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 2.6451612903225805, | |
| "grad_norm": 0.2944413423538208, | |
| "learning_rate": 4.347668008578187e-06, | |
| "loss": 1.0609, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 0.40398478507995605, | |
| "learning_rate": 3.8270656935122204e-06, | |
| "loss": 1.0409, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 2.688172043010753, | |
| "grad_norm": 0.15678246319293976, | |
| "learning_rate": 3.3385911956445625e-06, | |
| "loss": 1.2516, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 2.709677419354839, | |
| "grad_norm": 0.18234221637248993, | |
| "learning_rate": 2.8825342098122193e-06, | |
| "loss": 1.1226, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 2.731182795698925, | |
| "grad_norm": 0.19632849097251892, | |
| "learning_rate": 2.4591652053124607e-06, | |
| "loss": 1.0385, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 2.752688172043011, | |
| "grad_norm": 0.22743487358093262, | |
| "learning_rate": 2.068735265498204e-06, | |
| "loss": 0.9646, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 2.774193548387097, | |
| "grad_norm": 0.28680744767189026, | |
| "learning_rate": 1.711475938870494e-06, | |
| "loss": 0.9698, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 2.795698924731183, | |
| "grad_norm": 0.37023359537124634, | |
| "learning_rate": 1.3875991017562305e-06, | |
| "loss": 1.0446, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.817204301075269, | |
| "grad_norm": 0.17839553952217102, | |
| "learning_rate": 1.0972968326527323e-06, | |
| "loss": 1.2334, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 2.838709677419355, | |
| "grad_norm": 0.17378120124340057, | |
| "learning_rate": 8.407412983136427e-07, | |
| "loss": 1.1619, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 2.860215053763441, | |
| "grad_norm": 0.18641377985477448, | |
| "learning_rate": 6.180846516436054e-07, | |
| "loss": 1.0182, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 2.881720430107527, | |
| "grad_norm": 0.2144346982240677, | |
| "learning_rate": 4.294589414624692e-07, | |
| "loss": 0.9406, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 2.903225806451613, | |
| "grad_norm": 0.27374863624572754, | |
| "learning_rate": 2.7497603419232487e-07, | |
| "loss": 1.031, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 2.924731182795699, | |
| "grad_norm": 0.3471708297729492, | |
| "learning_rate": 1.5472754751400464e-07, | |
| "loss": 1.0266, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.946236559139785, | |
| "grad_norm": 0.24548716843128204, | |
| "learning_rate": 6.878479603226562e-08, | |
| "loss": 1.1736, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 2.967741935483871, | |
| "grad_norm": 0.18722181022167206, | |
| "learning_rate": 1.71987489819172e-08, | |
| "loss": 1.0664, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.989247311827957, | |
| "grad_norm": 0.27170056104660034, | |
| "learning_rate": 0.0, | |
| "loss": 0.9598, | |
| "step": 139 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 139, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.224234711308042e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |