| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.997593524392912, | |
| "eval_steps": 500, | |
| "global_step": 3426, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0, | |
| "eval_loss": 2.1641297340393066, | |
| "eval_runtime": 0.8244, | |
| "eval_samples_per_second": 24.261, | |
| "eval_steps_per_second": 2.426, | |
| "step": 0 | |
| }, | |
| { | |
| "epoch": 0.0008750820389411508, | |
| "grad_norm": 6.372805118560791, | |
| "learning_rate": 2.915451895043732e-08, | |
| "loss": 1.6434, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008750820389411508, | |
| "grad_norm": 6.278440475463867, | |
| "learning_rate": 2.915451895043732e-07, | |
| "loss": 1.7188, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.017501640778823015, | |
| "grad_norm": 6.023682594299316, | |
| "learning_rate": 5.830903790087464e-07, | |
| "loss": 1.7154, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02625246116823452, | |
| "grad_norm": 3.7778327465057373, | |
| "learning_rate": 8.746355685131196e-07, | |
| "loss": 1.6563, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03500328155764603, | |
| "grad_norm": 3.8672306537628174, | |
| "learning_rate": 1.1661807580174927e-06, | |
| "loss": 1.5743, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04375410194705753, | |
| "grad_norm": 2.53611159324646, | |
| "learning_rate": 1.4577259475218661e-06, | |
| "loss": 1.479, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05250492233646904, | |
| "grad_norm": 2.019819974899292, | |
| "learning_rate": 1.7492711370262391e-06, | |
| "loss": 1.4242, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.06125574272588055, | |
| "grad_norm": 1.7553459405899048, | |
| "learning_rate": 2.0408163265306125e-06, | |
| "loss": 1.3795, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.07000656311529206, | |
| "grad_norm": 1.4397658109664917, | |
| "learning_rate": 2.3323615160349855e-06, | |
| "loss": 1.3745, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.07875738350470357, | |
| "grad_norm": 1.5793251991271973, | |
| "learning_rate": 2.6239067055393585e-06, | |
| "loss": 1.3529, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.08750820389411507, | |
| "grad_norm": 1.5069832801818848, | |
| "learning_rate": 2.9154518950437323e-06, | |
| "loss": 1.3459, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.09625902428352658, | |
| "grad_norm": 1.4921239614486694, | |
| "learning_rate": 3.2069970845481052e-06, | |
| "loss": 1.2743, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.10500984467293809, | |
| "grad_norm": 1.4761064052581787, | |
| "learning_rate": 3.4985422740524782e-06, | |
| "loss": 1.2959, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.1137606650623496, | |
| "grad_norm": 1.4724879264831543, | |
| "learning_rate": 3.790087463556852e-06, | |
| "loss": 1.2773, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.1225114854517611, | |
| "grad_norm": 1.4954681396484375, | |
| "learning_rate": 4.081632653061225e-06, | |
| "loss": 1.2695, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.1312623058411726, | |
| "grad_norm": 1.4220881462097168, | |
| "learning_rate": 4.3731778425655976e-06, | |
| "loss": 1.2668, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.14001312623058412, | |
| "grad_norm": 1.4039027690887451, | |
| "learning_rate": 4.664723032069971e-06, | |
| "loss": 1.2231, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.14876394661999562, | |
| "grad_norm": 1.468996524810791, | |
| "learning_rate": 4.956268221574344e-06, | |
| "loss": 1.221, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.15751476700940714, | |
| "grad_norm": 1.5780359506607056, | |
| "learning_rate": 5.247813411078717e-06, | |
| "loss": 1.2403, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.16626558739881864, | |
| "grad_norm": 1.4090375900268555, | |
| "learning_rate": 5.539358600583091e-06, | |
| "loss": 1.2021, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.17501640778823013, | |
| "grad_norm": 1.489652395248413, | |
| "learning_rate": 5.8309037900874645e-06, | |
| "loss": 1.1731, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.18376722817764166, | |
| "grad_norm": 1.3121322393417358, | |
| "learning_rate": 6.122448979591837e-06, | |
| "loss": 1.1924, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.19251804856705315, | |
| "grad_norm": 1.3954280614852905, | |
| "learning_rate": 6.4139941690962105e-06, | |
| "loss": 1.2121, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.20126886895646467, | |
| "grad_norm": 1.4191243648529053, | |
| "learning_rate": 6.705539358600584e-06, | |
| "loss": 1.1973, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.21001968934587617, | |
| "grad_norm": 1.4373400211334229, | |
| "learning_rate": 6.9970845481049564e-06, | |
| "loss": 1.1701, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.2187705097352877, | |
| "grad_norm": 1.4669713973999023, | |
| "learning_rate": 7.28862973760933e-06, | |
| "loss": 1.1923, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.2275213301246992, | |
| "grad_norm": 1.4269640445709229, | |
| "learning_rate": 7.580174927113704e-06, | |
| "loss": 1.1713, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.23627215051411068, | |
| "grad_norm": 1.7535545825958252, | |
| "learning_rate": 7.871720116618077e-06, | |
| "loss": 1.1891, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.2450229709035222, | |
| "grad_norm": 1.599094271659851, | |
| "learning_rate": 8.16326530612245e-06, | |
| "loss": 1.1827, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.2537737912929337, | |
| "grad_norm": 1.8278971910476685, | |
| "learning_rate": 8.454810495626823e-06, | |
| "loss": 1.1402, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.2625246116823452, | |
| "grad_norm": 1.5127216577529907, | |
| "learning_rate": 8.746355685131195e-06, | |
| "loss": 1.1597, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.27127543207175675, | |
| "grad_norm": 1.6015137434005737, | |
| "learning_rate": 9.03790087463557e-06, | |
| "loss": 1.1398, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.28002625246116825, | |
| "grad_norm": 1.3944435119628906, | |
| "learning_rate": 9.329446064139942e-06, | |
| "loss": 1.1174, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.28877707285057974, | |
| "grad_norm": 1.2441927194595337, | |
| "learning_rate": 9.620991253644316e-06, | |
| "loss": 1.1416, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.29752789323999124, | |
| "grad_norm": 1.4111624956130981, | |
| "learning_rate": 9.912536443148689e-06, | |
| "loss": 1.1352, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.30627871362940273, | |
| "grad_norm": 1.3007830381393433, | |
| "learning_rate": 9.999872800049233e-06, | |
| "loss": 1.1208, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.3150295340188143, | |
| "grad_norm": 1.3299254179000854, | |
| "learning_rate": 9.999249795462235e-06, | |
| "loss": 1.142, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.3237803544082258, | |
| "grad_norm": 1.2524114847183228, | |
| "learning_rate": 9.998107687592326e-06, | |
| "loss": 1.1147, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.3325311747976373, | |
| "grad_norm": 1.4711254835128784, | |
| "learning_rate": 9.996446595031691e-06, | |
| "loss": 1.1182, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.34128199518704877, | |
| "grad_norm": 1.3600913286209106, | |
| "learning_rate": 9.994266690261928e-06, | |
| "loss": 1.1565, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.35003281557646027, | |
| "grad_norm": 1.3752756118774414, | |
| "learning_rate": 9.991568199636149e-06, | |
| "loss": 1.0803, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3587836359658718, | |
| "grad_norm": 1.2283236980438232, | |
| "learning_rate": 9.988351403355455e-06, | |
| "loss": 1.1371, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.3675344563552833, | |
| "grad_norm": 1.266762614250183, | |
| "learning_rate": 9.98461663543987e-06, | |
| "loss": 1.1027, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.3762852767446948, | |
| "grad_norm": 1.2771412134170532, | |
| "learning_rate": 9.980364283693638e-06, | |
| "loss": 1.0891, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.3850360971341063, | |
| "grad_norm": 1.2226617336273193, | |
| "learning_rate": 9.97559478966496e-06, | |
| "loss": 1.1105, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.39378691752351785, | |
| "grad_norm": 1.2738267183303833, | |
| "learning_rate": 9.970308648600142e-06, | |
| "loss": 1.114, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.40253773791292935, | |
| "grad_norm": 1.3344817161560059, | |
| "learning_rate": 9.964506409392178e-06, | |
| "loss": 1.1178, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.41128855830234085, | |
| "grad_norm": 1.3076369762420654, | |
| "learning_rate": 9.958188674523748e-06, | |
| "loss": 1.1205, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.42003937869175234, | |
| "grad_norm": 1.251315712928772, | |
| "learning_rate": 9.951356100004666e-06, | |
| "loss": 1.0795, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.42879019908116384, | |
| "grad_norm": 1.1784945726394653, | |
| "learning_rate": 9.94400939530375e-06, | |
| "loss": 1.1175, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.4375410194705754, | |
| "grad_norm": 1.219256043434143, | |
| "learning_rate": 9.93614932327517e-06, | |
| "loss": 1.0933, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.4462918398599869, | |
| "grad_norm": 1.398763656616211, | |
| "learning_rate": 9.927776700079217e-06, | |
| "loss": 1.1267, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.4550426602493984, | |
| "grad_norm": 1.25184166431427, | |
| "learning_rate": 9.918892395097572e-06, | |
| "loss": 1.1191, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.4637934806388099, | |
| "grad_norm": 1.3422456979751587, | |
| "learning_rate": 9.90949733084303e-06, | |
| "loss": 1.1003, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.47254430102822137, | |
| "grad_norm": 1.3312543630599976, | |
| "learning_rate": 9.8995924828637e-06, | |
| "loss": 1.1004, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4812951214176329, | |
| "grad_norm": 1.2708804607391357, | |
| "learning_rate": 9.889178879641717e-06, | |
| "loss": 1.1094, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.4900459418070444, | |
| "grad_norm": 1.3565181493759155, | |
| "learning_rate": 9.878257602486445e-06, | |
| "loss": 1.1252, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.4987967621964559, | |
| "grad_norm": 1.232021450996399, | |
| "learning_rate": 9.866829785422201e-06, | |
| "loss": 1.0976, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.5075475825858674, | |
| "grad_norm": 1.371229887008667, | |
| "learning_rate": 9.854896615070495e-06, | |
| "loss": 1.0941, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.516298402975279, | |
| "grad_norm": 1.35117769241333, | |
| "learning_rate": 9.842459330526826e-06, | |
| "loss": 1.1127, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.5250492233646904, | |
| "grad_norm": 1.234413504600525, | |
| "learning_rate": 9.829519223232006e-06, | |
| "loss": 1.09, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.533800043754102, | |
| "grad_norm": 1.2576864957809448, | |
| "learning_rate": 9.816077636838079e-06, | |
| "loss": 1.1094, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.5425508641435135, | |
| "grad_norm": 1.340950608253479, | |
| "learning_rate": 9.802135967068783e-06, | |
| "loss": 1.0889, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.5513016845329249, | |
| "grad_norm": 1.350095272064209, | |
| "learning_rate": 9.787695661574632e-06, | |
| "loss": 1.0661, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.5600525049223365, | |
| "grad_norm": 1.2629624605178833, | |
| "learning_rate": 9.772758219782601e-06, | |
| "loss": 1.0823, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.5688033253117479, | |
| "grad_norm": 1.234574556350708, | |
| "learning_rate": 9.757325192740425e-06, | |
| "loss": 1.0937, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5775541457011595, | |
| "grad_norm": 1.1775031089782715, | |
| "learning_rate": 9.741398182955548e-06, | |
| "loss": 1.1066, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.586304966090571, | |
| "grad_norm": 1.2653073072433472, | |
| "learning_rate": 9.724978844228717e-06, | |
| "loss": 1.0964, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.5950557864799825, | |
| "grad_norm": 1.2665278911590576, | |
| "learning_rate": 9.70806888148227e-06, | |
| "loss": 1.1007, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.603806606869394, | |
| "grad_norm": 1.2909613847732544, | |
| "learning_rate": 9.690670050583095e-06, | |
| "loss": 1.0973, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.6125574272588055, | |
| "grad_norm": 1.2396572828292847, | |
| "learning_rate": 9.672784158160307e-06, | |
| "loss": 1.0423, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.621308247648217, | |
| "grad_norm": 1.2569255828857422, | |
| "learning_rate": 9.654413061417655e-06, | |
| "loss": 1.0764, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.6300590680376286, | |
| "grad_norm": 1.3747756481170654, | |
| "learning_rate": 9.635558667940683e-06, | |
| "loss": 1.0563, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.63880988842704, | |
| "grad_norm": 1.2752923965454102, | |
| "learning_rate": 9.616222935498647e-06, | |
| "loss": 1.1084, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.6475607088164516, | |
| "grad_norm": 1.218320369720459, | |
| "learning_rate": 9.596407871841229e-06, | |
| "loss": 1.0674, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.656311529205863, | |
| "grad_norm": 1.2466013431549072, | |
| "learning_rate": 9.57611553449006e-06, | |
| "loss": 1.1048, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.6650623495952745, | |
| "grad_norm": 1.2956342697143555, | |
| "learning_rate": 9.555348030525076e-06, | |
| "loss": 1.0723, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.6738131699846861, | |
| "grad_norm": 1.29911208152771, | |
| "learning_rate": 9.534107516365725e-06, | |
| "loss": 1.0743, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.6825639903740975, | |
| "grad_norm": 1.2063359022140503, | |
| "learning_rate": 9.512396197547053e-06, | |
| "loss": 1.0717, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.6913148107635091, | |
| "grad_norm": 1.184462547302246, | |
| "learning_rate": 9.490216328490693e-06, | |
| "loss": 1.0895, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.7000656311529205, | |
| "grad_norm": 1.186874270439148, | |
| "learning_rate": 9.467570212270767e-06, | |
| "loss": 1.1476, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.7088164515423321, | |
| "grad_norm": 1.2901315689086914, | |
| "learning_rate": 9.444460200374747e-06, | |
| "loss": 1.0538, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.7175672719317436, | |
| "grad_norm": 1.2175313234329224, | |
| "learning_rate": 9.42088869245929e-06, | |
| "loss": 1.079, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.7263180923211551, | |
| "grad_norm": 1.1771786212921143, | |
| "learning_rate": 9.396858136101066e-06, | |
| "loss": 1.0756, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.7350689127105666, | |
| "grad_norm": 1.2814205884933472, | |
| "learning_rate": 9.372371026542595e-06, | |
