diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,3479 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 5.0, + "eval_steps": 500, + "global_step": 1560, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.016025641025641024, + "grad_norm": 6.815881626720253, + "learning_rate": 1.0256410256410257e-06, + "loss": 0.9447, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4387686252593994, + "step": 5, + "valid_targets_mean": 2775.8, + "valid_targets_min": 999 + }, + { + "epoch": 0.03205128205128205, + "grad_norm": 5.611249301686906, + "learning_rate": 2.307692307692308e-06, + "loss": 0.9287, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4332679808139801, + "step": 10, + "valid_targets_mean": 3038.2, + "valid_targets_min": 1540 + }, + { + "epoch": 0.04807692307692308, + "grad_norm": 4.11592117125762, + "learning_rate": 3.58974358974359e-06, + "loss": 0.9119, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4442380368709564, + "step": 15, + "valid_targets_mean": 2672.2, + "valid_targets_min": 1311 + }, + { + "epoch": 0.0641025641025641, + "grad_norm": 2.4171662234913085, + "learning_rate": 4.871794871794872e-06, + "loss": 0.8644, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.46687430143356323, + "step": 20, + "valid_targets_mean": 3144.1, + "valid_targets_min": 1686 + }, + { + "epoch": 0.08012820512820513, + "grad_norm": 1.5505428723718302, + "learning_rate": 6.153846153846155e-06, + "loss": 0.8381, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4025786817073822, + "step": 25, + "valid_targets_mean": 2429.6, + "valid_targets_min": 1163 + }, + { + "epoch": 0.09615384615384616, + "grad_norm": 1.0869371534735193, + "learning_rate": 7.435897435897437e-06, + "loss": 0.814, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4357997179031372, + "step": 30, + "valid_targets_mean": 3333.8, + "valid_targets_min": 2413 + }, + { + "epoch": 0.11217948717948718, + "grad_norm": 0.950971876737434, + "learning_rate": 8.717948717948719e-06, + "loss": 0.7732, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4076507091522217, + "step": 35, + "valid_targets_mean": 3007.8, + "valid_targets_min": 885 + }, + { + "epoch": 0.1282051282051282, + "grad_norm": 0.7582939151199518, + "learning_rate": 1e-05, + "loss": 0.7849, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4013994336128235, + "step": 40, + "valid_targets_mean": 2966.4, + "valid_targets_min": 1265 + }, + { + "epoch": 0.14423076923076922, + "grad_norm": 0.6819204506609364, + "learning_rate": 1.1282051282051283e-05, + "loss": 0.7513, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3817106783390045, + "step": 45, + "valid_targets_mean": 2599.5, + "valid_targets_min": 1385 + }, + { + "epoch": 0.16025641025641027, + "grad_norm": 0.6391357961237331, + "learning_rate": 1.2564102564102565e-05, + "loss": 0.74, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.37557607889175415, + "step": 50, + "valid_targets_mean": 2784.7, + "valid_targets_min": 1317 + }, + { + "epoch": 0.1762820512820513, + "grad_norm": 0.5551476820310136, + "learning_rate": 1.3846153846153847e-05, + "loss": 0.718, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3204777240753174, + "step": 55, + "valid_targets_mean": 2567.1, + "valid_targets_min": 1304 + }, + { + "epoch": 0.19230769230769232, + "grad_norm": 0.6077071082054613, + "learning_rate": 1.5128205128205129e-05, + "loss": 0.7196, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.35047832131385803, + "step": 60, + "valid_targets_mean": 2480.4, + "valid_targets_min": 1230 + }, + { + "epoch": 0.20833333333333334, + "grad_norm": 0.6028149033641808, + "learning_rate": 1.641025641025641e-05, + "loss": 0.7065, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3641282916069031, + "step": 65, + "valid_targets_mean": 2665.4, + "valid_targets_min": 1816 + }, + { + "epoch": 0.22435897435897437, + "grad_norm": 0.5364533780203461, + "learning_rate": 1.7692307692307694e-05, + "loss": 0.6856, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3332011103630066, + "step": 70, + "valid_targets_mean": 2644.1, + "valid_targets_min": 1139 + }, + { + "epoch": 0.2403846153846154, + "grad_norm": 0.5296770070879561, + "learning_rate": 1.8974358974358975e-05, + "loss": 0.6856, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3156450092792511, + "step": 75, + "valid_targets_mean": 2270.6, + "valid_targets_min": 601 + }, + { + "epoch": 0.2564102564102564, + "grad_norm": 0.5261553976863995, + "learning_rate": 2.025641025641026e-05, + "loss": 0.6877, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30606338381767273, + "step": 80, + "valid_targets_mean": 2451.2, + "valid_targets_min": 1292 + }, + { + "epoch": 0.2724358974358974, + "grad_norm": 0.49894510222939537, + "learning_rate": 2.153846153846154e-05, + "loss": 0.6759, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28691911697387695, + "step": 85, + "valid_targets_mean": 2653.8, + "valid_targets_min": 1086 + }, + { + "epoch": 0.28846153846153844, + "grad_norm": 0.5019026732938922, + "learning_rate": 2.2820512820512822e-05, + "loss": 0.6572, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3042421042919159, + "step": 90, + "valid_targets_mean": 2732.5, + "valid_targets_min": 1498 + }, + { + "epoch": 0.30448717948717946, + "grad_norm": 0.5087948648074874, + "learning_rate": 2.4102564102564103e-05, + "loss": 0.6372, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3090476393699646, + "step": 95, + "valid_targets_mean": 2741.7, + "valid_targets_min": 1313 + }, + { + "epoch": 0.32051282051282054, + "grad_norm": 0.5093262106093662, + "learning_rate": 2.5384615384615386e-05, + "loss": 0.654, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3319040834903717, + "step": 100, + "valid_targets_mean": 3045.6, + "valid_targets_min": 1493 + }, + { + "epoch": 0.33653846153846156, + "grad_norm": 0.5216909158213553, + "learning_rate": 2.6666666666666667e-05, + "loss": 0.6542, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3026030361652374, + "step": 105, + "valid_targets_mean": 2332.0, + "valid_targets_min": 1348 + }, + { + "epoch": 0.3525641025641026, + "grad_norm": 0.5594308657153152, + "learning_rate": 2.794871794871795e-05, + "loss": 0.6469, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3094126582145691, + "step": 110, + "valid_targets_mean": 2481.8, + "valid_targets_min": 1253 + }, + { + "epoch": 0.3685897435897436, + "grad_norm": 0.5697299366312509, + "learning_rate": 2.923076923076923e-05, + "loss": 0.6457, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.345976859331131, + "step": 115, + "valid_targets_mean": 2500.8, + "valid_targets_min": 1002 + }, + { + "epoch": 0.38461538461538464, + "grad_norm": 0.481422178674973, + "learning_rate": 3.0512820512820514e-05, + "loss": 0.6334, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3132162094116211, + "step": 120, + "valid_targets_mean": 3228.7, + "valid_targets_min": 1906 + }, + { + "epoch": 0.40064102564102566, + "grad_norm": 0.4861239944985147, + "learning_rate": 3.1794871794871795e-05, + "loss": 0.6196, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32609227299690247, + "step": 125, + "valid_targets_mean": 3174.4, + "valid_targets_min": 1632 + }, + { + "epoch": 0.4166666666666667, + "grad_norm": 0.5222849869950157, + "learning_rate": 3.307692307692308e-05, + "loss": 0.6388, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31350448727607727, + "step": 130, + "valid_targets_mean": 2809.1, + "valid_targets_min": 1061 + }, + { + "epoch": 0.4326923076923077, + "grad_norm": 0.5379788899851837, + "learning_rate": 3.435897435897436e-05, + "loss": 0.6313, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3066648244857788, + "step": 135, + "valid_targets_mean": 2802.4, + "valid_targets_min": 1190 + }, + { + "epoch": 0.44871794871794873, + "grad_norm": 0.5420454343087308, + "learning_rate": 3.5641025641025646e-05, + "loss": 0.6318, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30834197998046875, + "step": 140, + "valid_targets_mean": 2687.2, + "valid_targets_min": 1391 + }, + { + "epoch": 0.46474358974358976, + "grad_norm": 0.5214859365271105, + "learning_rate": 3.692307692307693e-05, + "loss": 0.6175, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33880215883255005, + "step": 145, + "valid_targets_mean": 2832.4, + "valid_targets_min": 829 + }, + { + "epoch": 0.4807692307692308, + "grad_norm": 0.46621326586782824, + "learning_rate": 3.820512820512821e-05, + "loss": 0.6109, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2916651666164398, + "step": 150, + "valid_targets_mean": 3312.8, + "valid_targets_min": 1660 + }, + { + "epoch": 0.4967948717948718, + "grad_norm": 0.5182511120208516, + "learning_rate": 3.948717948717949e-05, + "loss": 0.6284, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3359749913215637, + "step": 155, + "valid_targets_mean": 2667.7, + "valid_targets_min": 1182 + }, + { + "epoch": 0.5128205128205128, + "grad_norm": 0.5076341907952538, + "learning_rate": 3.999954938420724e-05, + "loss": 0.6205, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29317110776901245, + "step": 160, + "valid_targets_mean": 2856.9, + "valid_targets_min": 1035 + }, + { + "epoch": 0.5288461538461539, + "grad_norm": 0.4968550139113613, + "learning_rate": 3.9996795694563096e-05, + "loss": 0.6311, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3237057626247406, + "step": 165, + "valid_targets_mean": 3255.5, + "valid_targets_min": 930 + }, + { + "epoch": 0.5448717948717948, + "grad_norm": 0.5432075062569038, + "learning_rate": 3.9991539001644015e-05, + "loss": 0.6201, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3103916049003601, + "step": 170, + "valid_targets_mean": 2773.2, + "valid_targets_min": 1355 + }, + { + "epoch": 0.5608974358974359, + "grad_norm": 0.5320854842566946, + "learning_rate": 3.998377996343139e-05, + "loss": 0.6195, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30836841464042664, + "step": 175, + "valid_targets_mean": 2700.5, + "valid_targets_min": 1612 + }, + { + "epoch": 0.5769230769230769, + "grad_norm": 0.4787446461742092, + "learning_rate": 3.9973519551125746e-05, + "loss": 0.6079, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2759506106376648, + "step": 180, + "valid_targets_mean": 3075.6, + "valid_targets_min": 1686 + }, + { + "epoch": 0.592948717948718, + "grad_norm": 0.5135765340971998, + "learning_rate": 3.99607590490251e-05, + "loss": 0.6211, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3288421928882599, + "step": 185, + "valid_targets_mean": 3388.7, + "valid_targets_min": 1348 + }, + { + "epoch": 0.6089743589743589, + "grad_norm": 0.7381909636900029, + "learning_rate": 3.994550005436431e-05, + "loss": 0.6308, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31162500381469727, + "step": 190, + "valid_targets_mean": 2567.4, + "valid_targets_min": 1067 + }, + { + "epoch": 0.625, + "grad_norm": 0.526645919780682, + "learning_rate": 3.992774447711503e-05, + "loss": 0.6138, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31996357440948486, + "step": 195, + "valid_targets_mean": 2892.8, + "valid_targets_min": 1456 + }, + { + "epoch": 0.6410256410256411, + "grad_norm": 0.5012857363859938, + "learning_rate": 3.990749453974676e-05, + "loss": 0.6045, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2793130874633789, + "step": 200, + "valid_targets_mean": 2781.1, + "valid_targets_min": 1162 + }, + { + "epoch": 0.657051282051282, + "grad_norm": 0.5871837857805097, + "learning_rate": 3.9884752776948564e-05, + "loss": 0.6005, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.309184730052948, + "step": 205, + "valid_targets_mean": 2420.6, + "valid_targets_min": 900 + }, + { + "epoch": 0.6730769230769231, + "grad_norm": 0.5541790395378353, + "learning_rate": 3.985952203531184e-05, + "loss": 0.6318, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2788581848144531, + "step": 210, + "valid_targets_mean": 2503.8, + "valid_targets_min": 1417 + }, + { + "epoch": 0.6891025641025641, + "grad_norm": 0.5136635755544906, + "learning_rate": 3.983180547297404e-05, + "loss": 0.6149, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2936289310455322, + "step": 215, + "valid_targets_mean": 2670.8, + "valid_targets_min": 854 + }, + { + "epoch": 0.7051282051282052, + "grad_norm": 0.5526474169879116, + "learning_rate": 3.9801606559223286e-05, + "loss": 0.6043, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29138144850730896, + "step": 220, + "valid_targets_mean": 2094.0, + "valid_targets_min": 669 + }, + { + "epoch": 0.7211538461538461, + "grad_norm": 0.5005790560858441, + "learning_rate": 3.9768929074064206e-05, + "loss": 0.6046, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3046284317970276, + "step": 225, + "valid_targets_mean": 2856.6, + "valid_targets_min": 804 + }, + { + "epoch": 0.7371794871794872, + "grad_norm": 0.5319298199331578, + "learning_rate": 3.973377710774474e-05, + "loss": 0.5987, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2717939019203186, + "step": 230, + "valid_targets_mean": 2771.9, + "valid_targets_min": 1624 + }, + { + "epoch": 0.7532051282051282, + "grad_norm": 0.5462975979383605, + "learning_rate": 3.9696155060244166e-05, + "loss": 0.5994, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2797490358352661, + "step": 235, + "valid_targets_mean": 2263.8, + "valid_targets_min": 463 + }, + { + "epoch": 0.7692307692307693, + "grad_norm": 0.5625047884049433, + "learning_rate": 3.965606764072237e-05, + "loss": 0.6007, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2672920227050781, + "step": 240, + "valid_targets_mean": 2372.8, + "valid_targets_min": 886 + }, + { + "epoch": 0.7852564102564102, + "grad_norm": 0.5159405507851967, + "learning_rate": 3.96135198669304e-05, + "loss": 0.588, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26713523268699646, + "step": 245, + "valid_targets_mean": 2816.5, + "valid_targets_min": 1356 + }, + { + "epoch": 0.8012820512820513, + "grad_norm": 0.55075615565925, + "learning_rate": 3.956851706458236e-05, + "loss": 0.5946, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32864856719970703, + "step": 250, + "valid_targets_mean": 3047.9, + "valid_targets_min": 1028 + }, + { + "epoch": 0.8173076923076923, + "grad_norm": 0.4623296031936575, + "learning_rate": 3.952106486668884e-05, + "loss": 0.596, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26747170090675354, + "step": 255, + "valid_targets_mean": 2893.2, + "valid_targets_min": 1626 + }, + { + "epoch": 0.8333333333333334, + "grad_norm": 0.553339516613882, + "learning_rate": 3.9471169212851774e-05, + "loss": 0.5915, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3047725260257721, + "step": 260, + "valid_targets_mean": 2358.8, + "valid_targets_min": 1123 + }, + { + "epoch": 0.8493589743589743, + "grad_norm": 0.4923474896251558, + "learning_rate": 3.9418836348521045e-05, + "loss": 0.5974, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2745221257209778, + "step": 265, + "valid_targets_mean": 2917.7, + "valid_targets_min": 1276 + }, + { + "epoch": 0.8653846153846154, + "grad_norm": 0.5493468058767709, + "learning_rate": 3.936407282421267e-05, + "loss": 0.6092, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29619723558425903, + "step": 270, + "valid_targets_mean": 2358.4, + "valid_targets_min": 586 + }, + { + "epoch": 0.8814102564102564, + "grad_norm": 0.5064322514275459, + "learning_rate": 3.930688549468894e-05, + "loss": 0.5912, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2947567105293274, + "step": 275, + "valid_targets_mean": 2665.6, + "valid_targets_min": 1349 + }, + { + "epoch": 0.8974358974358975, + "grad_norm": 0.5308193147953131, + "learning_rate": 3.924728151810034e-05, + "loss": 0.6003, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29639333486557007, + "step": 280, + "valid_targets_mean": 2355.9, + "valid_targets_min": 875 + }, + { + "epoch": 0.9134615384615384, + "grad_norm": 0.5219547698552953, + "learning_rate": 3.9185268355089606e-05, + "loss": 0.6013, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3138948082923889, + "step": 285, + "valid_targets_mean": 2612.1, + "valid_targets_min": 1453 + }, + { + "epoch": 0.9294871794871795, + "grad_norm": 0.5477432963074507, + "learning_rate": 3.912085376785788e-05, + "loss": 0.5957, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2890666723251343, + "step": 290, + "valid_targets_mean": 2642.9, + "valid_targets_min": 1224 + }, + { + "epoch": 0.9455128205128205, + "grad_norm": 0.530177701496871, + "learning_rate": 3.9054045819193074e-05, + "loss": 0.5888, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2837531864643097, + "step": 295, + "valid_targets_mean": 2790.3, + "valid_targets_min": 723 + }, + { + "epoch": 0.9615384615384616, + "grad_norm": 0.5225578306865748, + "learning_rate": 3.898485287146068e-05, + "loss": 0.5877, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26135194301605225, + "step": 300, + "valid_targets_mean": 2409.1, + "valid_targets_min": 1671 + }, + { + "epoch": 0.9775641025641025, + "grad_norm": 0.8252101210922115, + "learning_rate": 3.8913283585557054e-05, + "loss": 0.597, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28889599442481995, + "step": 305, + "valid_targets_mean": 2582.6, + "valid_targets_min": 1479 + }, + { + "epoch": 0.9935897435897436, + "grad_norm": 0.49951767985407336, + "learning_rate": 3.8839346919825304e-05, + "loss": 0.5756, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2502233386039734, + "step": 310, + "valid_targets_mean": 2504.4, + "valid_targets_min": 905 + }, + { + "epoch": 1.0096153846153846, + "grad_norm": 0.4813342649549324, + "learning_rate": 3.876305212893399e-05, + "loss": 0.5868, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26282885670661926, + "step": 315, + "valid_targets_mean": 2381.0, + "valid_targets_min": 1152 + }, + { + "epoch": 1.0256410256410255, + "grad_norm": 0.5610493952941918, + "learning_rate": 3.868440876271871e-05, + "loss": 0.565, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28068363666534424, + "step": 320, + "valid_targets_mean": 2664.7, + "valid_targets_min": 920 + }, + { + "epoch": 1.0416666666666667, + "grad_norm": 0.5516739165526598, + "learning_rate": 3.860342666498677e-05, + "loss": 0.5807, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28632718324661255, + "step": 325, + "valid_targets_mean": 2595.9, + "valid_targets_min": 1536 + }, + { + "epoch": 1.0576923076923077, + "grad_norm": 0.7325610152519612, + "learning_rate": 3.8520115972284975e-05, + "loss": 0.5688, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2912096083164215, + "step": 330, + "valid_targets_mean": 2864.9, + "valid_targets_min": 1002 + }, + { + "epoch": 1.0737179487179487, + "grad_norm": 0.5182656532810745, + "learning_rate": 3.843448711263089e-05, + "loss": 0.5613, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29608041048049927, + "step": 335, + "valid_targets_mean": 2716.9, + "valid_targets_min": 1625 + }, + { + "epoch": 1.0897435897435896, + "grad_norm": 0.5019346432841351, + "learning_rate": 3.8346550804207544e-05, + "loss": 0.5701, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2656922936439514, + "step": 340, + "valid_targets_mean": 2720.6, + "valid_targets_min": 1109 + }, + { + "epoch": 1.1057692307692308, + "grad_norm": 0.520447499485171, + "learning_rate": 3.825631805402182e-05, + "loss": 0.5678, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27221739292144775, + "step": 345, + "valid_targets_mean": 2582.6, + "valid_targets_min": 1527 + }, + { + "epoch": 1.1217948717948718, + "grad_norm": 0.5141310110105949, + "learning_rate": 3.816380015652672e-05, + "loss": 0.5816, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3048447370529175, + "step": 350, + "valid_targets_mean": 2909.9, + "valid_targets_min": 1531 + }, + { + "epoch": 1.1378205128205128, + "grad_norm": 0.5273829773882149, + "learning_rate": 3.806900869220765e-05, + "loss": 0.5783, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32696932554244995, + "step": 355, + "valid_targets_mean": 2834.4, + "valid_targets_min": 1602 + }, + { + "epoch": 1.1538461538461537, + "grad_norm": 0.5074759814713276, + "learning_rate": 3.797195552613284e-05, + "loss": 0.5603, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.277593731880188, + "step": 360, + "valid_targets_mean": 3161.1, + "valid_targets_min": 1578 + }, + { + "epoch": 1.169871794871795, + "grad_norm": 0.5468399046124904, + "learning_rate": 3.787265280646825e-05, + "loss": 0.5523, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32178542017936707, + "step": 365, + "valid_targets_mean": 3033.4, + "valid_targets_min": 1516 + }, + { + "epoch": 1.185897435897436, + "grad_norm": 0.5410326948331472, + "learning_rate": 3.7771112962956936e-05, + "loss": 0.5578, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24614498019218445, + "step": 370, + "valid_targets_mean": 2371.9, + "valid_targets_min": 1089 + }, + { + "epoch": 1.2019230769230769, + "grad_norm": 0.5374891279143514, + "learning_rate": 3.7667348705363227e-05, + "loss": 0.5558, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27958446741104126, + "step": 375, + "valid_targets_mean": 2468.5, + "valid_targets_min": 1160 + }, + { + "epoch": 1.217948717948718, + "grad_norm": 0.5332748254418984, + "learning_rate": 3.7561373021881885e-05, + "loss": 0.5655, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3065238893032074, + "step": 380, + "valid_targets_mean": 3105.9, + "valid_targets_min": 1637 + }, + { + "epoch": 1.233974358974359, + "grad_norm": 0.5173446868519098, + "learning_rate": 3.745319917751229e-05, + "loss": 0.579, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3173753619194031, + "step": 385, + "valid_targets_mean": 2959.2, + "valid_targets_min": 1637 + }, + { + "epoch": 1.25, + "grad_norm": 0.529087530825308, + "learning_rate": 3.734284071239811e-05, + "loss": 0.5773, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3165820837020874, + "step": 390, + "valid_targets_mean": 2732.4, + "valid_targets_min": 1148 + }, + { + "epoch": 