| "loss": 1.0697, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.7438197330999782, | |
| "grad_norm": 1.2603764533996582, | |
| "learning_rate": 9.347429906433175e-06, | |
| "loss": 1.1079, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.7525705534893896, | |
| "grad_norm": 1.2228299379348755, | |
| "learning_rate": 9.322037365564844e-06, | |
| "loss": 1.0746, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.7613213738788012, | |
| "grad_norm": 1.1895763874053955, | |
| "learning_rate": 9.296196040603476e-06, | |
| "loss": 1.0831, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.7700721942682126, | |
| "grad_norm": 1.2288453578948975, | |
| "learning_rate": 9.269908614814995e-06, | |
| "loss": 1.0649, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.7788230146576242, | |
| "grad_norm": 1.1580345630645752, | |
| "learning_rate": 9.243177817786757e-06, | |
| "loss": 1.1053, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.7875738350470357, | |
| "grad_norm": 1.1613861322402954, | |
| "learning_rate": 9.216006425144118e-06, | |
| "loss": 1.0854, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.7963246554364471, | |
| "grad_norm": 1.0809310674667358, | |
| "learning_rate": 9.188397258262221e-06, | |
| "loss": 1.059, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.8050754758258587, | |
| "grad_norm": 1.2021125555038452, | |
| "learning_rate": 9.160353183973045e-06, | |
| "loss": 1.064, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.8138262962152701, | |
| "grad_norm": 1.286584496498108, | |
| "learning_rate": 9.131877114267711e-06, | |
| "loss": 1.0516, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.8225771166046817, | |
| "grad_norm": 1.177673578262329, | |
| "learning_rate": 9.102972005994123e-06, | |
| "loss": 1.0358, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.8313279369940932, | |
| "grad_norm": 1.2117091417312622, | |
| "learning_rate": 9.073640860549931e-06, | |
| "loss": 1.0701, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.8400787573835047, | |
| "grad_norm": 1.2018595933914185, | |
| "learning_rate": 9.043886723570885e-06, | |
| "loss": 1.0833, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.8488295777729162, | |
| "grad_norm": 1.1337984800338745, | |
| "learning_rate": 9.013712684614582e-06, | |
| "loss": 1.0785, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.8575803981623277, | |
| "grad_norm": 1.2217832803726196, | |
| "learning_rate": 8.983121876839658e-06, | |
| "loss": 1.0623, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.8663312185517392, | |
| "grad_norm": 1.2249287366867065, | |
| "learning_rate": 8.95211747668046e-06, | |
| "loss": 1.0285, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.8750820389411508, | |
| "grad_norm": 1.1673548221588135, | |
| "learning_rate": 8.920702703517205e-06, | |
| "loss": 1.0354, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.8838328593305622, | |
| "grad_norm": 1.1298819780349731, | |
| "learning_rate": 8.888880819341703e-06, | |
| "loss": 1.0538, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.8925836797199738, | |
| "grad_norm": 1.1600337028503418, | |
| "learning_rate": 8.856655128418638e-06, | |
| "loss": 1.0527, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.9013345001093852, | |
| "grad_norm": 1.2292169332504272, | |
| "learning_rate": 8.824028976942465e-06, | |
| "loss": 1.0613, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.9100853204987968, | |
| "grad_norm": 1.1878681182861328, | |
| "learning_rate": 8.791005752689958e-06, | |
| "loss": 1.0489, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.9188361408882083, | |
| "grad_norm": 1.2165477275848389, | |
| "learning_rate": 8.757588884668435e-06, | |
| "loss": 1.0675, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.9275869612776197, | |
| "grad_norm": 1.1682305335998535, | |
| "learning_rate": 8.723781842759696e-06, | |
| "loss": 1.084, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.9363377816670313, | |
| "grad_norm": 1.3387882709503174, | |
| "learning_rate": 8.689588137359731e-06, | |
| "loss": 1.0564, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.9450886020564427, | |
| "grad_norm": 1.2144585847854614, | |
| "learning_rate": 8.655011319014217e-06, | |
| "loss": 1.08, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.9538394224458543, | |
| "grad_norm": 1.2637611627578735, | |
| "learning_rate": 8.620054978049828e-06, | |
| "loss": 1.0716, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.9625902428352658, | |
| "grad_norm": 1.230635404586792, | |
| "learning_rate": 8.584722744201448e-06, | |
| "loss": 1.0429, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.9713410632246773, | |
| "grad_norm": 1.2495629787445068, | |
| "learning_rate": 8.549018286235255e-06, | |
| "loss": 1.0464, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.9800918836140888, | |
| "grad_norm": 1.1702367067337036, | |
| "learning_rate": 8.512945311567786e-06, | |
| "loss": 1.0476, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.9888427040035004, | |
| "grad_norm": 1.1367738246917725, | |
| "learning_rate": 8.476507565880959e-06, | |
| "loss": 1.0392, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.9975935243929118, | |
| "grad_norm": 1.1903005838394165, | |
| "learning_rate": 8.439708832733142e-06, | |
| "loss": 1.0414, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.006125574272588, | |
| "grad_norm": 1.310563325881958, | |
| "learning_rate": 8.402552933166288e-06, | |
| "loss": 0.9966, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.0148763946619996, | |
| "grad_norm": 1.2823761701583862, | |
| "learning_rate": 8.36504372530916e-06, | |
| "loss": 0.9961, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.023627215051411, | |
| "grad_norm": 1.340758204460144, | |
| "learning_rate": 8.327185103976731e-06, | |
| "loss": 0.9904, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.0323780354408225, | |
| "grad_norm": 1.2751260995864868, | |
| "learning_rate": 8.288981000265748e-06, | |
| "loss": 1.004, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.041128855830234, | |
| "grad_norm": 1.1516119241714478, | |
| "learning_rate": 8.25043538114655e-06, | |
| "loss": 0.9757, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.0498796762196456, | |
| "grad_norm": 1.206026315689087, | |
| "learning_rate": 8.211552249051149e-06, | |
| "loss": 0.9799, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.0586304966090572, | |
| "grad_norm": 1.2714760303497314, | |
| "learning_rate": 8.17233564145764e-06, | |
| "loss": 0.9731, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.0673813169984685, | |
| "grad_norm": 1.2047169208526611, | |
| "learning_rate": 8.132789630470952e-06, | |
| "loss": 0.971, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.07613213738788, | |
| "grad_norm": 1.282832384109497, | |
| "learning_rate": 8.092918322400028e-06, | |
| "loss": 0.9695, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.0848829577772916, | |
| "grad_norm": 1.2463730573654175, | |
| "learning_rate": 8.05272585733143e-06, | |
| "loss": 0.9727, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.0936337781667032, | |
| "grad_norm": 1.230220079421997, | |
| "learning_rate": 8.012216408699456e-06, | |
| "loss": 0.9909, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.1023845985561147, | |
| "grad_norm": 1.3110296726226807, | |
| "learning_rate": 7.97139418285279e-06, | |
| "loss": 0.9887, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.1111354189455263, | |
| "grad_norm": 1.234381914138794, | |
| "learning_rate": 7.930263418617714e-06, | |
| "loss": 1.0014, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.1198862393349376, | |
| "grad_norm": 1.2342872619628906, | |
| "learning_rate": 7.888828386857985e-06, | |
| "loss": 0.9795, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.1286370597243491, | |
| "grad_norm": 1.279505729675293, | |
| "learning_rate": 7.847093390031351e-06, | |
| "loss": 0.9647, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.1373878801137607, | |
| "grad_norm": 1.3248056173324585, | |
| "learning_rate": 7.8050627617428e-06, | |
| "loss": 0.9881, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.1461387005031722, | |
| "grad_norm": 1.234870433807373, | |
| "learning_rate": 7.76274086629458e-06, | |
| "loss": 0.9728, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.1548895208925836, | |
| "grad_norm": 1.2661969661712646, | |
| "learning_rate": 7.72013209823303e-06, | |
| "loss": 0.9716, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.1636403412819951, | |
| "grad_norm": 1.180408239364624, | |
| "learning_rate": 7.677240881892259e-06, | |
| "loss": 0.9641, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.1723911616714067, | |
| "grad_norm": 1.2939667701721191, | |
| "learning_rate": 7.634071670934734e-06, | |
| "loss": 0.9832, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.1811419820608182, | |
| "grad_norm": 1.1289607286453247, | |
| "learning_rate": 7.5906289478888475e-06, | |
| "loss": 0.9733, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.1898928024502298, | |
| "grad_norm": 1.2281969785690308, | |
| "learning_rate": 7.546917223683453e-06, | |
| "loss": 0.9645, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.1986436228396413, | |
| "grad_norm": 1.226364254951477, | |
| "learning_rate": 7.502941037179474e-06, | |
| "loss": 0.9668, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.2073944432290526, | |
| "grad_norm": 1.2580183744430542, | |
| "learning_rate": 7.4587049546986e-06, | |
| "loss": 1.0167, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.2161452636184642, | |
| "grad_norm": 1.261915683746338, | |
| "learning_rate": 7.414213569549145e-06, | |
| "loss": 0.9896, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.2248960840078758, | |
| "grad_norm": 1.2158194780349731, | |
| "learning_rate": 7.369471501549087e-06, | |
| "loss": 0.9893, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.2336469043972873, | |
| "grad_norm": 1.2699130773544312, | |
| "learning_rate": 7.324483396546371e-06, | |
| "loss": 1.0171, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.2423977247866986, | |
| "grad_norm": 1.1923282146453857, | |
| "learning_rate": 7.279253925936498e-06, | |
| "loss": 0.9802, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.2511485451761102, | |
| "grad_norm": 1.1498876810073853, | |
| "learning_rate": 7.233787786177464e-06, | |
| "loss": 0.9683, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.2598993655655217, | |
| "grad_norm": 1.2153609991073608, | |
| "learning_rate": 7.188089698302097e-06, | |
| "loss": 0.9916, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.2686501859549333, | |
| "grad_norm": 1.1934350728988647, | |
| "learning_rate": 7.142164407427849e-06, | |
| "loss": 0.9788, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.2774010063443448, | |
| "grad_norm": 1.2987148761749268, | |
| "learning_rate": 7.096016682264074e-06, | |
| "loss": 0.9865, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.2861518267337564, | |
| "grad_norm": 1.2581541538238525, | |
| "learning_rate": 7.049651314616862e-06, | |
| "loss": 0.9494, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.2949026471231677, | |
| "grad_norm": 1.257711410522461, | |
| "learning_rate": 7.003073118891487e-06, | |
| "loss": 0.9534, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.3036534675125793, | |
| "grad_norm": 1.3140227794647217, | |
| "learning_rate": 6.956286931592481e-06, | |
| "loss": 0.9888, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.3124042879019908, | |
| "grad_norm": 1.2232003211975098, | |
| "learning_rate": 6.909297610821444e-06, | |
| "loss": 0.9632, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.3211551082914024, | |
| "grad_norm": 1.190757393836975, | |
| "learning_rate": 6.86211003577259e-06, | |
| "loss": 0.9598, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.3299059286808137, | |
| "grad_norm": 1.180031657218933, | |
| "learning_rate": 6.81472910622611e-06, | |
| "loss": 0.9804, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.3386567490702252, | |
| "grad_norm": 1.2404285669326782, | |
| "learning_rate": 6.7671597420394e-06, | |
| "loss": 0.9915, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.3474075694596368, | |
| "grad_norm": 1.2322285175323486, | |
| "learning_rate": 6.719406882636196e-06, | |
| "loss": 0.9908, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.3561583898490484, | |
| "grad_norm": 1.1908921003341675, | |
| "learning_rate": 6.671475486493691e-06, | |
| "loss": 0.9716, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.36490921023846, | |
| "grad_norm": 1.2217786312103271, | |
| "learning_rate": 6.623370530627661e-06, | |
| "loss": 0.9602, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.3736600306278715, | |
| "grad_norm": 1.2622575759887695, | |
| "learning_rate": 6.575097010075665e-06, | |
| "loss": 0.9509, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.382410851017283, | |
| "grad_norm": 1.2419332265853882, | |
| "learning_rate": 6.526659937378392e-06, | |
| "loss": 0.9821, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.3911616714066943, | |
| "grad_norm": 1.3119240999221802, | |
| "learning_rate": 6.478064342059165e-06, | |
| "loss": 0.9823, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.3999124917961059, | |
| "grad_norm": 1.3174080848693848, | |
| "learning_rate": 6.429315270101702e-06, | |
| "loss": 0.9844, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.4086633121855174, | |
| "grad_norm": 1.2387707233428955, | |
| "learning_rate": 6.380417783426154e-06, | |
| "loss": 0.9545, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.4174141325749288, | |
| "grad_norm": 1.1914652585983276, | |
| "learning_rate": 6.331376959363501e-06, | |
| "loss": 0.9792, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.4261649529643403, | |
| "grad_norm": 1.2240160703659058, | |
| "learning_rate": 6.2821978901283386e-06, | |
| "loss": 0.9573, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.4349157733537519, | |
| "grad_norm": 1.2401772737503052, | |
| "learning_rate": 6.232885682290119e-06, | |
| "loss": 0.9669, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.4436665937431634, | |
| "grad_norm": 1.2027502059936523, | |
| "learning_rate": 6.183445456242905e-06, | |
| "loss": 0.9815, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.452417414132575, | |
| "grad_norm": 1.2344744205474854, | |
| "learning_rate": 6.13388234567369e-06, | |
| "loss": 0.9361, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.4611682345219865, | |
| "grad_norm": 1.2103825807571411, | |
| "learning_rate": 6.084201497029333e-06, | |
| "loss": 0.9744, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.469919054911398, | |
| "grad_norm": 1.3063782453536987, | |
| "learning_rate": 6.034408068982171e-06, | |
| "loss": 0.9681, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.4786698753008094, | |
| "grad_norm": 1.2130053043365479, | |
| "learning_rate": 5.984507231894364e-06, | |
| "loss": 0.9586, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.487420695690221, | |
| "grad_norm": 1.2866191864013672, | |
| "learning_rate": 5.934504167281022e-06, | |
| "loss": 0.9681, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.4961715160796325, | |
| "grad_norm": 1.3027279376983643, | |
| "learning_rate": 5.884404067272175e-06, | |