1.266025641025641, + "grad_norm": 0.5487377230267219, + "learning_rate": 3.7230311440132494e-05, + "loss": 0.5721, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27153822779655457, + "step": 395, + "valid_targets_mean": 2447.1, + "valid_targets_min": 789 + }, + { + "epoch": 1.282051282051282, + "grad_norm": 0.5533430874813212, + "learning_rate": 3.711562544602895e-05, + "loss": 0.5671, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30601072311401367, + "step": 400, + "valid_targets_mean": 2748.7, + "valid_targets_min": 1332 + }, + { + "epoch": 1.2980769230769231, + "grad_norm": 0.48601670442445527, + "learning_rate": 3.699879708535838e-05, + "loss": 0.5651, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2588644325733185, + "step": 405, + "valid_targets_mean": 2897.8, + "valid_targets_min": 707 + }, + { + "epoch": 1.314102564102564, + "grad_norm": 0.504375358351585, + "learning_rate": 3.687984098155212e-05, + "loss": 0.577, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2488366812467575, + "step": 410, + "valid_targets_mean": 2508.9, + "valid_targets_min": 1226 + }, + { + "epoch": 1.330128205128205, + "grad_norm": 0.4906340385259277, + "learning_rate": 3.6758772024371626e-05, + "loss": 0.563, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2608366310596466, + "step": 415, + "valid_targets_mean": 2512.8, + "valid_targets_min": 1285 + }, + { + "epoch": 1.3461538461538463, + "grad_norm": 0.5421789236972442, + "learning_rate": 3.663560536804465e-05, + "loss": 0.5626, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2818584740161896, + "step": 420, + "valid_targets_mean": 2951.8, + "valid_targets_min": 1396 + }, + { + "epoch": 1.3621794871794872, + "grad_norm": 0.5810036665050738, + "learning_rate": 3.65103564293684e-05, + "loss": 0.5765, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30127882957458496, + "step": 425, + "valid_targets_mean": 2694.9, + "valid_targets_min": 1274 + }, + { + "epoch": 1.3782051282051282, + "grad_norm": 0.47903358727394846, + "learning_rate": 3.638304088577984e-05, + "loss": 0.5727, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.256487101316452, + "step": 430, + "valid_targets_mean": 2993.9, + "valid_targets_min": 1075 + }, + { + "epoch": 1.3942307692307692, + "grad_norm": 0.498296258427307, + "learning_rate": 3.625367467339329e-05, + "loss": 0.5612, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2921409606933594, + "step": 435, + "valid_targets_mean": 3052.1, + "valid_targets_min": 1811 + }, + { + "epoch": 1.4102564102564101, + "grad_norm": 0.5769087450477887, + "learning_rate": 3.612227398500575e-05, + "loss": 0.5717, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29023826122283936, + "step": 440, + "valid_targets_mean": 2366.1, + "valid_targets_min": 575 + }, + { + "epoch": 1.4262820512820513, + "grad_norm": 0.5460088607922002, + "learning_rate": 3.598885526807003e-05, + "loss": 0.5783, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28189247846603394, + "step": 445, + "valid_targets_mean": 2592.5, + "valid_targets_min": 1339 + }, + { + "epoch": 1.4423076923076923, + "grad_norm": 0.5847261044409127, + "learning_rate": 3.585343522263599e-05, + "loss": 0.5667, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2702714204788208, + "step": 450, + "valid_targets_mean": 2488.9, + "valid_targets_min": 1134 + }, + { + "epoch": 1.4583333333333333, + "grad_norm": 0.5145747032617356, + "learning_rate": 3.571603079926024e-05, + "loss": 0.5757, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27167177200317383, + "step": 455, + "valid_targets_mean": 2837.7, + "valid_targets_min": 1581 + }, + { + "epoch": 1.4743589743589745, + "grad_norm": 0.5168757591741376, + "learning_rate": 3.5576659196884395e-05, + "loss": 0.5769, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2823517322540283, + "step": 460, + "valid_targets_mean": 2455.3, + "valid_targets_min": 1044 + }, + { + "epoch": 1.4903846153846154, + "grad_norm": 0.5330528082012648, + "learning_rate": 3.5435337860682304e-05, + "loss": 0.5621, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27999985218048096, + "step": 465, + "valid_targets_mean": 2777.6, + "valid_targets_min": 998 + }, + { + "epoch": 1.5064102564102564, + "grad_norm": 0.5412004827054049, + "learning_rate": 3.529208447987641e-05, + "loss": 0.57, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2921620309352875, + "step": 470, + "valid_targets_mean": 2736.0, + "valid_targets_min": 1853 + }, + { + "epoch": 1.5224358974358974, + "grad_norm": 0.5405160852428765, + "learning_rate": 3.5146916985523604e-05, + "loss": 0.5548, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28153955936431885, + "step": 475, + "valid_targets_mean": 2858.3, + "valid_targets_min": 1417 + }, + { + "epoch": 1.5384615384615383, + "grad_norm": 0.49547704458976594, + "learning_rate": 3.499985354827079e-05, + "loss": 0.5638, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2561967074871063, + "step": 480, + "valid_targets_mean": 2891.4, + "valid_targets_min": 1198 + }, + { + "epoch": 1.5544871794871795, + "grad_norm": 0.565058238537899, + "learning_rate": 3.485091257608047e-05, + "loss": 0.5656, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2924017906188965, + "step": 485, + "valid_targets_mean": 2463.4, + "valid_targets_min": 463 + }, + { + "epoch": 1.5705128205128205, + "grad_norm": 0.497498184926292, + "learning_rate": 3.4700112711926574e-05, + "loss": 0.5619, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27482593059539795, + "step": 490, + "valid_targets_mean": 2584.8, + "valid_targets_min": 1259 + }, + { + "epoch": 1.5865384615384617, + "grad_norm": 0.5287840023150424, + "learning_rate": 3.4547472831460976e-05, + "loss": 0.5585, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.297797828912735, + "step": 495, + "valid_targets_mean": 2593.3, + "valid_targets_min": 626 + }, + { + "epoch": 1.6025641025641026, + "grad_norm": 0.4679531756830481, + "learning_rate": 3.439301204065077e-05, + "loss": 0.5583, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25573623180389404, + "step": 500, + "valid_targets_mean": 2744.5, + "valid_targets_min": 1001 + }, + { + "epoch": 1.6185897435897436, + "grad_norm": 0.5475195616735163, + "learning_rate": 3.423674967338681e-05, + "loss": 0.5775, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3060113191604614, + "step": 505, + "valid_targets_mean": 2838.1, + "valid_targets_min": 1170 + }, + { + "epoch": 1.6346153846153846, + "grad_norm": 0.5235316750346353, + "learning_rate": 3.407870528906366e-05, + "loss": 0.5745, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.305141419172287, + "step": 510, + "valid_targets_mean": 2863.4, + "valid_targets_min": 847 + }, + { + "epoch": 1.6506410256410255, + "grad_norm": 0.4938595683656762, + "learning_rate": 3.391889867013134e-05, + "loss": 0.5709, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28400900959968567, + "step": 515, + "valid_targets_mean": 2622.2, + "valid_targets_min": 1127 + }, + { + "epoch": 1.6666666666666665, + "grad_norm": 0.4946310276214785, + "learning_rate": 3.375734981961918e-05, + "loss": 0.5653, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2853955626487732, + "step": 520, + "valid_targets_mean": 2732.2, + "valid_targets_min": 1004 + }, + { + "epoch": 1.6826923076923077, + "grad_norm": 0.4903094742766738, + "learning_rate": 3.359407895863199e-05, + "loss": 0.5541, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.273275226354599, + "step": 525, + "valid_targets_mean": 2855.8, + "valid_targets_min": 1604 + }, + { + "epoch": 1.6987179487179487, + "grad_norm": 0.5052281623241814, + "learning_rate": 3.342910652381902e-05, + "loss": 0.5421, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2506556808948517, + "step": 530, + "valid_targets_mean": 2644.6, + "valid_targets_min": 1101 + }, + { + "epoch": 1.7147435897435899, + "grad_norm": 0.5072487823218924, + "learning_rate": 3.326245316481591e-05, + "loss": 0.5595, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2633092999458313, + "step": 535, + "valid_targets_mean": 2397.3, + "valid_targets_min": 1313 + }, + { + "epoch": 1.7307692307692308, + "grad_norm": 0.5195874820272381, + "learning_rate": 3.30941397416599e-05, + "loss": 0.5556, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26943325996398926, + "step": 540, + "valid_targets_mean": 2366.9, + "valid_targets_min": 1427 + }, + { + "epoch": 1.7467948717948718, + "grad_norm": 0.5142077283080907, + "learning_rate": 3.2924187322178865e-05, + "loss": 0.5569, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27412715554237366, + "step": 545, + "valid_targets_mean": 2490.5, + "valid_targets_min": 980 + }, + { + "epoch": 1.7628205128205128, + "grad_norm": 0.47875545252541785, + "learning_rate": 3.275261717935417e-05, + "loss": 0.5694, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30112797021865845, + "step": 550, + "valid_targets_mean": 3342.2, + "valid_targets_min": 1873 + }, + { + "epoch": 1.7788461538461537, + "grad_norm": 0.4754964525049478, + "learning_rate": 3.2579450788657997e-05, + "loss": 0.5531, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3025818467140198, + "step": 555, + "valid_targets_mean": 2941.9, + "valid_targets_min": 1454 + }, + { + "epoch": 1.7948717948717947, + "grad_norm": 0.5071018173857419, + "learning_rate": 3.2404709825365204e-05, + "loss": 0.5759, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25834736227989197, + "step": 560, + "valid_targets_mean": 2617.4, + "valid_targets_min": 1589 + }, + { + "epoch": 1.810897435897436, + "grad_norm": 0.4734235774166826, + "learning_rate": 3.222841616184025e-05, + "loss": 0.5564, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2764870524406433, + "step": 565, + "valid_targets_mean": 2861.0, + "valid_targets_min": 1255 + }, + { + "epoch": 1.8269230769230769, + "grad_norm": 0.44588023212202244, + "learning_rate": 3.2050591864799406e-05, + "loss": 0.5423, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.248623788356781, + "step": 570, + "valid_targets_mean": 2789.8, + "valid_targets_min": 842 + }, + { + "epoch": 1.842948717948718, + "grad_norm": 0.5165230352269572, + "learning_rate": 3.187125919254869e-05, + "loss": 0.5627, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2903827428817749, + "step": 575, + "valid_targets_mean": 2986.1, + "valid_targets_min": 1230 + }, + { + "epoch": 1.858974358974359, + "grad_norm": 0.4542582048942946, + "learning_rate": 3.169044059219778e-05, + "loss": 0.5585, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2745620310306549, + "step": 580, + "valid_targets_mean": 3092.4, + "valid_targets_min": 1156 + }, + { + "epoch": 1.875, + "grad_norm": 0.47805109284658576, + "learning_rate": 3.1508158696850275e-05, + "loss": 0.5582, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2748388648033142, + "step": 585, + "valid_targets_mean": 2798.6, + "valid_targets_min": 1635 + }, + { + "epoch": 1.891025641025641, + "grad_norm": 0.5557249435844636, + "learning_rate": 3.132443632277075e-05, + "loss": 