| "loss": 0.975, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.5049223364690438, | |
| "grad_norm": 1.2601852416992188, | |
| "learning_rate": 5.834212134073645e-06, | |
| "loss": 0.9388, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.5136731568584554, | |
| "grad_norm": 1.2413978576660156, | |
| "learning_rate": 5.783933579426867e-06, | |
| "loss": 0.9723, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.522423977247867, | |
| "grad_norm": 1.2116490602493286, | |
| "learning_rate": 5.733573624067717e-06, | |
| "loss": 0.973, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.5311747976372785, | |
| "grad_norm": 1.2060950994491577, | |
| "learning_rate": 5.683137497184424e-06, | |
| "loss": 0.9375, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.53992561802669, | |
| "grad_norm": 1.2472314834594727, | |
| "learning_rate": 5.63263043587457e-06, | |
| "loss": 1.0112, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.5486764384161016, | |
| "grad_norm": 1.2653073072433472, | |
| "learning_rate": 5.582057684601308e-06, | |
| "loss": 0.9778, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.5574272588055131, | |
| "grad_norm": 1.2270764112472534, | |
| "learning_rate": 5.531424494648789e-06, | |
| "loss": 0.9807, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.5661780791949247, | |
| "grad_norm": 1.2816975116729736, | |
| "learning_rate": 5.480736123576886e-06, | |
| "loss": 0.9847, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.574928899584336, | |
| "grad_norm": 1.2429735660552979, | |
| "learning_rate": 5.429997834675276e-06, | |
| "loss": 0.9886, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.5836797199737476, | |
| "grad_norm": 1.3056929111480713, | |
| "learning_rate": 5.379214896416908e-06, | |
| "loss": 0.9514, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.592430540363159, | |
| "grad_norm": 1.155527949333191, | |
| "learning_rate": 5.328392581910962e-06, | |
| "loss": 0.9593, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.6011813607525704, | |
| "grad_norm": 1.2867212295532227, | |
| "learning_rate": 5.277536168355293e-06, | |
| "loss": 0.9722, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.609932181141982, | |
| "grad_norm": 1.3213430643081665, | |
| "learning_rate": 5.226650936488474e-06, | |
| "loss": 0.9588, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.6186830015313936, | |
| "grad_norm": 1.2772150039672852, | |
| "learning_rate": 5.1757421700414645e-06, | |
| "loss": 0.9729, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.627433821920805, | |
| "grad_norm": 1.2589421272277832, | |
| "learning_rate": 5.124815155188966e-06, | |
| "loss": 1.0171, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.6361846423102167, | |
| "grad_norm": 1.2326509952545166, | |
| "learning_rate": 5.073875180000523e-06, | |
| "loss": 0.9622, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.6449354626996282, | |
| "grad_norm": 1.1593009233474731, | |
| "learning_rate": 5.022927533891434e-06, | |
| "loss": 0.9584, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.6536862830890398, | |
| "grad_norm": 1.303597331047058, | |
| "learning_rate": 4.97197750707352e-06, | |
| "loss": 0.9994, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 1.662437103478451, | |
| "grad_norm": 1.2183244228363037, | |
| "learning_rate": 4.9210303900058025e-06, | |
| "loss": 0.9582, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.6711879238678626, | |
| "grad_norm": 1.2271572351455688, | |
| "learning_rate": 4.870091472845166e-06, | |
| "loss": 0.9775, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 1.679938744257274, | |
| "grad_norm": 1.222212791442871, | |
| "learning_rate": 4.819166044897047e-06, | |
| "loss": 0.9578, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 1.6886895646466855, | |
| "grad_norm": 1.2827479839324951, | |
| "learning_rate": 4.7682593940662156e-06, | |
| "loss": 0.982, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 1.697440385036097, | |
| "grad_norm": 1.2533340454101562, | |
| "learning_rate": 4.717376806307696e-06, | |
| "loss": 0.9735, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 1.7061912054255086, | |
| "grad_norm": 1.267432451248169, | |
| "learning_rate": 4.666523565077889e-06, | |
| "loss": 0.9598, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.7149420258149202, | |
| "grad_norm": 1.2450517416000366, | |
| "learning_rate": 4.615704950785965e-06, | |
| "loss": 0.9751, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 1.7236928462043317, | |
| "grad_norm": 1.2049565315246582, | |
| "learning_rate": 4.5649262402455705e-06, | |
| "loss": 0.9737, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 1.7324436665937433, | |
| "grad_norm": 1.2524805068969727, | |
| "learning_rate": 4.514192706126885e-06, | |
| "loss": 0.9662, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.7411944869831548, | |
| "grad_norm": 1.2566184997558594, | |
| "learning_rate": 4.463509616409144e-06, | |
| "loss": 0.9618, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 1.7499453073725662, | |
| "grad_norm": 1.176069736480713, | |
| "learning_rate": 4.4128822338336295e-06, | |
| "loss": 0.9521, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.7586961277619777, | |
| "grad_norm": 1.1588139533996582, | |
| "learning_rate": 4.362315815357197e-06, | |
| "loss": 0.9488, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 1.767446948151389, | |
| "grad_norm": 1.1984119415283203, | |
| "learning_rate": 4.311815611606423e-06, | |
| "loss": 0.9702, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 1.7761977685408006, | |
| "grad_norm": 1.1387180089950562, | |
| "learning_rate": 4.261386866332397e-06, | |
| "loss": 0.9528, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 1.7849485889302121, | |
| "grad_norm": 1.1617263555526733, | |
| "learning_rate": 4.211034815866228e-06, | |
| "loss": 0.951, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 1.7936994093196237, | |
| "grad_norm": 1.1776400804519653, | |
| "learning_rate": 4.160764688575321e-06, | |
| "loss": 0.9634, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.8024502297090352, | |
| "grad_norm": 1.25057852268219, | |
| "learning_rate": 4.110581704320495e-06, | |
| "loss": 0.9633, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 1.8112010500984468, | |
| "grad_norm": 1.2121225595474243, | |
| "learning_rate": 4.0604910739139575e-06, | |
| "loss": 0.9749, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 1.8199518704878583, | |
| "grad_norm": 1.3024873733520508, | |
| "learning_rate": 4.0104979985782394e-06, | |
| "loss": 0.9744, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 1.8287026908772699, | |
| "grad_norm": 1.1998348236083984, | |
| "learning_rate": 3.960607669406129e-06, | |
| "loss": 0.9622, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 1.8374535112666812, | |
| "grad_norm": 1.2691677808761597, | |
| "learning_rate": 3.9108252668216294e-06, | |
| "loss": 0.9702, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.8462043316560928, | |
| "grad_norm": 1.2797294855117798, | |
| "learning_rate": 3.861155960042057e-06, | |
| "loss": 0.9294, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 1.8549551520455043, | |
| "grad_norm": 1.2609057426452637, | |
| "learning_rate": 3.811604906541287e-06, | |
| "loss": 0.9624, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 1.8637059724349156, | |
| "grad_norm": 1.2273874282836914, | |
| "learning_rate": 3.762177251514218e-06, | |
| "loss": 0.9563, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 1.8724567928243272, | |
| "grad_norm": 1.225128173828125, | |
| "learning_rate": 3.7128781273425156e-06, | |
| "loss": 0.9587, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 1.8812076132137387, | |
| "grad_norm": 1.2284705638885498, | |
| "learning_rate": 3.6637126530616816e-06, | |
| "loss": 0.9401, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.8899584336031503, | |
| "grad_norm": 1.2802456617355347, | |
| "learning_rate": 3.6146859338295248e-06, | |
| "loss": 0.9386, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 1.8987092539925619, | |
| "grad_norm": 1.22531259059906, | |
| "learning_rate": 3.5658030603960437e-06, | |
| "loss": 0.9652, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 1.9074600743819734, | |
| "grad_norm": 1.2205225229263306, | |
| "learning_rate": 3.5170691085748325e-06, | |
| "loss": 0.9418, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 1.916210894771385, | |
| "grad_norm": 1.1588237285614014, | |
| "learning_rate": 3.4684891387160292e-06, | |
| "loss": 0.9624, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 1.9249617151607963, | |
| "grad_norm": 1.2422109842300415, | |
| "learning_rate": 3.42006819518086e-06, | |
| "loss": 0.9714, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.9337125355502078, | |
| "grad_norm": 1.2960257530212402, | |
| "learning_rate": 3.371811305817858e-06, | |
| "loss": 0.9521, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 1.9424633559396194, | |
| "grad_norm": 1.1022367477416992, | |
| "learning_rate": 3.3237234814407904e-06, | |
| "loss": 0.9775, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 1.9512141763290307, | |
| "grad_norm": 1.160735011100769, | |
| "learning_rate": 3.275809715308349e-06, | |
| "loss": 0.9497, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 1.9599649967184423, | |
| "grad_norm": 1.2306522130966187, | |
| "learning_rate": 3.2280749826056736e-06, | |
| "loss": 0.9544, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 1.9687158171078538, | |
| "grad_norm": 1.1739164590835571, | |
| "learning_rate": 3.180524239927749e-06, | |
| "loss": 0.9526, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.9774666374972654, | |
| "grad_norm": 1.2096645832061768, | |
| "learning_rate": 3.1331624247647276e-06, | |
| "loss": 0.9503, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 1.986217457886677, | |
| "grad_norm": 1.2347972393035889, | |
| "learning_rate": 3.085994454989232e-06, | |
| "loss": 0.9473, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 1.9949682782760885, | |
| "grad_norm": 1.28371000289917, | |
| "learning_rate": 3.039025228345725e-06, | |
| "loss": 0.9787, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 2.0035003281557646, | |
| "grad_norm": 1.1137796640396118, | |
| "learning_rate": 2.992259621941916e-06, | |
| "loss": 0.9234, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 2.012251148545176, | |
| "grad_norm": 1.3428007364273071, | |
| "learning_rate": 2.9457024917423594e-06, | |
| "loss": 0.8978, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.0210019689345877, | |
| "grad_norm": 1.2651664018630981, | |
| "learning_rate": 2.8993586720642307e-06, | |
| "loss": 0.8992, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 2.0297527893239993, | |
| "grad_norm": 1.1616652011871338, | |
| "learning_rate": 2.85323297507534e-06, | |
| "loss": 0.8695, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 2.038503609713411, | |
| "grad_norm": 1.2452731132507324, | |
| "learning_rate": 2.8073301902944518e-06, | |
| "loss": 0.901, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 2.047254430102822, | |
| "grad_norm": 1.2381782531738281, | |
| "learning_rate": 2.761655084093976e-06, | |
| "loss": 0.9096, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 2.0560052504922335, | |
| "grad_norm": 1.265434980392456, | |
| "learning_rate": 2.7162123992050316e-06, | |
| "loss": 0.8984, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.064756070881645, | |
| "grad_norm": 1.2683861255645752, | |
| "learning_rate": 2.671006854224982e-06, | |
| "loss": 0.8864, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 2.0735068912710566, | |
| "grad_norm": 1.2859617471694946, | |
| "learning_rate": 2.626043143127477e-06, | |
| "loss": 0.8728, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 2.082257711660468, | |
| "grad_norm": 1.3000761270523071, | |
| "learning_rate": 2.5813259347750445e-06, | |
| "loss": 0.932, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 2.0910085320498797, | |
| "grad_norm": 1.23272705078125, | |
| "learning_rate": 2.5368598724342974e-06, | |
| "loss": 0.9114, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 2.0997593524392912, | |
| "grad_norm": 1.243643879890442, | |
| "learning_rate": 2.4926495732937877e-06, | |
| "loss": 0.8987, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.108510172828703, | |
| "grad_norm": 1.2681580781936646, | |
| "learning_rate": 2.4486996279845864e-06, | |
| "loss": 0.9277, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 2.1172609932181143, | |
| "grad_norm": 1.2045451402664185, | |
| "learning_rate": 2.4050146001035983e-06, | |
| "loss": 0.8988, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 2.126011813607526, | |
| "grad_norm": 1.2446624040603638, | |
| "learning_rate": 2.361599025739695e-06, | |
| "loss": 0.889, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 2.134762633996937, | |
| "grad_norm": 1.298621654510498, | |
| "learning_rate": 2.318457413002718e-06, | |
| "loss": 0.898, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 2.1435134543863485, | |
| "grad_norm": 1.2545632123947144, | |
| "learning_rate": 2.2755942415553593e-06, | |
| "loss": 0.8675, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.15226427477576, | |
| "grad_norm": 1.1765785217285156, | |
| "learning_rate": 2.233013962148016e-06, | |
| "loss": 0.853, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 2.1610150951651717, | |
| "grad_norm": 1.3674156665802002, | |
| "learning_rate": 2.1907209961566418e-06, | |
| "loss": 0.9134, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 2.169765915554583, | |
| "grad_norm": 1.337667465209961, | |
| "learning_rate": 2.1487197351236443e-06, | |
| "loss": 0.8766, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 2.1785167359439948, | |
| "grad_norm": 1.240268349647522, | |
| "learning_rate": 2.107014540301886e-06, | |
| "loss": 0.8921, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 2.1872675563334063, | |
| "grad_norm": 1.2723026275634766, | |
| "learning_rate": 2.0656097422018346e-06, | |
| "loss": 0.8948, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.196018376722818, | |
| "grad_norm": 1.244240164756775, | |
| "learning_rate": 2.0245096401418853e-06, | |
| "loss": 0.902, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 2.2047691971122294, | |
| "grad_norm": 1.2332072257995605, | |
| "learning_rate": 1.9837185018019465e-06, | |
| "loss": 0.9042, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 2.213520017501641, | |
| "grad_norm": 1.3075368404388428, | |
| "learning_rate": 1.9432405627802963e-06, | |
| "loss": 0.8743, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 2.2222708378910525, | |
| "grad_norm": 1.2511454820632935, | |
| "learning_rate": 1.9030800261537742e-06, | |
| "loss": 0.8919, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 2.2310216582804636, | |
| "grad_norm": 1.2848820686340332, | |
| "learning_rate": 1.8632410620413505e-06, | |
| "loss": 0.8948, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.239772478669875, | |
| "grad_norm": 1.1853325366973877, | |
| "learning_rate": 1.823727807171115e-06, | |
| "loss": 0.9262, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 2.2485232990592867, | |
| "grad_norm": 1.2825355529785156, | |
| "learning_rate": 1.7845443644507355e-06, | |
| "loss": 0.8744, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 2.2572741194486983, | |