0.5595, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3172360956668854, + "step": 590, + "valid_targets_mean": 2392.4, + "valid_targets_min": 1421 + }, + { + "epoch": 1.907051282051282, + "grad_norm": 0.5342438102011375, + "learning_rate": 3.113929646652879e-05, + "loss": 0.5497, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28413185477256775, + "step": 595, + "valid_targets_mean": 2749.2, + "valid_targets_min": 1624 + }, + { + "epoch": 1.9230769230769231, + "grad_norm": 0.4903784085042197, + "learning_rate": 3.095276230212056e-05, + "loss": 0.5767, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24444545805454254, + "step": 600, + "valid_targets_mean": 2513.0, + "valid_targets_min": 1198 + }, + { + "epoch": 1.939102564102564, + "grad_norm": 0.4919753494428372, + "learning_rate": 3.076485717806808e-05, + "loss": 0.5701, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2963429093360901, + "step": 605, + "valid_targets_mean": 3101.6, + "valid_targets_min": 1700 + }, + { + "epoch": 1.9551282051282053, + "grad_norm": 0.5341843461050885, + "learning_rate": 3.057560461449665e-05, + "loss": 0.5727, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2879267930984497, + "step": 610, + "valid_targets_mean": 2705.4, + "valid_targets_min": 1013 + }, + { + "epoch": 1.9711538461538463, + "grad_norm": 0.4695007929493055, + "learning_rate": 3.038502830019092e-05, + "loss": 0.5563, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27306631207466125, + "step": 615, + "valid_targets_mean": 2854.8, + "valid_targets_min": 1127 + }, + { + "epoch": 1.9871794871794872, + "grad_norm": 0.47019937309397336, + "learning_rate": 3.019315208962968e-05, + "loss": 0.5622, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25543850660324097, + "step": 620, + "valid_targets_mean": 2876.4, + "valid_targets_min": 435 + }, + { + "epoch": 2.003205128205128, + "grad_norm": 0.5385068085538951, + "learning_rate": 3.0000000000000004e-05, + "loss": 0.5497, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3040393888950348, + "step": 625, + "valid_targets_mean": 2746.8, + "valid_targets_min": 1239 + }, + { + "epoch": 2.019230769230769, + "grad_norm": 0.5370993791663188, + "learning_rate": 2.9805596208191056e-05, + "loss": 0.5329, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2809440791606903, + "step": 630, + "valid_targets_mean": 2666.8, + "valid_targets_min": 1709 + }, + { + "epoch": 2.03525641025641, + "grad_norm": 0.5066634114490087, + "learning_rate": 2.960996504776783e-05, + "loss": 0.5337, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2636590600013733, + "step": 635, + "valid_targets_mean": 2582.0, + "valid_targets_min": 1011 + }, + { + "epoch": 2.051282051282051, + "grad_norm": 0.4662311214684164, + "learning_rate": 2.9413131005925296e-05, + "loss": 0.5285, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2907227575778961, + "step": 640, + "valid_targets_mean": 3295.8, + "valid_targets_min": 842 + }, + { + "epoch": 2.0673076923076925, + "grad_norm": 0.5210975495484659, + "learning_rate": 2.9215118720423375e-05, + "loss": 0.5335, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26075279712677, + "step": 645, + "valid_targets_mean": 2613.9, + "valid_targets_min": 1303 + }, + { + "epoch": 2.0833333333333335, + "grad_norm": 0.5089974927966768, + "learning_rate": 2.9015952976502994e-05, + "loss": 0.5387, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2646489143371582, + "step": 650, + "valid_targets_mean": 2547.9, + "valid_targets_min": 1075 + }, + { + "epoch": 2.0993589743589745, + "grad_norm": 0.45259300115190093, + "learning_rate": 2.8815658703783715e-05, + "loss": 0.5369, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29911428689956665, + "step": 655, + "valid_targets_mean": 3205.0, + "valid_targets_min": 1595 + }, + { + "epoch": 2.1153846153846154, + "grad_norm": 0.6431099125108005, + "learning_rate": 2.8614260973143318e-05, + "loss": 0.5274, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2597063183784485, + "step": 660, + "valid_targets_mean": 2840.9, + "valid_targets_min": 547 + }, + { + "epoch": 2.1314102564102564, + "grad_norm": 0.5633027672264547, + "learning_rate": 2.8411784993579633e-05, + "loss": 0.5305, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2584155201911926, + "step": 665, + "valid_targets_mean": 2480.5, + "valid_targets_min": 601 + }, + { + "epoch": 2.1474358974358974, + "grad_norm": 0.5062724588765886, + "learning_rate": 2.820825610905514e-05, + "loss": 0.5279, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2673356533050537, + "step": 670, + "valid_targets_mean": 2614.7, + "valid_targets_min": 1225 + }, + { + "epoch": 2.1634615384615383, + "grad_norm": 0.6211215546739832, + "learning_rate": 2.8003699795324674e-05, + "loss": 0.5296, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.267412394285202, + "step": 675, + "valid_targets_mean": 2339.8, + "valid_targets_min": 1190 + }, + { + "epoch": 2.1794871794871793, + "grad_norm": 0.47260767768104733, + "learning_rate": 2.7798141656746606e-05, + "loss": 0.5383, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26123547554016113, + "step": 680, + "valid_targets_mean": 2792.8, + "valid_targets_min": 1836 + }, + { + "epoch": 2.1955128205128207, + "grad_norm": 0.5167646182153408, + "learning_rate": 2.7591607423077932e-05, + "loss": 0.5229, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2804722487926483, + "step": 685, + "valid_targets_mean": 2891.1, + "valid_targets_min": 1396 + }, + { + "epoch": 2.2115384615384617, + "grad_norm": 0.45992088740325227, + "learning_rate": 2.738412294625369e-05, + "loss": 0.5256, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2086869180202484, + "step": 690, + "valid_targets_mean": 2911.8, + "valid_targets_min": 1076 + }, + { + "epoch": 2.2275641025641026, + "grad_norm": 0.5124490642725669, + "learning_rate": 2.717571419715107e-05, + "loss": 0.523, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25613564252853394, + "step": 695, + "valid_targets_mean": 2460.9, + "valid_targets_min": 1244 + }, + { + "epoch": 2.2435897435897436, + "grad_norm": 0.47586846790245935, + "learning_rate": 2.69664072623386e-05, + "loss": 0.5218, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24780923128128052, + "step": 700, + "valid_targets_mean": 2644.0, + "valid_targets_min": 1265 + }, + { + "epoch": 2.2596153846153846, + "grad_norm": 0.5114331024168999, + "learning_rate": 2.6756228340810946e-05, + "loss": 0.534, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31844857335090637, + "step": 705, + "valid_targets_mean": 3366.5, + "valid_targets_min": 1157 + }, + { + "epoch": 2.2756410256410255, + "grad_norm": 0.5412665077335189, + "learning_rate": 2.6545203740709502e-05, + "loss": 0.5414, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2947283089160919, + "step": 710, + "valid_targets_mean": 2735.8, + "valid_targets_min": 1140 + }, + { + "epoch": 2.2916666666666665, + "grad_norm": 0.48466975498534615, + "learning_rate": 2.6333359876029455e-05, + "loss": 0.5434, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28271210193634033, + "step": 715, + "valid_targets_mean": 2636.4, + "valid_targets_min": 1655 + }, + { + "epoch": 2.3076923076923075, + "grad_norm": 0.5118067123694199, + "learning_rate": 2.612072326331351e-05, + "loss": 0.533, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27647602558135986, + "step": 720, + "valid_targets_mean": 2821.6, + "valid_targets_min": 1448 + }, + { + "epoch": 2.323717948717949, + "grad_norm": 0.49715857445422257, + "learning_rate": 2.5907320518332827e-05, + "loss": 0.5356, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3041874170303345, + "step": 725, + "valid_targets_mean": 3134.6, + "valid_targets_min": 1678 + }, + { + "epoch": 2.33974358974359, + "grad_norm": 0.5298322972875337, + "learning_rate": 2.5693178352755497e-05, + "loss": 0.5266, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25558072328567505, + "step": 730, + "valid_targets_mean": 2448.6, + "valid_targets_min": 573 + }, + { + "epoch": 2.355769230769231, + "grad_norm": 0.5333606639898194, + "learning_rate": 2.547832357080305e-05, + "loss": 0.5215, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27683860063552856, + "step": 735, + "valid_targets_mean": 2680.8, + "valid_targets_min": 1742 + }, + { + "epoch": 2.371794871794872, + "grad_norm": 0.48051460695506026, + "learning_rate": 2.5262783065895377e-05, + "loss": 0.5202, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2652096450328827, + "step": 740, + "valid_targets_mean": 3207.2, + "valid_targets_min": 1816 + }, + { + "epoch": 2.3878205128205128, + "grad_norm": 0.4783808612705691, + "learning_rate": 2.5046583817284437e-05, + "loss": 0.532, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22841182351112366, + "step": 745, + "valid_targets_mean": 2587.9, + "valid_targets_min": 823 + }, + { + "epoch": 2.4038461538461537, + "grad_norm": 0.5391709249007991, + "learning_rate": 2.48297528866773e-05, + "loss": 0.5261, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2468860149383545, + "step": 750, + "valid_targets_mean": 2376.3, + "valid_targets_min": 892 + }, + { + "epoch": 2.4198717948717947, + "grad_norm": 0.5263511548452783, + "learning_rate": 2.4612317414848804e-05, + "loss": 0.5353, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2790845036506653, + "step": 755, + "valid_targets_mean": 2924.8, + "valid_targets_min": 1109 + }, + { + "epoch": 2.435897435897436, + "grad_norm": 0.49865538126497133, + "learning_rate": 2.4394304618244346e-05, + "loss": 0.5187, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2641284465789795, + "step": 760, + "valid_targets_mean": 2560.9, + "valid_targets_min": 1055 + }, + { + "epoch": 2.451923076923077, + "grad_norm": 0.4919863527000147, + "learning_rate": 2.4175741785573177e-05, + "loss": 0.5287, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2962533235549927, + "step": 765, + "valid_targets_mean": 2848.9, + "valid_targets_min": 1604 + }, + { + "epoch": 2.467948717948718, + "grad_norm": 0.5579768978590288, + "learning_rate": 2.39566562743927e-05, + "loss": 0.5379, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2721223831176758, + "step": 770, + "valid_targets_mean": 2622.4, + "valid_targets_min": 1480 + }, + { + "epoch": 2.483974358974359, + "grad_norm": 0.4917371422140305, + "learning_rate": 2.3737075507684103e-05, + "loss": 0.5519, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2823132872581482, + "step": 775, + "valid_targets_mean": 3060.8, + "valid_targets_min": 1002 + }, + { + "epoch": 2.5, + "grad_norm": 0.5183572497009322, + "learning_rate": 2.3517026970419786e-05, + "loss": 0.5285, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.243934765458107, + "step": 780, + "valid_targets_mean": 2466.1, + "valid_targets_min": 1244 + }, + { + "epoch": 2.516025641025641, + "grad_norm": 0.5689569336565411, + "learning_rate": 2.3296538206123134e-05, + "loss": 0.5385, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29714637994766235, + "step": 785, + "valid_targets_mean": 2579.8, + "valid_targets_min": 769 + }, + { + "epoch": 2.532051282051282, + "grad_norm": 0.4852903583594617, + "learning_rate": 2.307563681342081e-05, + "loss": 0.5375, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30132395029067993, + "step": 790, + "valid_targets_mean": 2919.2, + "valid_targets_min": 1185 + }, + { + "epoch": 2.5480769230769234, + "grad_norm": 0.5048000072384919, + "learning_rate": 2.285435044258829e-05, + "loss": 0.5226, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26704657077789307, + "step": 795, + "valid_targets_mean": 3074.5, + "valid_targets_min": 970 + }, + { + "epoch": 2.564102564102564, + "grad_norm": 0.49317222290343393, + "learning_rate": 2.263270679208883e-05, + "loss": 0.5175, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24643100798130035, + "step": 800, + "valid_targets_mean": 2436.0, + "valid_targets_min": 585 + }, + { + "epoch": 2.5801282051282053, + "grad_norm": 0.4737520930741298, + "learning_rate": 2.2410733605106462e-05, + "loss": 0.5415, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26149970293045044, + "step": 805, + "valid_targets_mean": 2902.1, + "valid_targets_min": 1304 + }, + { + "epoch": 2.5961538461538463, + "grad_norm": 0.5121758444175616, + "learning_rate": 2.2188458666073382e-05, + "loss": 0.5318, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2612760066986084, + "step": 810, + "valid_targets_mean": 3133.7, + "valid_targets_min": 1373 + }, + { + "epoch": 2.6121794871794872, + "grad_norm": 0.5045481361963446, + "learning_rate": 2.1965909797192143e-05, + "loss": 0.5313, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25052016973495483, + "step": 815, + "valid_targets_mean": 2793.5, + "valid_targets_min": 1061 + }, + { + "epoch": 2.628205128205128, + "grad_norm": 0.49866393126043274, + "learning_rate": 2.174311485495317e-05, + "loss": 0.5437, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26840710639953613, + "step": 820, + "valid_targets_mean": 2718.1, + "valid_targets_min": 1183 + }, + { + "epoch": 2.644230769230769, + "grad_norm": 0.5328425522492785, + "learning_rate": 2.1520101726647922e-05, + "loss": 0.5389, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2338663637638092, + "step": 825, + "valid_targets_mean": 2070.3, + "valid_targets_min": 924 + }, + { + "epoch": 2.66025641025641, + "grad_norm": 0.4935178661564159, + "learning_rate": 2.1296898326878282e-05, + "loss": 0.5293, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2643035054206848, + "step": 830, + "valid_targets_mean": 2890.3, + "valid_targets_min": 966 + }, + { + "epoch": 2.676282051282051, + "grad_norm": 0.5137500688356019, + "learning_rate": 2.1073532594062432e-05, + "loss": 0.5259, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31183022260665894, + "step": 835, + "valid_targets_mean": 2903.3, + "valid_targets_min": 1411 + }, + { + "epoch": 2.6923076923076925, + "grad_norm": 0.5790430984783882, + "learning_rate": 2.0850032486937838e-05, + "loss": 0.5397, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2541824281215668, + "step": 840, + "valid_targets_mean": 2222.1, + "valid_targets_min": 1056 + }, + { + "epoch": 2.7083333333333335, + "grad_norm": 0.5089728268620257, + "learning_rate": 2.0626425981061608e-05, + "loss": 0.5361, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26173079013824463, + "step": 845, + "valid_targets_mean": 3020.2, + "valid_targets_min": 1193 + }, + { + "epoch": 2.7243589743589745, + "grad_norm": 0.5063766494827193, + "learning_rate": 2.0402741065308808e-05, + "loss": 0.5235, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28143495321273804, + "step": 850, + "valid_targets_mean": 2642.5, + "valid_targets_min": 1303 + }, + { + "epoch": 2.7403846153846154, + "grad_norm": 0.5309100876021802, + "learning_rate": 2.0179005738369098e-05, + "loss": 0.5509, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.267280638217926, + "step": 855, + "valid_targets_mean": 2583.9, + "valid_targets_min": 1500 + }, + { + "epoch": 2.7564102564102564, + "grad_norm": 0.5281817313172733, + "learning_rate": 1.995524800524211e-05, + "loss": 0.5513, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2864803671836853, + "step": 860, + "valid_targets_mean": 2709.2, + "valid_targets_min": 963 + }, + { + "epoch": 2.7724358974358974, + "grad_norm": 0.4852294842051727, + "learning_rate": 1.9731495873732055e-05, + "loss": 0.5443, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24208587408065796, + "step": 865, + "valid_targets_mean": 2688.0, + "valid_targets_min": 1304 + }, + { + "epoch": 2.7884615384615383, + "grad_norm": 0.623697654841149, + "learning_rate": 1.9507777350941996e-05, + "loss": 0.5374, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2755206227302551, + "step": 870, + "valid_targets_mean": 2876.7, + "valid_targets_min": 1219 + }, + { + "epoch": 2.8044871794871797, + "grad_norm": 0.47700443954173843, + "learning_rate": 1.9284120439768192e-05, + "loss": 0.5417, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26930078864097595, + "step": 875, + "valid_targets_mean": 2995.6, + "valid_targets_min": 1731 + }, + { + "epoch": 2.8205128205128203, + "grad_norm": 0.5340023662623198, + "learning_rate": 1.9060553135394957e-05, + "loss": 0.527, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22173257172107697, + "step": 880, + "valid_targets_mean": 2209.4, + "valid_targets_min": 1215 + }, + { + "epoch": 2.8365384615384617, + "grad_norm": 0.49369388993999563, + "learning_rate": 1.8837103421790486e-05, + "loss": 0.5235, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2612769901752472, + "step": 885, + "valid_targets_mean": 2726.4, + "valid_targets_min": 520 + }, + { + "epoch": 2.8525641025641026, + "grad_norm": 0.5014272176624708, + "learning_rate": 1.861379926820414e-05, + "loss": 0.552, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28455230593681335, + "step": 890, + "valid_targets_mean": 2690.3, + "valid_targets_min": 1033 + }, + { + "epoch": 2.8685897435897436, + "grad_norm": 0.5121877748715592, + "learning_rate": 1.8390668625665483e-05, + "loss": 0.5258, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2517583966255188, + "step": 895, + "valid_targets_mean": 2481.9, + "valid_targets_min": 586 + }, + { + "epoch": 2.8846153846153846, + "grad_norm": 0.46919528692963297, + "learning_rate": 1.8167739423485668e-05, + "loss": 0.5261, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25650352239608765, + "step": 900, + "valid_targets_mean": 2941.4, + "valid_targets_min": 1404 + }, + { + "epoch": 2.9006410256410255, + "grad_norm": 0.5019379963438835, + "learning_rate": 1.794503956576152e-05, + "loss": 0.5485, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27527838945388794, + "step": 905, + "valid_targets_mean": 2664.9, + "valid_targets_min": 1325 + }, + { + "epoch": 2.9166666666666665, + "grad_norm": 0.4855003398140857, + "learning_rate": 1.7722596927882758e-05, + "loss": 0.5284, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23271912336349487, + "step": 910, + "valid_targets_mean": 2628.2, + "valid_targets_min": 1311 + }, + { + "epoch": 2.9326923076923075, + "grad_norm": 0.4877223370952336, + "learning_rate": 1.7500439353042834e-05, + "loss": 0.5624, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26730355620384216, + "step": 915, + "valid_targets_mean": 2775.1, + "valid_targets_min": 756 + }, + { + "epoch": 2.948717948717949, + "grad_norm": 0.5401641959987764, + "learning_rate": 1.727859464875381e-05, + "loss": 0.542, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27316930890083313, + "step": 920, + "valid_targets_mean": 2610.7, + "valid_targets_min": 669 + }, + { + "epoch": 2.96474358974359, + "grad_norm": 0.467235720702022, + "learning_rate": 1.7057090583365678e-05, + "loss": 0.5179, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26495471596717834, + "step": 925, + "valid_targets_mean": 3275.5, + "valid_targets_min": 886 + }, + { + "epoch": 2.980769230769231, + "grad_norm": 0.49024323938310016, + "learning_rate": 1.6835954882590567e-05, + "loss": 0.5313, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27755841612815857, + "step": 930, + "valid_targets_mean": 2761.4, + "valid_targets_min": 1173 + }, + { + "epoch": 2.996794871794872, + "grad_norm": 0.48260465618458004, + "learning_rate": 1.6615215226032332e-05, + "loss": 0.5237, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2788291275501251, + "step": 935, + "valid_targets_mean": 3261.9, + "valid_targets_min": 1666 + }, + { + "epoch": 3.0128205128205128, + "grad_norm": 0.4901412147381482, + "learning_rate": 1.6394899243721887e-05, + "loss": 0.5119, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22973302006721497, + "step": 940, + "valid_targets_mean": 3016.4, + "valid_targets_min": 1109 + }, + { + "epoch": 3.0288461538461537, + "grad_norm": 0.5016178718406432, + "learning_rate": 1.6175034512658753e-05, + "loss": 0.522, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2631143629550934, + "step": 945, + "valid_targets_mean": 3169.8, + "valid_targets_min": 1670 + }, + { + "epoch": 3.0448717948717947, + "grad_norm": 0.5207415731021814, + "learning_rate": 1.5955648553359247e-05, + "loss": 0.5081, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2500995099544525, + "step": 950, + "valid_targets_mean": 2769.1, + "valid_targets_min": 1530 + }, + { + "epoch": 3.0608974358974357, + "grad_norm": 0.4952883381542296, + "learning_rate": 1.5736768826411683e-05, + "loss": 0.5318, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25875523686408997, + "step": 955, + "valid_targets_mean": 2645.2, + "valid_targets_min": 1112 + }, + { + "epoch": 3.076923076923077, + "grad_norm": 0.5381528529606109, + "learning_rate": 1.5518422729039188e-05, + "loss": 0.5134, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2681347727775574, + "step": 960, + "valid_targets_mean": 2322.7, + "valid_targets_min": 886 + }, + { + "epoch": 3.092948717948718, + "grad_norm": 0.5019141279039122, + "learning_rate": 1.5300637591670357e-05, + "loss": 0.5126, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23699134588241577, + "step": 965, + "valid_targets_mean": 2605.8, + "valid_targets_min": 1386 + }, + { + "epoch": 3.108974358974359, + "grad_norm": 0.5284324796248476, + "learning_rate": 1.5083440674518302e-05, + "loss": 0.5212, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25518831610679626, + "step": 970, + "valid_targets_mean": 2437.3, + "valid_targets_min": 1194 + }, + { + "epoch": 3.125, + "grad_norm": 0.5545288550997762, + "learning_rate": 1.4866859164168466e-05, + "loss": 0.524, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25043103098869324, + "step": 975, + "valid_targets_mean": 2586.2, + "valid_targets_min": 1323 + }, + { + "epoch": 3.141025641025641, + "grad_norm": 0.5150871184607682, + "learning_rate": 1.4650920170175704e-05, + "loss": 0.5181, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24903708696365356, + "step": 980, + "valid_targets_mean": 2856.2, + "valid_targets_min": 1416 + }, + { + "epoch": 3.157051282051282, + "grad_norm": 0.51903581494101, + "learning_rate": 1.443565072167095e-05, + "loss": 0.5278, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26731231808662415, + "step": 985, + "valid_targets_mean": 2566.1, + "valid_targets_min": 1303 + }, + { + "epoch": 3.173076923076923, + "grad_norm": 0.5466632124851736, + "learning_rate": 1.4221077763977984e-05, + "loss": 0.5165, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24252647161483765, + "step": 990, + "valid_targets_mean": 2442.4, + "valid_targets_min": 659 + }, + { + "epoch": 3.189102564102564, + "grad_norm": 0.5466556370515516, + "learning_rate": 1.4007228155240696e-05, + "loss": 0.5124, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26542800664901733, + "step": 995, + "valid_targets_mean": 2529.8, + "valid_targets_min": 1625 + }, + { + "epoch": 3.2051282051282053, + "grad_norm": 0.5069311466573636, + "learning_rate": 1.37941286630612e-05, + "loss": 0.5114, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2801700830459595, + "step": 1000, + "valid_targets_mean": 2792.9, + "valid_targets_min": 1377 + }, + { + "epoch": 3.2211538461538463, + "grad_norm": 0.6250859912027822, + "learning_rate": 1.3581805961149371e-05, + "loss": 0.508, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25172659754753113, + "step": 1005, + "valid_targets_mean": 2743.6, + "valid_targets_min": 1259 + }, + { + "epoch": 3.2371794871794872, + "grad_norm": 0.5199458882968508, + "learning_rate": 1.3370286625984089e-05, + "loss": 0.5113, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2833164632320404, + "step": 1010, + "valid_targets_mean": 2770.8, + "valid_targets_min": 1114 + }, + { + "epoch": 3.253205128205128, + "grad_norm": 0.4596497760610937, + "learning_rate": 1.3159597133486628e-05, + "loss": 0.503, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23313070833683014, + "step": 1015, + "valid_targets_mean": 2712.9, + "valid_targets_min": 842 + }, + { + "epoch": 3.269230769230769, + "grad_norm": 0.5167847220688774, + "learning_rate": 1.2949763855706678e-05, + "loss": 0.5055, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2619097828865051, + "step": 1020, + "valid_targets_mean": 2816.8, + "valid_targets_min": 1647 + }, + { + "epoch": 3.28525641025641, + "grad_norm": 0.4909239667561551, + "learning_rate": 1.274081305752135e-05, + "loss": 0.516, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2434234470129013, + "step": 1025, + "valid_targets_mean": 2543.8, + "valid_targets_min": 1362 + }, + { + "epoch": 3.301282051282051, + "grad_norm": 0.46584291867859573, + "learning_rate": 1.2532770893347582e-05, + "loss": 0.5095, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23388712108135223, + "step": 1030, + "valid_targets_mean": 2714.8, + "valid_targets_min": 1576 + }, + { + "epoch": 3.3173076923076925, + "grad_norm": 0.502132146117417, + "learning_rate": 1.2325663403868406e-05, + "loss": 0.4977, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2469586580991745, + "step": 1035, + "valid_targets_mean": 2671.7, + "valid_targets_min": 1413 + }, + { + "epoch": 3.3333333333333335, + "grad_norm": 0.5571585854450248, + "learning_rate": 1.2119516512773424e-05, + "loss": 0.5168, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2683925926685333, + "step": 1040, + "valid_targets_mean": 2520.4, + "valid_targets_min": 756 + }, + { + "epoch": 3.3493589743589745, + "grad_norm": 0.5336762575664108, + "learning_rate": 1.1914356023513904e-05, + "loss": 0.5198, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2457447350025177, + "step": 1045, + "valid_targets_mean": 2380.1, + "valid_targets_min": 1242 + }, + { + "epoch": 3.3653846153846154, + "grad_norm": 0.5667115296323573, + "learning_rate": 1.1710207616073001e-05, + "loss": 0.514, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28504765033721924, + "step": 1050, + "valid_targets_mean": 2624.0, + "valid_targets_min": 567 + }, + { + "epoch": 3.3814102564102564, + "grad_norm": 0.5099600194173104, + "learning_rate": 1.1507096843751372e-05, + "loss": 0.5127, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2691630721092224, + "step": 1055, + "valid_targets_mean": 2777.8, + "valid_targets_min": 683 + }, + { + "epoch": 3.3974358974358974, + "grad_norm": 0.533137614413181, + "learning_rate": 1.1305049129968637e-05, + "loss": 0.5129, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21904993057250977, + "step": 1060, + "valid_targets_mean": 2296.4, + "valid_targets_min": 657 + }, + { + "epoch": 3.4134615384615383, + "grad_norm": 0.5146746955617865, + "learning_rate": 1.110408976508118e-05, + "loss": 0.5059, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2554003596305847, + "step": 1065, + "valid_targets_mean": 2615.2, + "valid_targets_min": 1338 + }, + { + "epoch": 3.4294871794871793, + "grad_norm": 0.569499082119217, + "learning_rate": 1.090424390321648e-05, + "loss": 0.5266, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2377270758152008, + "step": 1070, + "valid_targets_mean": 2141.8, + "valid_targets_min": 651 + }, + { + "epoch": 3.4455128205128207, + "grad_norm": 0.4835581841975967, + "learning_rate": 1.070553655912463e-05, + "loss": 0.5122, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2490512728691101, + "step": 1075, + "valid_targets_mean": 2829.5, + "valid_targets_min": 1276 + }, + { + "epoch": 3.4615384615384617, + "grad_norm": 0.5182915999209446, + "learning_rate": 1.0507992605047193e-05, + "loss": 0.5121, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28138262033462524, + "step": 1080, + "valid_targets_mean": 3092.6, + "valid_targets_min": 1552 + }, + { + "epoch": 3.4775641025641026, + "grad_norm": 0.5102091196096896, + "learning_rate": 1.0311636767603952e-05, + "loss": 0.5219, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24811017513275146, + "step": 1085, + "valid_targets_mean": 2473.1, + "valid_targets_min": 1263 + }, + { + "epoch": 3.4935897435897436, + "grad_norm": 0.5505424626120781, + "learning_rate": 1.0116493624697862e-05, + "loss": 0.5251, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25646162033081055, + "step": 1090, + "valid_targets_mean": 2409.8, + "valid_targets_min": 1381 + }, + { + "epoch": 3.5096153846153846, + "grad_norm": 0.5453130427654166, + "learning_rate": 9.922587602438657e-06, + "loss": 0.5186, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.267898827791214, + "step": 1095, + "valid_targets_mean": 2698.6, + "valid_targets_min": 751 + }, + { + "epoch": 3.5256410256410255, + "grad_norm": 0.5011760790811203, + "learning_rate": 9.729942972085401e-06, + "loss": 0.5121, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2532099187374115, + "step": 1100, + "valid_targets_mean": 2729.1, + "valid_targets_min": 1471 + }, + { + "epoch": 3.5416666666666665, + "grad_norm": 0.5063405039437282, + "learning_rate": 9.538583847008452e-06, + "loss": 0.5183, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2585946321487427, + "step": 1105, + "valid_targets_mean": 2670.7, + "valid_targets_min": 1420 + }, + { + "epoch": 3.5576923076923075, + "grad_norm": 0.4878211464301314, + "learning_rate": 9.348534179671202e-06, + "loss": 0.4917, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25911808013916016, + "step": 1110, + "valid_targets_mean": 3231.8, + "valid_targets_min": 1715 + }, + { + "epoch": 3.573717948717949, + "grad_norm": 0.48958046086249296, + "learning_rate": 9.159817758631923e-06, + "loss": 0.5104, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26951271295547485, + "step": 1115, + "valid_targets_mean": 2618.6, + "valid_targets_min": 1116 + }, + { + "epoch": 3.58974358974359, + "grad_norm": 0.4839908494296597, + "learning_rate": 8.972458205566168e-06, + "loss": 0.5116, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25361812114715576, + "step": 1120, + "valid_targets_mean": 2789.6, + "valid_targets_min": 1187 + }, + { + "epoch": 3.605769230769231, + "grad_norm": 0.5535168080897457, + "learning_rate": 8.786478972310023e-06, + "loss": 0.5066, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26389190554618835, + "step": 1125, + "valid_targets_mean": 2648.0, + "valid_targets_min": 1379 + }, + { + "epoch": 3.621794871794872, + "grad_norm": 0.5080355494863202, + "learning_rate": 8.601903337924646e-06, + "loss": 0.5118, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2588704526424408, + "step": 1130, + "valid_targets_mean": 2575.2, + "valid_targets_min": 1334 + }, + { + "epoch": 3.6378205128205128, + "grad_norm": 0.49125868635329756, + "learning_rate": 8.418754405782423e-06, + "loss": 0.5154, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24010632932186127, + "step": 1135, + "valid_targets_mean": 3124.4, + "valid_targets_min": 1320 + }, + { + "epoch": 3.6538461538461537, + "grad_norm": 0.522430677893815, + "learning_rate": 8.237055100675092e-06, + "loss": 0.5173, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2740614414215088, + "step": 1140, + "valid_targets_mean": 2463.3, + "valid_targets_min": 956 + }, + { + "epoch": 3.6698717948717947, + "grad_norm": 0.5095945655848667, + "learning_rate": 8.056828165944282e-06, + "loss": 0.5138, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24778661131858826, + "step": 1145, + "valid_targets_mean": 2634.4, + "valid_targets_min": 1133 + }, + { + "epoch": 3.685897435897436, + "grad_norm": 0.5205507571372584, + "learning_rate": 7.878096160634675e-06, + "loss": 0.5065, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2700071632862091, + "step": 1150, + "valid_targets_mean": 2605.7, + "valid_targets_min": 961 + }, + { + "epoch": 3.7019230769230766, + "grad_norm": 0.5022199361254235, + "learning_rate": 7.700881456670342e-06, + "loss": 0.5224, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25807714462280273, + "step": 1155, + "valid_targets_mean": 2745.6, + "valid_targets_min": 1467 + }, + { + "epoch": 3.717948717948718, + "grad_norm": 0.5204165070548322, + "learning_rate": 7.525206236054385e-06, + "loss": 0.5144, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26541176438331604, + "step": 1160, + "valid_targets_mean": 2881.2, + "valid_targets_min": 1531 + }, + { + "epoch": 3.733974358974359, + "grad_norm": 0.4847397653951549, + "learning_rate": 7.3510924880924575e-06, + "loss": 0.5147, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2076713740825653, + "step": 1165, + "valid_targets_mean": 2522.8, + "valid_targets_min": 1030 + }, + { + "epoch": 3.75, + "grad_norm": 0.49758404101693104, + "learning_rate": 7.178562006640337e-06, + "loss": 0.5296, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2572706639766693, + "step": 1170, + "valid_targets_mean": 2866.5, + "valid_targets_min": 776 + }, + { + "epoch": 3.766025641025641, + "grad_norm": 0.4865801190784059, + "learning_rate": 7.0076363873759865e-06, + "loss": 0.5041, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2694791853427887, + "step": 1175, + "valid_targets_mean": 3112.3, + "valid_targets_min": 1177 + }, + { + "epoch": 3.782051282051282, + "grad_norm": 0.4289474023794107, + "learning_rate": 6.838337025096424e-06, + "loss": 0.4976, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24875801801681519, + "step": 1180, + "valid_targets_mean": 3378.5, + "valid_targets_min": 2046 + }, + { + "epoch": 3.7980769230769234, + "grad_norm": 0.49994513336481994, + "learning_rate": 6.67068511103971e-06, + "loss": 0.5071, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25759273767471313, + "step": 1185, + "valid_targets_mean": 2810.8, + "valid_targets_min": 1317 + }, + { + "epoch": 3.814102564102564, + "grad_norm": 0.5145087645637461, + "learning_rate": 6.504701630232475e-06, + "loss": 0.5001, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24269188940525055, + "step": 1190, + "valid_targets_mean": 2315.5, + "valid_targets_min": 873 + }, + { + "epoch": 3.8301282051282053, + "grad_norm": 0.5053964635592612, + "learning_rate": 6.340407358863167e-06, + "loss": 0.5106, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27107882499694824, + "step": 1195, + "valid_targets_mean": 2939.9, + "valid_targets_min": 1553 + }, + { + "epoch": 3.8461538461538463, + "grad_norm": 0.48788700328753865, + "learning_rate": 6.177822861681557e-06, + "loss": 0.512, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24716894328594208, + "step": 1200, + "valid_targets_mean": 2915.2, + "valid_targets_min": 1249 + }, + { + "epoch": 3.8621794871794872, + "grad_norm": 0.5375256238281615, + "learning_rate": 6.016968489424572e-06, + "loss": 0.5223, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2848437428474426, + "step": 1205, + "valid_targets_mean": 2627.2, + "valid_targets_min": 1600 + }, + { + "epoch": 3.878205128205128, + "grad_norm": 0.44335134280100075, + "learning_rate": 5.857864376269051e-06, + "loss": 0.5028, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24449265003204346, + "step": 1210, + "valid_targets_mean": 3266.6, + "valid_targets_min": 1885 + }, + { + "epoch": 3.894230769230769, + "grad_norm": 0.4877168445363961, + "learning_rate": 5.700530437311509e-06, + "loss": 0.5187, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2589573860168457, + "step": 1215, + "valid_targets_mean": 2644.8, + "valid_targets_min": 707 + }, + { + "epoch": 3.91025641025641, + "grad_norm": 0.5115202360691214, + "learning_rate": 5.544986366075371e-06, + "loss": 0.5202, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2709648311138153, + "step": 1220, + "valid_targets_mean": 3078.7, + "valid_targets_min": 1332 + }, + { + "epoch": 3.926282051282051, + "grad_norm": 0.46525331541379866, + "learning_rate": 5.39125163204594e-06, + "loss": 0.503, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2524239420890808, + "step": 1225, + "valid_targets_mean": 3194.6, + "valid_targets_min": 1622 + }, + { + "epoch": 3.9423076923076925, + "grad_norm": 0.5134129988969063, + "learning_rate": 5.239345478233364e-06, + "loss": 0.5013, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27747344970703125, + "step": 1230, + "valid_targets_mean": 2960.1, + "valid_targets_min": 1354 + }, + { + "epoch": 3.9583333333333335, + "grad_norm": 0.490211374373267, + "learning_rate": 5.089286918764031e-06, + "loss": 0.5082, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24609319865703583, + "step": 1235, + "valid_targets_mean": 3089.4, + "valid_targets_min": 2019 + }, + { + "epoch": 3.9743589743589745, + "grad_norm": 0.468208756136862, + "learning_rate": 4.941094736500522e-06, + "loss": 0.4994, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23819640278816223, + "step": 1240, + "valid_targets_mean": 2995.7, + "valid_targets_min": 1084 + }, + { + "epoch": 3.9903846153846154, + "grad_norm": 0.5090017684601932, + "learning_rate": 4.794787480690597e-06, + "loss": 0.5204, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2769983112812042, + "step": 1245, + "valid_targets_mean": 2845.6, + "valid_targets_min": 918 + }, + { + "epoch": 4.006410256410256, + "grad_norm": 0.47177334327622505, + "learning_rate": 4.650383464645338e-06, + "loss": 0.5117, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2771986126899719, + "step": 1250, + "valid_targets_mean": 3190.9, + "valid_targets_min": 1686 + }, + { + "epoch": 4.022435897435898, + "grad_norm": 0.5007827975128015, + "learning_rate": 4.507900763446911e-06, + "loss": 0.5032, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2388410121202469, + "step": 1255, + "valid_targets_mean": 2622.6, + "valid_targets_min": 1570 + }, + { + "epoch": 4.038461538461538, + "grad_norm": 0.48832340426080595, + "learning_rate": 4.367357211686072e-06, + "loss": 0.5037, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24476832151412964, + "step": 1260, + "valid_targets_mean": 2671.0, + "valid_targets_min": 625 + }, + { + "epoch": 4.05448717948718, + "grad_norm": 0.4933650082162676, + "learning_rate": 4.228770401229824e-06, + "loss": 0.5025, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2521268129348755, + "step": 1265, + "valid_targets_mean": 2734.1, + "valid_targets_min": 723 + }, + { + "epoch": 4.07051282051282, + "grad_norm": 0.5079581171636282, + "learning_rate": 4.092157679019442e-06, + "loss": 0.4991, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26366621255874634, + "step": 1270, + "valid_targets_mean": 3056.3, + "valid_targets_min": 1040 + }, + { + "epoch": 4.086538461538462, + "grad_norm": 0.49596862412646814, + "learning_rate": 3.957536144899123e-06, + "loss": 0.513, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24502941966056824, + "step": 1275, + "valid_targets_mean": 2438.4, + "valid_targets_min": 1514 + }, + { + "epoch": 4.102564102564102, + "grad_norm": 0.5145129088905596, + "learning_rate": 3.8249226494756445e-06, + "loss": 0.4966, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25848090648651123, + "step": 1280, + "valid_targets_mean": 2680.9, + "valid_targets_min": 1795 + }, + { + "epoch": 4.118589743589744, + "grad_norm": 0.5037020159257446, + "learning_rate": 3.694333792009115e-06, + "loss": 0.5052, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22823549807071686, + "step": 1285, + "valid_targets_mean": 2669.3, + "valid_targets_min": 1027 + }, + { + "epoch": 4.134615384615385, + "grad_norm": 0.5060562285691546, + "learning_rate": 3.565785918335292e-06, + "loss": 0.5004, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24490630626678467, + "step": 1290, + "valid_targets_mean": 2757.9, + "valid_targets_min": 1501 + }, + { + "epoch": 4.1506410256410255, + "grad_norm": 0.4710171907923074, + "learning_rate": 3.43929511881953e-06, + "loss": 0.4886, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2338603436946869, + "step": 1295, + "valid_targets_mean": 2913.7, + "valid_targets_min": 751 + }, + { + "epoch": 4.166666666666667, + "grad_norm": 0.555995128304345, + "learning_rate": 3.3148772263427743e-06, + "loss": 0.5163, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.275356650352478, + "step": 1300, + "valid_targets_mean": 2518.2, + "valid_targets_min": 1317 + }, + { + "epoch": 4.1826923076923075, + "grad_norm": 0.759342398332849, + "learning_rate": 3.1925478143197418e-06, + "loss": 0.5048, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25380346179008484, + "step": 1305, + "valid_targets_mean": 2555.2, + "valid_targets_min": 1653 + }, + { + "epoch": 4.198717948717949, + "grad_norm": 0.5072591530204905, + "learning_rate": 3.0723221947495907e-06, + "loss": 0.5003, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21137681603431702, + "step": 1310, + "valid_targets_mean": 2328.9, + "valid_targets_min": 947 + }, + { + "epoch": 4.214743589743589, + "grad_norm": 0.5180541192988526, + "learning_rate": 2.954215416299331e-06, + "loss": 0.5043, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24892708659172058, + "step": 1315, + "valid_targets_mean": 2616.9, + "valid_targets_min": 1347 + }, + { + "epoch": 4.230769230769231, + "grad_norm": 0.5132267950452266, + "learning_rate": 2.838242262420148e-06, + "loss": 0.5036, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25727900862693787, + "step": 1320, + "valid_targets_mean": 2797.8, + "valid_targets_min": 1104 + }, + { + "epoch": 4.246794871794872, + "grad_norm": 0.5184018388606338, + "learning_rate": 2.7244172494969978e-06, + "loss": 0.509, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24256864190101624, + "step": 1325, + "valid_targets_mean": 2934.1, + "valid_targets_min": 1565 + }, + { + "epoch": 4.262820512820513, + "grad_norm": 0.5252899074313376, + "learning_rate": 2.6127546250315438e-06, + "loss": 0.5098, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24076877534389496, + "step": 1330, + "valid_targets_mean": 2428.9, + "valid_targets_min": 1041 + }, + { + "epoch": 4.278846153846154, + "grad_norm": 0.5108531790336086, + "learning_rate": 2.503268365858831e-06, + "loss": 0.5041, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2731473445892334, + "step": 1335, + "valid_targets_mean": 2923.9, + "valid_targets_min": 1600 + }, + { + "epoch": 4.294871794871795, + "grad_norm": 0.5321025763349825, + "learning_rate": 2.3959721763977805e-06, + "loss": 0.5123, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2777535617351532, + "step": 1340, + "valid_targets_mean": 2700.3, + "valid_targets_min": 1464 + }, + { + "epoch": 4.310897435897436, + "grad_norm": 0.6033522758194312, + "learning_rate": 2.2908794869358044e-06, + "loss": 0.4999, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2676275968551636, + "step": 1345, + "valid_targets_mean": 2569.8, + "valid_targets_min": 1370 + }, + { + "epoch": 4.326923076923077, + "grad_norm": 0.5333361636057034, + "learning_rate": 2.188003451947747e-06, + "loss": 0.5061, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2521325349807739, + "step": 1350, + "valid_targets_mean": 2529.4, + "valid_targets_min": 1258 + }, + { + "epoch": 4.342948717948718, + "grad_norm": 0.5123708912901537, + "learning_rate": 2.0873569484493305e-06, + "loss": 0.4987, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2577575743198395, + "step": 1355, + "valid_targets_mean": 2865.8, + "valid_targets_min": 1345 + }, + { + "epoch": 4.358974358974359, + "grad_norm": 0.5054990888582566, + "learning_rate": 1.9889525743853323e-06, + "loss": 0.4986, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2553538978099823, + "step": 1360, + "valid_targets_mean": 2472.9, + "valid_targets_min": 1112 + }, + { + "epoch": 4.375, + "grad_norm": 0.5678085203826202, + "learning_rate": 1.8928026470526917e-06, + "loss": 0.5114, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2596340775489807, + "step": 1365, + "valid_targets_mean": 2628.6, + "valid_targets_min": 1276 + }, + { + "epoch": 4.391025641025641, + "grad_norm": 0.4893241712584916, + "learning_rate": 1.7989192015587776e-06, + "loss": 0.4961, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23696500062942505, + "step": 1370, + "valid_targets_mean": 2611.6, + "valid_targets_min": 1075 + }, + { + "epoch": 4.407051282051282, + "grad_norm": 0.5091429153868882, + "learning_rate": 1.7073139893149092e-06, + "loss": 0.4983, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2712137699127197, + "step": 1375, + "valid_targets_mean": 2809.8, + "valid_targets_min": 1100 + }, + { + "epoch": 4.423076923076923, + "grad_norm": 0.572290095970069, + "learning_rate": 1.6179984765654743e-06, + "loss": 0.4996, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23557832837104797, + "step": 1380, + "valid_targets_mean": 2475.9, + "valid_targets_min": 1168 + }, + { + "epoch": 4.439102564102564, + "grad_norm": 0.5831706000170874, + "learning_rate": 1.5309838429526714e-06, + "loss": 0.5145, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23172461986541748, + "step": 1385, + "valid_targets_mean": 2455.8, + "valid_targets_min": 1304 + }, + { + "epoch": 4.455128205128205, + "grad_norm": 0.5140434264325106, + "learning_rate": 1.4462809801171428e-06, + "loss": 0.5168, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26323020458221436, + "step": 1390, + "valid_targets_mean": 2694.8, + "valid_targets_min": 1232 + }, + { + "epoch": 4.471153846153846, + "grad_norm": 0.5019615556872697, + "learning_rate": 1.3639004903346954e-06, + "loss": 0.5077, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24197980761528015, + "step": 1395, + "valid_targets_mean": 2674.9, + "valid_targets_min": 1334 + }, + { + "epoch": 4.487179487179487, + "grad_norm": 0.5101131304847485, + "learning_rate": 1.2838526851891864e-06, + "loss": 0.5002, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25778818130493164, + "step": 1400, + "valid_targets_mean": 2975.2, + "valid_targets_min": 1695 + }, + { + "epoch": 4.503205128205128, + "grad_norm": 0.494280031757431, + "learning_rate": 1.2061475842818337e-06, + "loss": 0.5081, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28242138028144836, + "step": 1405, + "valid_targets_mean": 3291.9, + "valid_targets_min": 1955 + }, + { + "epoch": 4.519230769230769, + "grad_norm": 0.4859471337024964, + "learning_rate": 1.1307949139770446e-06, + "loss": 0.4874, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2636260688304901, + "step": 1410, + "valid_targets_mean": 3025.3, + "valid_targets_min": 1367 + }, + { + "epoch": 4.535256410256411, + "grad_norm": 0.5115875440836855, + "learning_rate": 1.057804106184992e-06, + "loss": 0.5068, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23679447174072266, + "step": 1415, + "valid_targets_mean": 2550.2, + "valid_targets_min": 1379 + }, + { + "epoch": 4.551282051282051, + "grad_norm": 0.49992052994826364, + "learning_rate": 9.871842971809853e-07, + "loss": 0.509, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28533414006233215, + "step": 1420, + "valid_targets_mean": 2607.3, + "valid_targets_min": 1195 + }, + { + "epoch": 4.5673076923076925, + "grad_norm": 0.4701214147504648, + "learning_rate": 9.189443264619102e-07, + "loss": 0.5015, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25728708505630493, + "step": 1425, + "valid_targets_mean": 3328.6, + "valid_targets_min": 1109 + }, + { + "epoch": 4.583333333333333, + "grad_norm": 0.5676405967354015, + "learning_rate": 8.530927356397778e-07, + "loss": 0.5017, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25928235054016113, + "step": 1430, + "valid_targets_mean": 2991.0, + "valid_targets_min": 1104 + }, + { + "epoch": 4.5993589743589745, + "grad_norm": 0.5627440521581173, + "learning_rate": 7.896377673725553e-07, + "loss": 0.4988, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22346052527427673, + "step": 1435, + "valid_targets_mean": 2614.4, + "valid_targets_min": 891 + }, + { + "epoch": 4.615384615384615, + "grad_norm": 0.504256223200527, + "learning_rate": 7.285873643324514e-07, + "loss": 0.5065, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2539728283882141, + "step": 1440, + "valid_targets_mean": 2844.8, + "valid_targets_min": 1455 + }, + { + "epoch": 4.631410256410256, + "grad_norm": 0.4925495661414361, + "learning_rate": 6.69949168211721e-07, + "loss": 0.4925, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22008481621742249, + "step": 1445, + "valid_targets_mean": 2975.1, + "valid_targets_min": 1312 + }, + { + "epoch": 4.647435897435898, + "grad_norm": 0.5494300003962139, + "learning_rate": 6.137305187661513e-07, + "loss": 0.5076, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2878986895084381, + "step": 1450, + "valid_targets_mean": 2772.4, + "valid_targets_min": 826 + }, + { + "epoch": 4.663461538461538, + "grad_norm": 0.5039636136792486, + "learning_rate": 5.599384528963425e-07, + "loss": 0.5015, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26579830050468445, + "step": 1455, + "valid_targets_mean": 2956.1, + "valid_targets_min": 1082 + }, + { + "epoch": 4.67948717948718, + "grad_norm": 0.513557568224406, + "learning_rate": 5.085797037669072e-07, + "loss": 0.5119, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26804548501968384, + "step": 1460, + "valid_targets_mean": 2945.6, + "valid_targets_min": 1373 + }, + { + "epoch": 4.69551282051282, + "grad_norm": 0.48755150331371994, + "learning_rate": 4.5966069996365993e-07, + "loss": 0.4999, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25493526458740234, + "step": 1465, + "valid_targets_mean": 2965.0, + "valid_targets_min": 989 + }, + { + "epoch": 4.711538461538462, + "grad_norm": 0.48996510478060334, + "learning_rate": 4.1318756468897047e-07, + "loss": 0.4846, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26829373836517334, + "step": 1470, + "valid_targets_mean": 3138.4, + "valid_targets_min": 1490 + }, + { + "epoch": 4.727564102564102, + "grad_norm": 0.540400619341247, + "learning_rate": 3.691661149953096e-07, + "loss": 0.5104, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25986146926879883, + "step": 1475, + "valid_targets_mean": 2386.2, + "valid_targets_min": 920 + }, + { + "epoch": 4.743589743589744, + "grad_norm": 0.5070305685741632, + "learning_rate": 3.2760186105712964e-07, + "loss": 0.5052, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23506109416484833, + "step": 1480, + "valid_targets_mean": 2560.3, + "valid_targets_min": 1303 + }, + { + "epoch": 4.759615384615385, + "grad_norm": 0.5067194990562207, + "learning_rate": 2.8850000548115155e-07, + "loss": 0.499, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23781508207321167, + "step": 1485, + "valid_targets_mean": 2701.9, + "valid_targets_min": 1124 + }, + { + "epoch": 4.7756410256410255, + "grad_norm": 0.5156995328819387, + "learning_rate": 2.518654426551592e-07, + "loss": 0.496, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27351832389831543, + "step": 1490, + "valid_targets_mean": 2925.2, + "valid_targets_min": 1015 + }, + { + "epoch": 4.791666666666667, + "grad_norm": 0.5599487612757731, + "learning_rate": 2.1770275813536746e-07, + "loss": 0.497, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25763094425201416, + "step": 1495, + "valid_targets_mean": 2683.4, + "valid_targets_min": 1528 + }, + { + "epoch": 4.8076923076923075, + "grad_norm": 0.5385472477141456, + "learning_rate": 1.8601622807244312e-07, + "loss": 0.5143, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2503165900707245, + "step": 1500, + "valid_targets_mean": 2487.6, + "valid_targets_min": 1024 + }, + { + "epoch": 4.823717948717949, + "grad_norm": 0.4726991070974343, + "learning_rate": 1.5680981867625566e-07, + "loss": 0.5138, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24221260845661163, + "step": 1505, + "valid_targets_mean": 2659.8, + "valid_targets_min": 1354 + }, + { + "epoch": 4.839743589743589, + "grad_norm": 0.5461272895850571, + "learning_rate": 1.3008718571943636e-07, + "loss": 0.5003, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2652673125267029, + "step": 1510, + "valid_targets_mean": 2697.6, + "valid_targets_min": 1328 + }, + { + "epoch": 4.855769230769231, + "grad_norm": 0.5253200725248618, + "learning_rate": 1.058516740797777e-07, + "loss": 0.4958, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25738656520843506, + "step": 1515, + "valid_targets_mean": 3106.8, + "valid_targets_min": 1145 + }, + { + "epoch": 4.871794871794872, + "grad_norm": 0.47676822865399054, + "learning_rate": 8.410631732155062e-08, + "loss": 0.5006, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26309454441070557, + "step": 1520, + "valid_targets_mean": 3213.8, + "valid_targets_min": 1801 + }, + { + "epoch": 4.887820512820513, + "grad_norm": 0.49678497951381617, + "learning_rate": 6.485383731580142e-08, + "loss": 0.4982, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.258725643157959, + "step": 1525, + "valid_targets_mean": 2639.6, + "valid_targets_min": 1130 + }, + { + "epoch": 4.903846153846154, + "grad_norm": 0.5295930413362215, + "learning_rate": 4.809664389964441e-08, + "loss": 0.5071, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24629537761211395, + "step": 1530, + "valid_targets_mean": 2534.1, + "valid_targets_min": 1371 + }, + { + "epoch": 4.919871794871795, + "grad_norm": 0.5092910461235371, + "learning_rate": 3.383683457463649e-08, + "loss": 0.4957, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28113991022109985, + "step": 1535, + "valid_targets_mean": 2743.7, + "valid_targets_min": 855 + }, + { + "epoch": 4.935897435897436, + "grad_norm": 0.5052772865334241, + "learning_rate": 2.207619424421381e-08, + "loss": 0.507, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22167710959911346, + "step": 1540, + "valid_targets_mean": 2326.5, + "valid_targets_min": 828 + }, + { + "epoch": 4.951923076923077, + "grad_norm": 0.5286114149755486, + "learning_rate": 1.281619499029274e-08, + "loss": 0.5125, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25723177194595337, + "step": 1545, + "valid_targets_mean": 2850.4, + "valid_targets_min": 1711 + }, + { + "epoch": 4.967948717948718, + "grad_norm": 0.5176168142039531, + "learning_rate": 6.057995888997248e-09, + "loss": 0.4946, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2718939483165741, + "step": 1550, + "valid_targets_mean": 2905.1, + "valid_targets_min": 1289 + }, + { + "epoch": 4.983974358974359, + "grad_norm": 0.5326959721897664, + "learning_rate": 1.8024428655794012e-09, + "loss": 0.5016, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3013129234313965, + "step": 1555, + "valid_targets_mean": 3025.9, + "valid_targets_min": 1624 + }, + { + "epoch": 5.0, + "grad_norm": 0.5167254136149978, + "learning_rate": 5.00685885418406e-11, + "loss": 0.4965, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21225425601005554, + "step": 1560, + "valid_targets_mean": 2312.1, + "valid_targets_min": 769 + }, + { + "epoch": 5.0, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21225425601005554, + "step": 1560, + "total_flos": 6.613684893274604e+17, + "train_loss": 0.5546745132177304, + "train_runtime": 11827.5918, + "train_samples_per_second": 4.22, + "train_steps_per_second": 0.132, + "valid_targets_mean": 2312.1, + "valid_targets_min": 769 + } + ], + "logging_steps": 5, + "max_steps": 1560, + "num_input_tokens_seen": 0, + "num_train_epochs": 5, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 6.613684893274604e+17, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}