| "grad_norm": 1.2997822761535645, | |
| "learning_rate": 1.7456948025414288e-06, | |
| "loss": 0.9153, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 2.26602493983811, | |
| "grad_norm": 1.2967973947525024, | |
| "learning_rate": 1.7071831554354822e-06, | |
| "loss": 0.8702, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 2.2747757602275214, | |
| "grad_norm": 1.2506667375564575, | |
| "learning_rate": 1.6690134220373866e-06, | |
| "loss": 0.9055, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.283526580616933, | |
| "grad_norm": 1.2581911087036133, | |
| "learning_rate": 1.631189565748596e-06, | |
| "loss": 0.8953, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 2.2922774010063445, | |
| "grad_norm": 1.2846101522445679, | |
| "learning_rate": 1.5937155140559885e-06, | |
| "loss": 0.8857, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 2.301028221395756, | |
| "grad_norm": 1.317673683166504, | |
| "learning_rate": 1.556595158124049e-06, | |
| "loss": 0.8936, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 2.309779041785167, | |
| "grad_norm": 1.3196806907653809, | |
| "learning_rate": 1.5198323523908266e-06, | |
| "loss": 0.8889, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 2.3185298621745787, | |
| "grad_norm": 1.2810077667236328, | |
| "learning_rate": 1.483430914167701e-06, | |
| "loss": 0.9023, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.3272806825639902, | |
| "grad_norm": 1.3437639474868774, | |
| "learning_rate": 1.447394623243018e-06, | |
| "loss": 0.9372, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 2.336031502953402, | |
| "grad_norm": 1.2470753192901611, | |
| "learning_rate": 1.41172722148959e-06, | |
| "loss": 0.8692, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 2.3447823233428133, | |
| "grad_norm": 1.226980209350586, | |
| "learning_rate": 1.3764324124761735e-06, | |
| "loss": 0.9238, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 2.353533143732225, | |
| "grad_norm": 1.2623775005340576, | |
| "learning_rate": 1.3415138610828987e-06, | |
| "loss": 0.8965, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 2.3622839641216364, | |
| "grad_norm": 1.1816771030426025, | |
| "learning_rate": 1.3069751931207164e-06, | |
| "loss": 0.863, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.371034784511048, | |
| "grad_norm": 1.0856176614761353, | |
| "learning_rate": 1.2728199949549143e-06, | |
| "loss": 0.8694, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 2.3797856049004595, | |
| "grad_norm": 1.1386377811431885, | |
| "learning_rate": 1.2390518131327173e-06, | |
| "loss": 0.8742, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 2.388536425289871, | |
| "grad_norm": 1.1820670366287231, | |
| "learning_rate": 1.205674154015033e-06, | |
| "loss": 0.8797, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 2.3972872456792826, | |
| "grad_norm": 1.3150583505630493, | |
| "learning_rate": 1.1726904834123582e-06, | |
| "loss": 0.9003, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 2.4060380660686937, | |
| "grad_norm": 1.30619478225708, | |
| "learning_rate": 1.1401042262249056e-06, | |
| "loss": 0.9086, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.4147888864581053, | |
| "grad_norm": 1.3051726818084717, | |
| "learning_rate": 1.10791876608698e-06, | |
| "loss": 0.8602, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 2.423539706847517, | |
| "grad_norm": 1.225061297416687, | |
| "learning_rate": 1.0761374450156203e-06, | |
| "loss": 0.8669, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 2.4322905272369284, | |
| "grad_norm": 1.1500948667526245, | |
| "learning_rate": 1.04476356306359e-06, | |
| "loss": 0.8733, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 2.44104134762634, | |
| "grad_norm": 1.2501094341278076, | |
| "learning_rate": 1.0138003779767102e-06, | |
| "loss": 0.9025, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 2.4497921680157515, | |
| "grad_norm": 1.1673866510391235, | |
| "learning_rate": 9.832511048555825e-07, | |
| "loss": 0.918, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.458542988405163, | |
| "grad_norm": 1.2898247241973877, | |
| "learning_rate": 9.53118915821748e-07, | |
| "loss": 0.8854, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 2.4672938087945746, | |
| "grad_norm": 1.201026201248169, | |
| "learning_rate": 9.234069396883066e-07, | |
| "loss": 0.9144, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 2.476044629183986, | |
| "grad_norm": 1.1994189023971558, | |
| "learning_rate": 8.941182616350319e-07, | |
| "loss": 0.889, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 2.4847954495733973, | |
| "grad_norm": 1.2872554063796997, | |
| "learning_rate": 8.652559228880169e-07, | |
| "loss": 0.8989, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 2.493546269962809, | |
| "grad_norm": 1.286000370979309, | |
| "learning_rate": 8.368229204038897e-07, | |
| "loss": 0.9244, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.5022970903522204, | |
| "grad_norm": 1.227409839630127, | |
| "learning_rate": 8.088222065586132e-07, | |
| "loss": 0.9, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 2.511047910741632, | |
| "grad_norm": 1.2374303340911865, | |
| "learning_rate": 7.812566888409228e-07, | |
| "loss": 0.9094, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 2.5197987311310435, | |
| "grad_norm": 1.3504208326339722, | |
| "learning_rate": 7.541292295504327e-07, | |
| "loss": 0.8945, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 2.528549551520455, | |
| "grad_norm": 1.203357219696045, | |
| "learning_rate": 7.27442645500413e-07, | |
| "loss": 0.8938, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 2.5373003719098666, | |
| "grad_norm": 1.2361124753952026, | |
| "learning_rate": 7.011997077253113e-07, | |
| "loss": 0.8804, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.546051192299278, | |
| "grad_norm": 1.2695845365524292, | |
| "learning_rate": 6.75403141193014e-07, | |
| "loss": 0.901, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 2.5548020126886897, | |
| "grad_norm": 1.3083370923995972, | |
| "learning_rate": 6.500556245218986e-07, | |
| "loss": 0.903, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 2.5635528330781012, | |
| "grad_norm": 1.253382921218872, | |
| "learning_rate": 6.25159789702695e-07, | |
| "loss": 0.8944, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 2.5723036534675128, | |
| "grad_norm": 1.2046483755111694, | |
| "learning_rate": 6.007182218251883e-07, | |
| "loss": 0.8654, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 2.5810544738569243, | |
| "grad_norm": 1.286098837852478, | |
| "learning_rate": 5.767334588097979e-07, | |
| "loss": 0.8889, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.5898052942463354, | |
| "grad_norm": 1.212591290473938, | |
| "learning_rate": 5.532079911440419e-07, | |
| "loss": 0.8724, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 2.598556114635747, | |
| "grad_norm": 1.2204617261886597, | |
| "learning_rate": 5.301442616239371e-07, | |
| "loss": 0.8696, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 2.6073069350251585, | |
| "grad_norm": 1.1826462745666504, | |
| "learning_rate": 5.075446651003535e-07, | |
| "loss": 0.8542, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 2.61605775541457, | |
| "grad_norm": 1.270414113998413, | |
| "learning_rate": 4.854115482303345e-07, | |
| "loss": 0.8681, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 2.6248085758039816, | |
| "grad_norm": 1.2174789905548096, | |
| "learning_rate": 4.6374720923343275e-07, | |
| "loss": 0.9086, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.633559396193393, | |
| "grad_norm": 1.2752397060394287, | |
| "learning_rate": 4.4255389765307077e-07, | |
| "loss": 0.8978, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 2.6423102165828047, | |
| "grad_norm": 1.2312636375427246, | |
| "learning_rate": 4.2183381412295666e-07, | |
| "loss": 0.8834, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 2.6510610369722163, | |
| "grad_norm": 1.1354252099990845, | |
| "learning_rate": 4.015891101385794e-07, | |
| "loss": 0.8863, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 2.6598118573616274, | |
| "grad_norm": 1.2844924926757812, | |
| "learning_rate": 3.8182188783380534e-07, | |
| "loss": 0.8984, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 2.668562677751039, | |
| "grad_norm": 1.2184648513793945, | |
| "learning_rate": 3.6253419976260027e-07, | |
| "loss": 0.904, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.6773134981404505, | |
| "grad_norm": 1.2693063020706177, | |
| "learning_rate": 3.437280486858996e-07, | |
| "loss": 0.8813, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 2.686064318529862, | |
| "grad_norm": 1.3009639978408813, | |
| "learning_rate": 3.2540538736364977e-07, | |
| "loss": 0.8851, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 2.6948151389192736, | |
| "grad_norm": 1.2336896657943726, | |
| "learning_rate": 3.075681183520407e-07, | |
| "loss": 0.8994, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 2.703565959308685, | |
| "grad_norm": 1.2099403142929077, | |
| "learning_rate": 2.9021809380595066e-07, | |
| "loss": 0.899, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 2.7123167796980967, | |
| "grad_norm": 1.1838281154632568, | |
| "learning_rate": 2.7335711528663047e-07, | |
| "loss": 0.8951, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.7210676000875083, | |
| "grad_norm": 1.2549548149108887, | |
| "learning_rate": 2.569869335746267e-07, | |
| "loss": 0.8795, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 2.72981842047692, | |
| "grad_norm": 1.189772605895996, | |
| "learning_rate": 2.411092484879957e-07, | |
| "loss": 0.8632, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 2.7385692408663314, | |
| "grad_norm": 1.4026386737823486, | |
| "learning_rate": 2.2572570870579845e-07, | |
| "loss": 0.905, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 2.747320061255743, | |
| "grad_norm": 1.2943899631500244, | |
| "learning_rate": 2.1083791159690615e-07, | |
| "loss": 0.8858, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 2.7560708816451545, | |
| "grad_norm": 1.3465709686279297, | |
| "learning_rate": 1.9644740305413636e-07, | |
| "loss": 0.8963, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.764821702034566, | |
| "grad_norm": 1.2346800565719604, | |
| "learning_rate": 1.8255567733373392e-07, | |
| "loss": 0.8928, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 2.773572522423977, | |
| "grad_norm": 1.255714774131775, | |
| "learning_rate": 1.691641769002139e-07, | |
| "loss": 0.8977, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 2.7823233428133887, | |
| "grad_norm": 1.3175842761993408, | |
| "learning_rate": 1.5627429227657864e-07, | |
| "loss": 0.9226, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 2.7910741632028, | |
| "grad_norm": 1.333662748336792, | |
| "learning_rate": 1.4388736189993257e-07, | |
| "loss": 0.8932, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 2.7998249835922118, | |
| "grad_norm": 1.2100019454956055, | |
| "learning_rate": 1.320046719825069e-07, | |
| "loss": 0.8886, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.8085758039816233, | |
| "grad_norm": 1.2037444114685059, | |
| "learning_rate": 1.2062745637809737e-07, | |
| "loss": 0.8912, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 2.817326624371035, | |
| "grad_norm": 1.2076711654663086, | |
| "learning_rate": 1.0975689645395193e-07, | |
| "loss": 0.8627, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 2.8260774447604464, | |
| "grad_norm": 1.2997610569000244, | |
| "learning_rate": 9.939412096809986e-08, | |
| "loss": 0.899, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 2.8348282651498575, | |
| "grad_norm": 1.171425461769104, | |
| "learning_rate": 8.954020595214441e-08, | |
| "loss": 0.9032, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 2.843579085539269, | |
| "grad_norm": 1.2966265678405762, | |
| "learning_rate": 8.019617459953333e-08, | |
| "loss": 0.8888, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.8523299059286806, | |
| "grad_norm": 1.27556574344635, | |
| "learning_rate": 7.136299715931438e-08, | |
| "loss": 0.8775, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 2.861080726318092, | |
| "grad_norm": 1.2437477111816406, | |
| "learning_rate": 6.304159083538652e-08, | |
| "loss": 0.8696, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 2.8698315467075037, | |
| "grad_norm": 1.2502518892288208, | |
| "learning_rate": 5.523281969126271e-08, | |
| "loss": 0.8915, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 2.8785823670969153, | |
| "grad_norm": 1.250112771987915, | |
| "learning_rate": 4.7937494560349464e-08, | |
| "loss": 0.8792, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 2.887333187486327, | |
| "grad_norm": 1.2390645742416382, | |
| "learning_rate": 4.115637296174979e-08, | |
| "loss": 0.8995, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.8960840078757384, | |
| "grad_norm": 1.2656710147857666, | |
| "learning_rate": 3.4890159021608305e-08, | |
| "loss": 0.8741, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 2.90483482826515, | |
| "grad_norm": 1.149044156074524, | |
| "learning_rate": 2.9139503399996382e-08, | |
| "loss": 0.8869, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 2.9135856486545615, | |
| "grad_norm": 1.2886285781860352, | |
| "learning_rate": 2.3905003223349544e-08, | |
| "loss": 0.8875, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 2.922336469043973, | |
| "grad_norm": 1.2327951192855835, | |
| "learning_rate": 1.9187202022464822e-08, | |
| "loss": 0.8899, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 2.9310872894333846, | |
| "grad_norm": 1.3543472290039062, | |
| "learning_rate": 1.4986589676063147e-08, | |
| "loss": 0.8787, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.939838109822796, | |
| "grad_norm": 1.3077126741409302, | |
| "learning_rate": 1.1303602359919452e-08, | |
| "loss": 0.9173, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 2.9485889302122072, | |
| "grad_norm": 1.254578948020935, | |
| "learning_rate": 8.1386225015756e-09, | |
| "loss": 0.8375, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 2.957339750601619, | |
| "grad_norm": 1.3026719093322754, | |
| "learning_rate": 5.491978740627679e-09, | |
| "loss": 0.8827, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 2.9660905709910304, | |
| "grad_norm": 1.2489607334136963, | |
| "learning_rate": 3.3639458946027603e-09, | |
| "loss": 0.8835, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 2.974841391380442, | |
| "grad_norm": 1.264045000076294, | |
| "learning_rate": 1.7547449304222696e-09, | |
| "loss": 0.8905, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.9835922117698535, | |
| "grad_norm": 1.2398723363876343, | |
| "learning_rate": 6.645429414570137e-10, | |
| "loss": 0.8854, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 2.992343032159265, | |
| "grad_norm": 1.2165299654006958, | |
| "learning_rate": 9.345313017938839e-11, | |
| "loss": 0.8911, | |
| "step": 3420 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 3426, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500.0, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.3232347449707725e+17, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |