{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 55.55555555555556, "eval_steps": 500, "global_step": 20000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.027777777777777776, "grad_norm": 3.59006667137146, "learning_rate": 3.0000000000000004e-07, "loss": 1.2083, "step": 10 }, { "epoch": 0.05555555555555555, "grad_norm": 2.1800248622894287, "learning_rate": 6.333333333333333e-07, "loss": 1.1089, "step": 20 }, { "epoch": 0.08333333333333333, "grad_norm": 2.553022861480713, "learning_rate": 9.666666666666668e-07, "loss": 1.1052, "step": 30 }, { "epoch": 0.1111111111111111, "grad_norm": 2.1491830348968506, "learning_rate": 1.3e-06, "loss": 1.0947, "step": 40 }, { "epoch": 0.1388888888888889, "grad_norm": 1.3348098993301392, "learning_rate": 1.6333333333333333e-06, "loss": 1.0842, "step": 50 }, { "epoch": 0.16666666666666666, "grad_norm": 1.8711786270141602, "learning_rate": 1.9666666666666668e-06, "loss": 1.0763, "step": 60 }, { "epoch": 0.19444444444444445, "grad_norm": 1.2430402040481567, "learning_rate": 2.3e-06, "loss": 1.0739, "step": 70 }, { "epoch": 0.2222222222222222, "grad_norm": 1.5116122961044312, "learning_rate": 2.6333333333333337e-06, "loss": 1.0647, "step": 80 }, { "epoch": 0.25, "grad_norm": 1.6797914505004883, "learning_rate": 2.966666666666667e-06, "loss": 1.0751, "step": 90 }, { "epoch": 0.2777777777777778, "grad_norm": 1.9885735511779785, "learning_rate": 3.3e-06, "loss": 1.0687, "step": 100 }, { "epoch": 0.3055555555555556, "grad_norm": 1.9800221920013428, "learning_rate": 3.633333333333334e-06, "loss": 1.0519, "step": 110 }, { "epoch": 0.3333333333333333, "grad_norm": 1.7728772163391113, "learning_rate": 3.966666666666667e-06, "loss": 1.0462, "step": 120 }, { "epoch": 0.3611111111111111, "grad_norm": 1.4425629377365112, "learning_rate": 4.2999999999999995e-06, "loss": 1.0251, "step": 130 }, { "epoch": 0.3888888888888889, "grad_norm": 1.7924060821533203, "learning_rate": 4.633333333333334e-06, "loss": 1.0207, "step": 140 }, { "epoch": 0.4166666666666667, "grad_norm": 1.4595309495925903, "learning_rate": 4.966666666666667e-06, "loss": 1.0306, "step": 150 }, { "epoch": 0.4444444444444444, "grad_norm": 1.9495493173599243, "learning_rate": 5.3e-06, "loss": 1.0254, "step": 160 }, { "epoch": 0.4722222222222222, "grad_norm": 2.152066469192505, "learning_rate": 5.633333333333333e-06, "loss": 1.0155, "step": 170 }, { "epoch": 0.5, "grad_norm": 1.3608466386795044, "learning_rate": 5.9666666666666666e-06, "loss": 0.9986, "step": 180 }, { "epoch": 0.5277777777777778, "grad_norm": 1.9830243587493896, "learning_rate": 6.300000000000001e-06, "loss": 0.9964, "step": 190 }, { "epoch": 0.5555555555555556, "grad_norm": 1.215932011604309, "learning_rate": 6.633333333333333e-06, "loss": 0.9851, "step": 200 }, { "epoch": 0.5833333333333334, "grad_norm": 1.6745388507843018, "learning_rate": 6.966666666666667e-06, "loss": 0.982, "step": 210 }, { "epoch": 0.6111111111111112, "grad_norm": 1.904631495475769, "learning_rate": 7.2999999999999996e-06, "loss": 0.9702, "step": 220 }, { "epoch": 0.6388888888888888, "grad_norm": 1.1576166152954102, "learning_rate": 7.633333333333334e-06, "loss": 0.9707, "step": 230 }, { "epoch": 0.6666666666666666, "grad_norm": 2.0446488857269287, "learning_rate": 7.966666666666666e-06, "loss": 0.9564, "step": 240 }, { "epoch": 0.6944444444444444, "grad_norm": 2.2964532375335693, "learning_rate": 8.3e-06, "loss": 0.9574, "step": 250 }, { "epoch": 0.7222222222222222, "grad_norm": 1.6763256788253784, "learning_rate": 8.633333333333334e-06, "loss": 0.9376, "step": 260 }, { "epoch": 0.75, "grad_norm": 2.8731932640075684, "learning_rate": 8.966666666666668e-06, "loss": 0.8871, "step": 270 }, { "epoch": 0.7777777777777778, "grad_norm": 3.743459463119507, "learning_rate": 9.3e-06, "loss": 0.8861, "step": 280 }, { "epoch": 0.8055555555555556, "grad_norm": 8.091812133789062, "learning_rate": 9.633333333333335e-06, "loss": 0.7753, "step": 290 }, { "epoch": 0.8333333333333334, "grad_norm": 4.634387493133545, "learning_rate": 9.966666666666667e-06, "loss": 0.7883, "step": 300 }, { "epoch": 0.8611111111111112, "grad_norm": 5.9085845947265625, "learning_rate": 1.03e-05, "loss": 0.7242, "step": 310 }, { "epoch": 0.8888888888888888, "grad_norm": 11.413772583007812, "learning_rate": 1.0633333333333334e-05, "loss": 0.6172, "step": 320 }, { "epoch": 0.9166666666666666, "grad_norm": 6.14446496963501, "learning_rate": 1.0966666666666666e-05, "loss": 0.6092, "step": 330 }, { "epoch": 0.9444444444444444, "grad_norm": 8.040328979492188, "learning_rate": 1.13e-05, "loss": 0.501, "step": 340 }, { "epoch": 0.9722222222222222, "grad_norm": 7.827601909637451, "learning_rate": 1.1633333333333334e-05, "loss": 0.5132, "step": 350 }, { "epoch": 1.0, "grad_norm": 5.841759204864502, "learning_rate": 1.1966666666666668e-05, "loss": 0.4441, "step": 360 }, { "epoch": 1.0277777777777777, "grad_norm": 9.356619834899902, "learning_rate": 1.23e-05, "loss": 0.4025, "step": 370 }, { "epoch": 1.0555555555555556, "grad_norm": 5.747879505157471, "learning_rate": 1.2633333333333333e-05, "loss": 0.3777, "step": 380 }, { "epoch": 1.0833333333333333, "grad_norm": 3.9728035926818848, "learning_rate": 1.2966666666666669e-05, "loss": 0.3962, "step": 390 }, { "epoch": 1.1111111111111112, "grad_norm": 6.551274299621582, "learning_rate": 1.3300000000000001e-05, "loss": 0.3038, "step": 400 }, { "epoch": 1.1388888888888888, "grad_norm": 7.673063278198242, "learning_rate": 1.3633333333333334e-05, "loss": 0.3287, "step": 410 }, { "epoch": 1.1666666666666667, "grad_norm": 5.012998104095459, "learning_rate": 1.3966666666666666e-05, "loss": 0.3205, "step": 420 }, { "epoch": 1.1944444444444444, "grad_norm": 10.47907829284668, "learning_rate": 1.43e-05, "loss": 0.2884, "step": 430 }, { "epoch": 1.2222222222222223, "grad_norm": 6.80259895324707, "learning_rate": 1.4633333333333334e-05, "loss": 0.3059, "step": 440 }, { "epoch": 1.25, "grad_norm": 5.941986083984375, "learning_rate": 1.4966666666666668e-05, "loss": 0.2714, "step": 450 }, { "epoch": 1.2777777777777777, "grad_norm": 2.913792133331299, "learning_rate": 1.53e-05, "loss": 0.2255, "step": 460 }, { "epoch": 1.3055555555555556, "grad_norm": 3.0305910110473633, "learning_rate": 1.563333333333333e-05, "loss": 0.2198, "step": 470 }, { "epoch": 1.3333333333333333, "grad_norm": 4.068256378173828, "learning_rate": 1.5966666666666667e-05, "loss": 0.1895, "step": 480 }, { "epoch": 1.3611111111111112, "grad_norm": 4.522433280944824, "learning_rate": 1.63e-05, "loss": 0.2061, "step": 490 }, { "epoch": 1.3888888888888888, "grad_norm": 3.094316244125366, "learning_rate": 1.6633333333333336e-05, "loss": 0.1516, "step": 500 }, { "epoch": 1.4166666666666667, "grad_norm": 2.77016282081604, "learning_rate": 1.6966666666666668e-05, "loss": 0.162, "step": 510 }, { "epoch": 1.4444444444444444, "grad_norm": 4.687958240509033, "learning_rate": 1.73e-05, "loss": 0.1433, "step": 520 }, { "epoch": 1.4722222222222223, "grad_norm": 2.3259494304656982, "learning_rate": 1.7633333333333336e-05, "loss": 0.1662, "step": 530 }, { "epoch": 1.5, "grad_norm": 2.9830076694488525, "learning_rate": 1.796666666666667e-05, "loss": 0.1855, "step": 540 }, { "epoch": 1.5277777777777777, "grad_norm": 3.5648300647735596, "learning_rate": 1.83e-05, "loss": 0.1375, "step": 550 }, { "epoch": 1.5555555555555556, "grad_norm": 1.300215244293213, "learning_rate": 1.8633333333333333e-05, "loss": 0.1465, "step": 560 }, { "epoch": 1.5833333333333335, "grad_norm": 3.337721586227417, "learning_rate": 1.896666666666667e-05, "loss": 0.1521, "step": 570 }, { "epoch": 1.6111111111111112, "grad_norm": 1.9488636255264282, "learning_rate": 1.93e-05, "loss": 0.1404, "step": 580 }, { "epoch": 1.6388888888888888, "grad_norm": 2.3286383152008057, "learning_rate": 1.9633333333333334e-05, "loss": 0.0969, "step": 590 }, { "epoch": 1.6666666666666665, "grad_norm": 2.8948934078216553, "learning_rate": 1.9966666666666666e-05, "loss": 0.1082, "step": 600 }, { "epoch": 1.6944444444444444, "grad_norm": 2.5086934566497803, "learning_rate": 2.0300000000000002e-05, "loss": 0.125, "step": 610 }, { "epoch": 1.7222222222222223, "grad_norm": 2.498093366622925, "learning_rate": 2.0633333333333335e-05, "loss": 0.0965, "step": 620 }, { "epoch": 1.75, "grad_norm": 2.0224499702453613, "learning_rate": 2.0966666666666667e-05, "loss": 0.0779, "step": 630 }, { "epoch": 1.7777777777777777, "grad_norm": 2.556365966796875, "learning_rate": 2.13e-05, "loss": 0.069, "step": 640 }, { "epoch": 1.8055555555555556, "grad_norm": 2.6082143783569336, "learning_rate": 2.1633333333333332e-05, "loss": 0.0812, "step": 650 }, { "epoch": 1.8333333333333335, "grad_norm": 2.0550155639648438, "learning_rate": 2.1966666666666668e-05, "loss": 0.0728, "step": 660 }, { "epoch": 1.8611111111111112, "grad_norm": 1.6575927734375, "learning_rate": 2.23e-05, "loss": 0.017, "step": 670 }, { "epoch": 1.8888888888888888, "grad_norm": 2.869657516479492, "learning_rate": 2.2633333333333336e-05, "loss": 0.0522, "step": 680 }, { "epoch": 1.9166666666666665, "grad_norm": 2.900102138519287, "learning_rate": 2.2966666666666668e-05, "loss": 0.0304, "step": 690 }, { "epoch": 1.9444444444444444, "grad_norm": 2.6903531551361084, "learning_rate": 2.3300000000000004e-05, "loss": 0.0381, "step": 700 }, { "epoch": 1.9722222222222223, "grad_norm": 2.07124400138855, "learning_rate": 2.3633333333333336e-05, "loss": 0.0336, "step": 710 }, { "epoch": 2.0, "grad_norm": 2.217904806137085, "learning_rate": 2.396666666666667e-05, "loss": 0.0178, "step": 720 }, { "epoch": 2.0277777777777777, "grad_norm": 1.9478322267532349, "learning_rate": 2.43e-05, "loss": 0.0446, "step": 730 }, { "epoch": 2.0555555555555554, "grad_norm": 1.422187089920044, "learning_rate": 2.4633333333333334e-05, "loss": 0.0149, "step": 740 }, { "epoch": 2.0833333333333335, "grad_norm": 1.93894624710083, "learning_rate": 2.496666666666667e-05, "loss": 0.0187, "step": 750 }, { "epoch": 2.111111111111111, "grad_norm": 2.6775286197662354, "learning_rate": 2.5300000000000002e-05, "loss": 0.0157, "step": 760 }, { "epoch": 2.138888888888889, "grad_norm": 2.170353889465332, "learning_rate": 2.5633333333333338e-05, "loss": 0.0123, "step": 770 }, { "epoch": 2.1666666666666665, "grad_norm": 1.964930534362793, "learning_rate": 2.5966666666666667e-05, "loss": 0.0262, "step": 780 }, { "epoch": 2.1944444444444446, "grad_norm": 2.2398481369018555, "learning_rate": 2.6300000000000002e-05, "loss": -0.0064, "step": 790 }, { "epoch": 2.2222222222222223, "grad_norm": 2.082385778427124, "learning_rate": 2.663333333333333e-05, "loss": 0.0207, "step": 800 }, { "epoch": 2.25, "grad_norm": 1.4442280530929565, "learning_rate": 2.6966666666666667e-05, "loss": 0.0008, "step": 810 }, { "epoch": 2.2777777777777777, "grad_norm": 2.1997814178466797, "learning_rate": 2.7300000000000003e-05, "loss": 0.0231, "step": 820 }, { "epoch": 2.3055555555555554, "grad_norm": 1.4828358888626099, "learning_rate": 2.7633333333333332e-05, "loss": 0.0045, "step": 830 }, { "epoch": 2.3333333333333335, "grad_norm": 1.487001657485962, "learning_rate": 2.7966666666666668e-05, "loss": -0.0163, "step": 840 }, { "epoch": 2.361111111111111, "grad_norm": 1.8799312114715576, "learning_rate": 2.83e-05, "loss": -0.015, "step": 850 }, { "epoch": 2.388888888888889, "grad_norm": 1.8355900049209595, "learning_rate": 2.8633333333333336e-05, "loss": 0.0045, "step": 860 }, { "epoch": 2.4166666666666665, "grad_norm": 1.9812779426574707, "learning_rate": 2.8966666666666668e-05, "loss": -0.0056, "step": 870 }, { "epoch": 2.4444444444444446, "grad_norm": 1.3251404762268066, "learning_rate": 2.93e-05, "loss": -0.0077, "step": 880 }, { "epoch": 2.4722222222222223, "grad_norm": 2.347377061843872, "learning_rate": 2.9633333333333336e-05, "loss": 0.0002, "step": 890 }, { "epoch": 2.5, "grad_norm": 2.022001266479492, "learning_rate": 2.9966666666666672e-05, "loss": -0.0011, "step": 900 }, { "epoch": 2.5277777777777777, "grad_norm": 1.3376007080078125, "learning_rate": 3.03e-05, "loss": 0.0003, "step": 910 }, { "epoch": 2.5555555555555554, "grad_norm": 1.476718783378601, "learning_rate": 3.063333333333334e-05, "loss": -0.0042, "step": 920 }, { "epoch": 2.5833333333333335, "grad_norm": 1.6696317195892334, "learning_rate": 3.096666666666666e-05, "loss": 0.0144, "step": 930 }, { "epoch": 2.611111111111111, "grad_norm": 1.8459622859954834, "learning_rate": 3.13e-05, "loss": -0.0344, "step": 940 }, { "epoch": 2.638888888888889, "grad_norm": 1.2552207708358765, "learning_rate": 3.1633333333333334e-05, "loss": -0.0223, "step": 950 }, { "epoch": 2.6666666666666665, "grad_norm": 2.417234420776367, "learning_rate": 3.196666666666667e-05, "loss": -0.008, "step": 960 }, { "epoch": 2.6944444444444446, "grad_norm": 1.226636528968811, "learning_rate": 3.2300000000000006e-05, "loss": -0.0143, "step": 970 }, { "epoch": 2.7222222222222223, "grad_norm": 1.2197294235229492, "learning_rate": 3.263333333333333e-05, "loss": -0.0231, "step": 980 }, { "epoch": 2.75, "grad_norm": 1.7816338539123535, "learning_rate": 3.296666666666667e-05, "loss": -0.0332, "step": 990 }, { "epoch": 2.7777777777777777, "grad_norm": 1.9632608890533447, "learning_rate": 3.33e-05, "loss": -0.025, "step": 1000 }, { "epoch": 2.8055555555555554, "grad_norm": 1.5374140739440918, "learning_rate": 3.3633333333333335e-05, "loss": -0.0047, "step": 1010 }, { "epoch": 2.8333333333333335, "grad_norm": 1.6534368991851807, "learning_rate": 3.396666666666667e-05, "loss": -0.0289, "step": 1020 }, { "epoch": 2.861111111111111, "grad_norm": 1.0268762111663818, "learning_rate": 3.430000000000001e-05, "loss": -0.0239, "step": 1030 }, { "epoch": 2.888888888888889, "grad_norm": 1.093058705329895, "learning_rate": 3.463333333333333e-05, "loss": -0.0401, "step": 1040 }, { "epoch": 2.9166666666666665, "grad_norm": 1.7956583499908447, "learning_rate": 3.496666666666667e-05, "loss": -0.0253, "step": 1050 }, { "epoch": 2.9444444444444446, "grad_norm": 1.7291741371154785, "learning_rate": 3.53e-05, "loss": -0.0151, "step": 1060 }, { "epoch": 2.9722222222222223, "grad_norm": 1.2337664365768433, "learning_rate": 3.563333333333334e-05, "loss": -0.0444, "step": 1070 }, { "epoch": 3.0, "grad_norm": 1.7078052759170532, "learning_rate": 3.596666666666667e-05, "loss": -0.0174, "step": 1080 }, { "epoch": 3.0277777777777777, "grad_norm": 1.571593999862671, "learning_rate": 3.63e-05, "loss": -0.0247, "step": 1090 }, { "epoch": 3.0555555555555554, "grad_norm": 1.5376014709472656, "learning_rate": 3.6633333333333334e-05, "loss": -0.0389, "step": 1100 }, { "epoch": 3.0833333333333335, "grad_norm": 1.6260725259780884, "learning_rate": 3.6966666666666666e-05, "loss": -0.0313, "step": 1110 }, { "epoch": 3.111111111111111, "grad_norm": 1.4571807384490967, "learning_rate": 3.73e-05, "loss": -0.05, "step": 1120 }, { "epoch": 3.138888888888889, "grad_norm": 1.2193491458892822, "learning_rate": 3.763333333333334e-05, "loss": -0.0326, "step": 1130 }, { "epoch": 3.1666666666666665, "grad_norm": 1.2061558961868286, "learning_rate": 3.796666666666667e-05, "loss": -0.0437, "step": 1140 }, { "epoch": 3.1944444444444446, "grad_norm": 1.6744608879089355, "learning_rate": 3.83e-05, "loss": -0.033, "step": 1150 }, { "epoch": 3.2222222222222223, "grad_norm": 1.3061261177062988, "learning_rate": 3.8633333333333335e-05, "loss": -0.0444, "step": 1160 }, { "epoch": 3.25, "grad_norm": 1.491563081741333, "learning_rate": 3.896666666666667e-05, "loss": -0.0315, "step": 1170 }, { "epoch": 3.2777777777777777, "grad_norm": 1.4821584224700928, "learning_rate": 3.9300000000000007e-05, "loss": -0.0469, "step": 1180 }, { "epoch": 3.3055555555555554, "grad_norm": 1.3603414297103882, "learning_rate": 3.963333333333333e-05, "loss": -0.0335, "step": 1190 }, { "epoch": 3.3333333333333335, "grad_norm": 1.196789026260376, "learning_rate": 3.996666666666667e-05, "loss": -0.0267, "step": 1200 }, { "epoch": 3.361111111111111, "grad_norm": 1.1366238594055176, "learning_rate": 4.0300000000000004e-05, "loss": -0.0222, "step": 1210 }, { "epoch": 3.388888888888889, "grad_norm": 1.4032021760940552, "learning_rate": 4.0633333333333336e-05, "loss": -0.033, "step": 1220 }, { "epoch": 3.4166666666666665, "grad_norm": 0.8007150888442993, "learning_rate": 4.096666666666667e-05, "loss": -0.0504, "step": 1230 }, { "epoch": 3.4444444444444446, "grad_norm": 1.5290610790252686, "learning_rate": 4.13e-05, "loss": -0.0323, "step": 1240 }, { "epoch": 3.4722222222222223, "grad_norm": 2.4610958099365234, "learning_rate": 4.1633333333333333e-05, "loss": -0.0428, "step": 1250 }, { "epoch": 3.5, "grad_norm": 1.0869654417037964, "learning_rate": 4.196666666666667e-05, "loss": -0.0525, "step": 1260 }, { "epoch": 3.5277777777777777, "grad_norm": 1.712296962738037, "learning_rate": 4.23e-05, "loss": -0.055, "step": 1270 }, { "epoch": 3.5555555555555554, "grad_norm": 1.252585530281067, "learning_rate": 4.263333333333334e-05, "loss": -0.0346, "step": 1280 }, { "epoch": 3.5833333333333335, "grad_norm": 1.5438263416290283, "learning_rate": 4.296666666666666e-05, "loss": -0.0422, "step": 1290 }, { "epoch": 3.611111111111111, "grad_norm": 1.658010482788086, "learning_rate": 4.33e-05, "loss": -0.0388, "step": 1300 }, { "epoch": 3.638888888888889, "grad_norm": 1.7152831554412842, "learning_rate": 4.3633333333333335e-05, "loss": -0.0464, "step": 1310 }, { "epoch": 3.6666666666666665, "grad_norm": 1.8405520915985107, "learning_rate": 4.396666666666667e-05, "loss": -0.0395, "step": 1320 }, { "epoch": 3.6944444444444446, "grad_norm": 1.282498836517334, "learning_rate": 4.43e-05, "loss": -0.028, "step": 1330 }, { "epoch": 3.7222222222222223, "grad_norm": 1.3419854640960693, "learning_rate": 4.463333333333334e-05, "loss": -0.0269, "step": 1340 }, { "epoch": 3.75, "grad_norm": 1.4038687944412231, "learning_rate": 4.496666666666667e-05, "loss": -0.0499, "step": 1350 }, { "epoch": 3.7777777777777777, "grad_norm": 1.2779887914657593, "learning_rate": 4.53e-05, "loss": -0.0289, "step": 1360 }, { "epoch": 3.8055555555555554, "grad_norm": 1.7231802940368652, "learning_rate": 4.5633333333333336e-05, "loss": -0.0523, "step": 1370 }, { "epoch": 3.8333333333333335, "grad_norm": 1.2351782321929932, "learning_rate": 4.596666666666667e-05, "loss": -0.0372, "step": 1380 }, { "epoch": 3.861111111111111, "grad_norm": 1.0968230962753296, "learning_rate": 4.630000000000001e-05, "loss": -0.0431, "step": 1390 }, { "epoch": 3.888888888888889, "grad_norm": 0.7615843415260315, "learning_rate": 4.663333333333333e-05, "loss": -0.056, "step": 1400 }, { "epoch": 3.9166666666666665, "grad_norm": 1.0493243932724, "learning_rate": 4.696666666666667e-05, "loss": -0.0138, "step": 1410 }, { "epoch": 3.9444444444444446, "grad_norm": 1.298606514930725, "learning_rate": 4.73e-05, "loss": -0.0436, "step": 1420 }, { "epoch": 3.9722222222222223, "grad_norm": 1.3641277551651, "learning_rate": 4.763333333333334e-05, "loss": -0.0385, "step": 1430 }, { "epoch": 4.0, "grad_norm": 1.3234436511993408, "learning_rate": 4.796666666666667e-05, "loss": -0.0642, "step": 1440 }, { "epoch": 4.027777777777778, "grad_norm": 0.7827696204185486, "learning_rate": 4.83e-05, "loss": -0.0452, "step": 1450 }, { "epoch": 4.055555555555555, "grad_norm": 1.0622494220733643, "learning_rate": 4.8633333333333334e-05, "loss": -0.0505, "step": 1460 }, { "epoch": 4.083333333333333, "grad_norm": 1.3039312362670898, "learning_rate": 4.8966666666666667e-05, "loss": -0.0673, "step": 1470 }, { "epoch": 4.111111111111111, "grad_norm": 1.0623434782028198, "learning_rate": 4.93e-05, "loss": -0.0591, "step": 1480 }, { "epoch": 4.138888888888889, "grad_norm": 1.300879716873169, "learning_rate": 4.963333333333334e-05, "loss": -0.0551, "step": 1490 }, { "epoch": 4.166666666666667, "grad_norm": 1.308798909187317, "learning_rate": 4.996666666666667e-05, "loss": -0.046, "step": 1500 }, { "epoch": 4.194444444444445, "grad_norm": 1.0091146230697632, "learning_rate": 5.03e-05, "loss": -0.07, "step": 1510 }, { "epoch": 4.222222222222222, "grad_norm": 1.5920844078063965, "learning_rate": 5.0633333333333335e-05, "loss": -0.0538, "step": 1520 }, { "epoch": 4.25, "grad_norm": 1.2166138887405396, "learning_rate": 5.0966666666666674e-05, "loss": -0.0456, "step": 1530 }, { "epoch": 4.277777777777778, "grad_norm": 1.1909031867980957, "learning_rate": 5.130000000000001e-05, "loss": -0.0616, "step": 1540 }, { "epoch": 4.305555555555555, "grad_norm": 1.2878278493881226, "learning_rate": 5.163333333333333e-05, "loss": -0.0413, "step": 1550 }, { "epoch": 4.333333333333333, "grad_norm": 0.8010730743408203, "learning_rate": 5.196666666666667e-05, "loss": -0.0704, "step": 1560 }, { "epoch": 4.361111111111111, "grad_norm": 0.7422463893890381, "learning_rate": 5.2300000000000004e-05, "loss": -0.0605, "step": 1570 }, { "epoch": 4.388888888888889, "grad_norm": 1.2650811672210693, "learning_rate": 5.2633333333333336e-05, "loss": -0.058, "step": 1580 }, { "epoch": 4.416666666666667, "grad_norm": 1.2036734819412231, "learning_rate": 5.296666666666666e-05, "loss": -0.0532, "step": 1590 }, { "epoch": 4.444444444444445, "grad_norm": 1.100925087928772, "learning_rate": 5.330000000000001e-05, "loss": -0.0488, "step": 1600 }, { "epoch": 4.472222222222222, "grad_norm": 0.8113247752189636, "learning_rate": 5.3633333333333334e-05, "loss": -0.0394, "step": 1610 }, { "epoch": 4.5, "grad_norm": 1.0056072473526, "learning_rate": 5.3966666666666666e-05, "loss": -0.0505, "step": 1620 }, { "epoch": 4.527777777777778, "grad_norm": 1.6927176713943481, "learning_rate": 5.4300000000000005e-05, "loss": -0.0401, "step": 1630 }, { "epoch": 4.555555555555555, "grad_norm": 1.207577109336853, "learning_rate": 5.463333333333334e-05, "loss": -0.0594, "step": 1640 }, { "epoch": 4.583333333333333, "grad_norm": 0.5374363660812378, "learning_rate": 5.496666666666666e-05, "loss": -0.0639, "step": 1650 }, { "epoch": 4.611111111111111, "grad_norm": 1.5611162185668945, "learning_rate": 5.530000000000001e-05, "loss": -0.0453, "step": 1660 }, { "epoch": 4.638888888888889, "grad_norm": 1.2640613317489624, "learning_rate": 5.5633333333333335e-05, "loss": -0.058, "step": 1670 }, { "epoch": 4.666666666666667, "grad_norm": 0.9046183228492737, "learning_rate": 5.596666666666667e-05, "loss": -0.0724, "step": 1680 }, { "epoch": 4.694444444444445, "grad_norm": 1.4021344184875488, "learning_rate": 5.63e-05, "loss": -0.0357, "step": 1690 }, { "epoch": 4.722222222222222, "grad_norm": 0.9998553991317749, "learning_rate": 5.663333333333334e-05, "loss": -0.0577, "step": 1700 }, { "epoch": 4.75, "grad_norm": 0.8816908001899719, "learning_rate": 5.696666666666667e-05, "loss": -0.0662, "step": 1710 }, { "epoch": 4.777777777777778, "grad_norm": 1.0441832542419434, "learning_rate": 5.73e-05, "loss": -0.0461, "step": 1720 }, { "epoch": 4.805555555555555, "grad_norm": 1.0150343179702759, "learning_rate": 5.7633333333333336e-05, "loss": -0.068, "step": 1730 }, { "epoch": 4.833333333333333, "grad_norm": 1.064945936203003, "learning_rate": 5.796666666666667e-05, "loss": -0.0503, "step": 1740 }, { "epoch": 4.861111111111111, "grad_norm": 0.9658017754554749, "learning_rate": 5.83e-05, "loss": -0.0694, "step": 1750 }, { "epoch": 4.888888888888889, "grad_norm": 0.7725010514259338, "learning_rate": 5.863333333333334e-05, "loss": -0.0687, "step": 1760 }, { "epoch": 4.916666666666667, "grad_norm": 0.7685675024986267, "learning_rate": 5.896666666666667e-05, "loss": -0.0652, "step": 1770 }, { "epoch": 4.944444444444445, "grad_norm": 1.1630629301071167, "learning_rate": 5.93e-05, "loss": -0.0794, "step": 1780 }, { "epoch": 4.972222222222222, "grad_norm": 1.0213478803634644, "learning_rate": 5.9633333333333344e-05, "loss": -0.0514, "step": 1790 }, { "epoch": 5.0, "grad_norm": 0.8022200465202332, "learning_rate": 5.996666666666667e-05, "loss": -0.0594, "step": 1800 }, { "epoch": 5.027777777777778, "grad_norm": 0.7570667266845703, "learning_rate": 6.03e-05, "loss": -0.0645, "step": 1810 }, { "epoch": 5.055555555555555, "grad_norm": 1.3488258123397827, "learning_rate": 6.063333333333333e-05, "loss": -0.0704, "step": 1820 }, { "epoch": 5.083333333333333, "grad_norm": 1.2108361721038818, "learning_rate": 6.0966666666666674e-05, "loss": -0.0583, "step": 1830 }, { "epoch": 5.111111111111111, "grad_norm": 0.9981511235237122, "learning_rate": 6.13e-05, "loss": -0.0601, "step": 1840 }, { "epoch": 5.138888888888889, "grad_norm": 0.8483412265777588, "learning_rate": 6.163333333333333e-05, "loss": -0.0655, "step": 1850 }, { "epoch": 5.166666666666667, "grad_norm": 1.1261223554611206, "learning_rate": 6.196666666666668e-05, "loss": -0.0605, "step": 1860 }, { "epoch": 5.194444444444445, "grad_norm": 0.6597011685371399, "learning_rate": 6.23e-05, "loss": -0.0755, "step": 1870 }, { "epoch": 5.222222222222222, "grad_norm": 0.9884948134422302, "learning_rate": 6.263333333333333e-05, "loss": -0.0613, "step": 1880 }, { "epoch": 5.25, "grad_norm": 1.3865379095077515, "learning_rate": 6.296666666666667e-05, "loss": -0.0471, "step": 1890 }, { "epoch": 5.277777777777778, "grad_norm": 0.9377394318580627, "learning_rate": 6.330000000000001e-05, "loss": -0.0598, "step": 1900 }, { "epoch": 5.305555555555555, "grad_norm": 0.9896470904350281, "learning_rate": 6.363333333333334e-05, "loss": -0.0704, "step": 1910 }, { "epoch": 5.333333333333333, "grad_norm": 0.8680184483528137, "learning_rate": 6.396666666666667e-05, "loss": -0.0832, "step": 1920 }, { "epoch": 5.361111111111111, "grad_norm": 0.7604486346244812, "learning_rate": 6.43e-05, "loss": -0.0634, "step": 1930 }, { "epoch": 5.388888888888889, "grad_norm": 0.8691123723983765, "learning_rate": 6.463333333333334e-05, "loss": -0.0822, "step": 1940 }, { "epoch": 5.416666666666667, "grad_norm": 1.1164014339447021, "learning_rate": 6.496666666666667e-05, "loss": -0.059, "step": 1950 }, { "epoch": 5.444444444444445, "grad_norm": 0.9757416844367981, "learning_rate": 6.53e-05, "loss": -0.0611, "step": 1960 }, { "epoch": 5.472222222222222, "grad_norm": 1.3875832557678223, "learning_rate": 6.563333333333333e-05, "loss": -0.06, "step": 1970 }, { "epoch": 5.5, "grad_norm": 0.6562498807907104, "learning_rate": 6.596666666666667e-05, "loss": -0.0614, "step": 1980 }, { "epoch": 5.527777777777778, "grad_norm": 0.8936523199081421, "learning_rate": 6.630000000000001e-05, "loss": -0.0636, "step": 1990 }, { "epoch": 5.555555555555555, "grad_norm": 0.8032233715057373, "learning_rate": 6.663333333333333e-05, "loss": -0.0724, "step": 2000 }, { "epoch": 5.583333333333333, "grad_norm": 0.9884873032569885, "learning_rate": 6.696666666666666e-05, "loss": -0.074, "step": 2010 }, { "epoch": 5.611111111111111, "grad_norm": 0.7313695549964905, "learning_rate": 6.730000000000001e-05, "loss": -0.0808, "step": 2020 }, { "epoch": 5.638888888888889, "grad_norm": 0.8927887678146362, "learning_rate": 6.763333333333334e-05, "loss": -0.0694, "step": 2030 }, { "epoch": 5.666666666666667, "grad_norm": 1.3876607418060303, "learning_rate": 6.796666666666666e-05, "loss": -0.0789, "step": 2040 }, { "epoch": 5.694444444444445, "grad_norm": 0.5337174534797668, "learning_rate": 6.83e-05, "loss": -0.0728, "step": 2050 }, { "epoch": 5.722222222222222, "grad_norm": 0.8354716897010803, "learning_rate": 6.863333333333334e-05, "loss": -0.0722, "step": 2060 }, { "epoch": 5.75, "grad_norm": 1.063969612121582, "learning_rate": 6.896666666666667e-05, "loss": -0.0662, "step": 2070 }, { "epoch": 5.777777777777778, "grad_norm": 0.9810799956321716, "learning_rate": 6.93e-05, "loss": -0.0571, "step": 2080 }, { "epoch": 5.805555555555555, "grad_norm": 0.8555030822753906, "learning_rate": 6.963333333333334e-05, "loss": -0.0745, "step": 2090 }, { "epoch": 5.833333333333333, "grad_norm": 0.8982958197593689, "learning_rate": 6.996666666666667e-05, "loss": -0.0697, "step": 2100 }, { "epoch": 5.861111111111111, "grad_norm": 0.611605167388916, "learning_rate": 7.03e-05, "loss": -0.0622, "step": 2110 }, { "epoch": 5.888888888888889, "grad_norm": 0.8661583065986633, "learning_rate": 7.063333333333333e-05, "loss": -0.0497, "step": 2120 }, { "epoch": 5.916666666666667, "grad_norm": 0.8051588535308838, "learning_rate": 7.096666666666667e-05, "loss": -0.0525, "step": 2130 }, { "epoch": 5.944444444444445, "grad_norm": 0.8949028253555298, "learning_rate": 7.13e-05, "loss": -0.0699, "step": 2140 }, { "epoch": 5.972222222222222, "grad_norm": 0.9154114127159119, "learning_rate": 7.163333333333334e-05, "loss": -0.0783, "step": 2150 }, { "epoch": 6.0, "grad_norm": 0.7446164488792419, "learning_rate": 7.196666666666668e-05, "loss": -0.0552, "step": 2160 }, { "epoch": 6.027777777777778, "grad_norm": 0.7678980827331543, "learning_rate": 7.23e-05, "loss": -0.0825, "step": 2170 }, { "epoch": 6.055555555555555, "grad_norm": 1.0784051418304443, "learning_rate": 7.263333333333334e-05, "loss": -0.0882, "step": 2180 }, { "epoch": 6.083333333333333, "grad_norm": 0.8895697593688965, "learning_rate": 7.296666666666667e-05, "loss": -0.0769, "step": 2190 }, { "epoch": 6.111111111111111, "grad_norm": 0.6490553617477417, "learning_rate": 7.33e-05, "loss": -0.0804, "step": 2200 }, { "epoch": 6.138888888888889, "grad_norm": 1.0531326532363892, "learning_rate": 7.363333333333334e-05, "loss": -0.0603, "step": 2210 }, { "epoch": 6.166666666666667, "grad_norm": 1.2256500720977783, "learning_rate": 7.396666666666667e-05, "loss": -0.0708, "step": 2220 }, { "epoch": 6.194444444444445, "grad_norm": 0.5784858465194702, "learning_rate": 7.43e-05, "loss": -0.0644, "step": 2230 }, { "epoch": 6.222222222222222, "grad_norm": 0.7431011199951172, "learning_rate": 7.463333333333334e-05, "loss": -0.0761, "step": 2240 }, { "epoch": 6.25, "grad_norm": 0.6309260725975037, "learning_rate": 7.496666666666667e-05, "loss": -0.0727, "step": 2250 }, { "epoch": 6.277777777777778, "grad_norm": 0.7811299562454224, "learning_rate": 7.53e-05, "loss": -0.065, "step": 2260 }, { "epoch": 6.305555555555555, "grad_norm": 0.7168146967887878, "learning_rate": 7.563333333333333e-05, "loss": -0.0662, "step": 2270 }, { "epoch": 6.333333333333333, "grad_norm": 0.8230806589126587, "learning_rate": 7.596666666666668e-05, "loss": -0.0775, "step": 2280 }, { "epoch": 6.361111111111111, "grad_norm": 0.7440839409828186, "learning_rate": 7.630000000000001e-05, "loss": -0.0668, "step": 2290 }, { "epoch": 6.388888888888889, "grad_norm": 1.095524787902832, "learning_rate": 7.663333333333333e-05, "loss": -0.0723, "step": 2300 }, { "epoch": 6.416666666666667, "grad_norm": 1.0188788175582886, "learning_rate": 7.696666666666668e-05, "loss": -0.0771, "step": 2310 }, { "epoch": 6.444444444444445, "grad_norm": 0.9254186153411865, "learning_rate": 7.730000000000001e-05, "loss": -0.059, "step": 2320 }, { "epoch": 6.472222222222222, "grad_norm": 1.076367974281311, "learning_rate": 7.763333333333334e-05, "loss": -0.0702, "step": 2330 }, { "epoch": 6.5, "grad_norm": 0.8928374648094177, "learning_rate": 7.796666666666666e-05, "loss": -0.0769, "step": 2340 }, { "epoch": 6.527777777777778, "grad_norm": 0.869167149066925, "learning_rate": 7.83e-05, "loss": -0.0621, "step": 2350 }, { "epoch": 6.555555555555555, "grad_norm": 0.619438111782074, "learning_rate": 7.863333333333334e-05, "loss": -0.0791, "step": 2360 }, { "epoch": 6.583333333333333, "grad_norm": 0.6670336127281189, "learning_rate": 7.896666666666667e-05, "loss": -0.0668, "step": 2370 }, { "epoch": 6.611111111111111, "grad_norm": 0.7099744081497192, "learning_rate": 7.93e-05, "loss": -0.0784, "step": 2380 }, { "epoch": 6.638888888888889, "grad_norm": 0.7188742756843567, "learning_rate": 7.963333333333334e-05, "loss": -0.0605, "step": 2390 }, { "epoch": 6.666666666666667, "grad_norm": 0.8846904635429382, "learning_rate": 7.996666666666667e-05, "loss": -0.0836, "step": 2400 }, { "epoch": 6.694444444444445, "grad_norm": 1.0989766120910645, "learning_rate": 8.030000000000001e-05, "loss": -0.0604, "step": 2410 }, { "epoch": 6.722222222222222, "grad_norm": 0.5285449624061584, "learning_rate": 8.063333333333333e-05, "loss": -0.0705, "step": 2420 }, { "epoch": 6.75, "grad_norm": 0.896422266960144, "learning_rate": 8.096666666666667e-05, "loss": -0.0782, "step": 2430 }, { "epoch": 6.777777777777778, "grad_norm": 0.8955458402633667, "learning_rate": 8.13e-05, "loss": -0.074, "step": 2440 }, { "epoch": 6.805555555555555, "grad_norm": 0.6945486068725586, "learning_rate": 8.163333333333334e-05, "loss": -0.0756, "step": 2450 }, { "epoch": 6.833333333333333, "grad_norm": 0.6047375202178955, "learning_rate": 8.196666666666668e-05, "loss": -0.0686, "step": 2460 }, { "epoch": 6.861111111111111, "grad_norm": 0.6410619020462036, "learning_rate": 8.23e-05, "loss": -0.0835, "step": 2470 }, { "epoch": 6.888888888888889, "grad_norm": 0.7192192077636719, "learning_rate": 8.263333333333334e-05, "loss": -0.0796, "step": 2480 }, { "epoch": 6.916666666666667, "grad_norm": 0.932910680770874, "learning_rate": 8.296666666666667e-05, "loss": -0.0592, "step": 2490 }, { "epoch": 6.944444444444445, "grad_norm": 0.8369176983833313, "learning_rate": 8.33e-05, "loss": -0.0714, "step": 2500 }, { "epoch": 6.972222222222222, "grad_norm": 0.7967250347137451, "learning_rate": 8.363333333333334e-05, "loss": -0.0743, "step": 2510 }, { "epoch": 7.0, "grad_norm": 0.7530558705329895, "learning_rate": 8.396666666666667e-05, "loss": -0.0867, "step": 2520 }, { "epoch": 7.027777777777778, "grad_norm": 0.7421709299087524, "learning_rate": 8.43e-05, "loss": -0.0884, "step": 2530 }, { "epoch": 7.055555555555555, "grad_norm": 0.8251081705093384, "learning_rate": 8.463333333333335e-05, "loss": -0.062, "step": 2540 }, { "epoch": 7.083333333333333, "grad_norm": 1.0071446895599365, "learning_rate": 8.496666666666667e-05, "loss": -0.0828, "step": 2550 }, { "epoch": 7.111111111111111, "grad_norm": 0.5009008049964905, "learning_rate": 8.53e-05, "loss": -0.066, "step": 2560 }, { "epoch": 7.138888888888889, "grad_norm": 0.8243054747581482, "learning_rate": 8.563333333333333e-05, "loss": -0.0683, "step": 2570 }, { "epoch": 7.166666666666667, "grad_norm": 0.588026762008667, "learning_rate": 8.596666666666668e-05, "loss": -0.0738, "step": 2580 }, { "epoch": 7.194444444444445, "grad_norm": 0.6533929109573364, "learning_rate": 8.63e-05, "loss": -0.088, "step": 2590 }, { "epoch": 7.222222222222222, "grad_norm": 0.8693581819534302, "learning_rate": 8.663333333333333e-05, "loss": -0.0699, "step": 2600 }, { "epoch": 7.25, "grad_norm": 0.6362013816833496, "learning_rate": 8.696666666666668e-05, "loss": -0.0801, "step": 2610 }, { "epoch": 7.277777777777778, "grad_norm": 0.8143713474273682, "learning_rate": 8.730000000000001e-05, "loss": -0.0901, "step": 2620 }, { "epoch": 7.305555555555555, "grad_norm": 0.702124834060669, "learning_rate": 8.763333333333334e-05, "loss": -0.0686, "step": 2630 }, { "epoch": 7.333333333333333, "grad_norm": 1.0381027460098267, "learning_rate": 8.796666666666667e-05, "loss": -0.0687, "step": 2640 }, { "epoch": 7.361111111111111, "grad_norm": 0.6594991683959961, "learning_rate": 8.83e-05, "loss": -0.0865, "step": 2650 }, { "epoch": 7.388888888888889, "grad_norm": 0.9706031680107117, "learning_rate": 8.863333333333334e-05, "loss": -0.0854, "step": 2660 }, { "epoch": 7.416666666666667, "grad_norm": 0.7076573371887207, "learning_rate": 8.896666666666667e-05, "loss": -0.0755, "step": 2670 }, { "epoch": 7.444444444444445, "grad_norm": 0.5413500666618347, "learning_rate": 8.93e-05, "loss": -0.0807, "step": 2680 }, { "epoch": 7.472222222222222, "grad_norm": 0.8320577144622803, "learning_rate": 8.963333333333333e-05, "loss": -0.0881, "step": 2690 }, { "epoch": 7.5, "grad_norm": 0.4946286976337433, "learning_rate": 8.996666666666667e-05, "loss": -0.0771, "step": 2700 }, { "epoch": 7.527777777777778, "grad_norm": 0.6335960030555725, "learning_rate": 9.030000000000001e-05, "loss": -0.0846, "step": 2710 }, { "epoch": 7.555555555555555, "grad_norm": 0.6049409508705139, "learning_rate": 9.063333333333333e-05, "loss": -0.071, "step": 2720 }, { "epoch": 7.583333333333333, "grad_norm": 0.6953684091567993, "learning_rate": 9.096666666666666e-05, "loss": -0.0972, "step": 2730 }, { "epoch": 7.611111111111111, "grad_norm": 0.7790083885192871, "learning_rate": 9.130000000000001e-05, "loss": -0.0936, "step": 2740 }, { "epoch": 7.638888888888889, "grad_norm": 0.8471085429191589, "learning_rate": 9.163333333333334e-05, "loss": -0.0835, "step": 2750 }, { "epoch": 7.666666666666667, "grad_norm": 0.8508561253547668, "learning_rate": 9.196666666666666e-05, "loss": -0.0958, "step": 2760 }, { "epoch": 7.694444444444445, "grad_norm": 0.7866385579109192, "learning_rate": 9.230000000000001e-05, "loss": -0.0907, "step": 2770 }, { "epoch": 7.722222222222222, "grad_norm": 0.9133675694465637, "learning_rate": 9.263333333333334e-05, "loss": -0.0879, "step": 2780 }, { "epoch": 7.75, "grad_norm": 0.716550350189209, "learning_rate": 9.296666666666667e-05, "loss": -0.0814, "step": 2790 }, { "epoch": 7.777777777777778, "grad_norm": 0.594226062297821, "learning_rate": 9.33e-05, "loss": -0.0853, "step": 2800 }, { "epoch": 7.805555555555555, "grad_norm": 0.6017398834228516, "learning_rate": 9.363333333333334e-05, "loss": -0.0909, "step": 2810 }, { "epoch": 7.833333333333333, "grad_norm": 0.6105267405509949, "learning_rate": 9.396666666666667e-05, "loss": -0.088, "step": 2820 }, { "epoch": 7.861111111111111, "grad_norm": 0.8928349614143372, "learning_rate": 9.43e-05, "loss": -0.0743, "step": 2830 }, { "epoch": 7.888888888888889, "grad_norm": 0.6611433029174805, "learning_rate": 9.463333333333333e-05, "loss": -0.0808, "step": 2840 }, { "epoch": 7.916666666666667, "grad_norm": 0.7590584754943848, "learning_rate": 9.496666666666667e-05, "loss": -0.0855, "step": 2850 }, { "epoch": 7.944444444444445, "grad_norm": 0.7500903606414795, "learning_rate": 9.53e-05, "loss": -0.0852, "step": 2860 }, { "epoch": 7.972222222222222, "grad_norm": 0.6876359581947327, "learning_rate": 9.563333333333334e-05, "loss": -0.0745, "step": 2870 }, { "epoch": 8.0, "grad_norm": 0.6954237818717957, "learning_rate": 9.596666666666668e-05, "loss": -0.0875, "step": 2880 }, { "epoch": 8.027777777777779, "grad_norm": 0.6060745120048523, "learning_rate": 9.63e-05, "loss": -0.0799, "step": 2890 }, { "epoch": 8.055555555555555, "grad_norm": 0.6287747621536255, "learning_rate": 9.663333333333334e-05, "loss": -0.0819, "step": 2900 }, { "epoch": 8.083333333333334, "grad_norm": 0.6438542604446411, "learning_rate": 9.696666666666667e-05, "loss": -0.0872, "step": 2910 }, { "epoch": 8.11111111111111, "grad_norm": 0.7381275296211243, "learning_rate": 9.730000000000001e-05, "loss": -0.083, "step": 2920 }, { "epoch": 8.13888888888889, "grad_norm": 0.6046106219291687, "learning_rate": 9.763333333333334e-05, "loss": -0.0841, "step": 2930 }, { "epoch": 8.166666666666666, "grad_norm": 0.8178691864013672, "learning_rate": 9.796666666666667e-05, "loss": -0.0695, "step": 2940 }, { "epoch": 8.194444444444445, "grad_norm": 0.6135996580123901, "learning_rate": 9.83e-05, "loss": -0.0676, "step": 2950 }, { "epoch": 8.222222222222221, "grad_norm": 0.8107286095619202, "learning_rate": 9.863333333333334e-05, "loss": -0.0781, "step": 2960 }, { "epoch": 8.25, "grad_norm": 0.8435459136962891, "learning_rate": 9.896666666666667e-05, "loss": -0.0836, "step": 2970 }, { "epoch": 8.277777777777779, "grad_norm": 0.4758634865283966, "learning_rate": 9.93e-05, "loss": -0.0938, "step": 2980 }, { "epoch": 8.305555555555555, "grad_norm": 0.49465373158454895, "learning_rate": 9.963333333333333e-05, "loss": -0.0857, "step": 2990 }, { "epoch": 8.333333333333334, "grad_norm": 0.8318036794662476, "learning_rate": 9.996666666666668e-05, "loss": -0.0688, "step": 3000 }, { "epoch": 8.36111111111111, "grad_norm": 0.5224122405052185, "learning_rate": 9.999999384858465e-05, "loss": -0.096, "step": 3010 }, { "epoch": 8.38888888888889, "grad_norm": 0.7248770594596863, "learning_rate": 9.999997258443473e-05, "loss": -0.0891, "step": 3020 }, { "epoch": 8.416666666666666, "grad_norm": 0.5976536273956299, "learning_rate": 9.999993613161331e-05, "loss": -0.0893, "step": 3030 }, { "epoch": 8.444444444444445, "grad_norm": 1.001853108406067, "learning_rate": 9.999988449013146e-05, "loss": -0.0874, "step": 3040 }, { "epoch": 8.472222222222221, "grad_norm": 0.704267144203186, "learning_rate": 9.99998176600049e-05, "loss": -0.0833, "step": 3050 }, { "epoch": 8.5, "grad_norm": 0.619638979434967, "learning_rate": 9.999973564125389e-05, "loss": -0.0991, "step": 3060 }, { "epoch": 8.527777777777779, "grad_norm": 0.6297928690910339, "learning_rate": 9.999963843390335e-05, "loss": -0.0949, "step": 3070 }, { "epoch": 8.555555555555555, "grad_norm": 0.8818594217300415, "learning_rate": 9.999952603798282e-05, "loss": -0.0681, "step": 3080 }, { "epoch": 8.583333333333334, "grad_norm": 0.5903865694999695, "learning_rate": 9.999939845352646e-05, "loss": -0.0914, "step": 3090 }, { "epoch": 8.61111111111111, "grad_norm": 0.6764851808547974, "learning_rate": 9.999925568057298e-05, "loss": -0.092, "step": 3100 }, { "epoch": 8.63888888888889, "grad_norm": 0.590708315372467, "learning_rate": 9.999909771916578e-05, "loss": -0.0917, "step": 3110 }, { "epoch": 8.666666666666666, "grad_norm": 0.7586330771446228, "learning_rate": 9.999892456935285e-05, "loss": -0.0917, "step": 3120 }, { "epoch": 8.694444444444445, "grad_norm": 0.6225112080574036, "learning_rate": 9.999873623118679e-05, "loss": -0.0884, "step": 3130 }, { "epoch": 8.722222222222221, "grad_norm": 0.5763458013534546, "learning_rate": 9.999853270472479e-05, "loss": -0.0916, "step": 3140 }, { "epoch": 8.75, "grad_norm": 0.6602080464363098, "learning_rate": 9.999831399002871e-05, "loss": -0.0957, "step": 3150 }, { "epoch": 8.777777777777779, "grad_norm": 0.6241950392723083, "learning_rate": 9.999808008716494e-05, "loss": -0.0997, "step": 3160 }, { "epoch": 8.805555555555555, "grad_norm": 0.712907075881958, "learning_rate": 9.999783099620459e-05, "loss": -0.0927, "step": 3170 }, { "epoch": 8.833333333333334, "grad_norm": 0.6818045377731323, "learning_rate": 9.999756671722328e-05, "loss": -0.0979, "step": 3180 }, { "epoch": 8.86111111111111, "grad_norm": 0.7351495623588562, "learning_rate": 9.99972872503013e-05, "loss": -0.1006, "step": 3190 }, { "epoch": 8.88888888888889, "grad_norm": 0.7318455576896667, "learning_rate": 9.999699259552359e-05, "loss": -0.0879, "step": 3200 }, { "epoch": 8.916666666666666, "grad_norm": 0.5421245098114014, "learning_rate": 9.99966827529796e-05, "loss": -0.1018, "step": 3210 }, { "epoch": 8.944444444444445, "grad_norm": 0.5177657008171082, "learning_rate": 9.999635772276348e-05, "loss": -0.1006, "step": 3220 }, { "epoch": 8.972222222222221, "grad_norm": 0.6163880825042725, "learning_rate": 9.999601750497396e-05, "loss": -0.1049, "step": 3230 }, { "epoch": 9.0, "grad_norm": 0.5744783282279968, "learning_rate": 9.99956620997144e-05, "loss": -0.1046, "step": 3240 }, { "epoch": 9.027777777777779, "grad_norm": 0.5593729019165039, "learning_rate": 9.999529150709275e-05, "loss": -0.1012, "step": 3250 }, { "epoch": 9.055555555555555, "grad_norm": 0.6273961067199707, "learning_rate": 9.999490572722158e-05, "loss": -0.1094, "step": 3260 }, { "epoch": 9.083333333333334, "grad_norm": 0.6853123903274536, "learning_rate": 9.99945047602181e-05, "loss": -0.1011, "step": 3270 }, { "epoch": 9.11111111111111, "grad_norm": 0.7186084985733032, "learning_rate": 9.99940886062041e-05, "loss": -0.1046, "step": 3280 }, { "epoch": 9.13888888888889, "grad_norm": 0.8748005032539368, "learning_rate": 9.999365726530599e-05, "loss": -0.0976, "step": 3290 }, { "epoch": 9.166666666666666, "grad_norm": 0.5652263164520264, "learning_rate": 9.999321073765481e-05, "loss": -0.0894, "step": 3300 }, { "epoch": 9.194444444444445, "grad_norm": 0.5907758474349976, "learning_rate": 9.99927490233862e-05, "loss": -0.1119, "step": 3310 }, { "epoch": 9.222222222222221, "grad_norm": 0.7235530614852905, "learning_rate": 9.999227212264043e-05, "loss": -0.1109, "step": 3320 }, { "epoch": 9.25, "grad_norm": 0.48440590500831604, "learning_rate": 9.999178003556236e-05, "loss": -0.0969, "step": 3330 }, { "epoch": 9.277777777777779, "grad_norm": 0.6958463788032532, "learning_rate": 9.999127276230146e-05, "loss": -0.1078, "step": 3340 }, { "epoch": 9.305555555555555, "grad_norm": 0.5558679103851318, "learning_rate": 9.999075030301184e-05, "loss": -0.1065, "step": 3350 }, { "epoch": 9.333333333333334, "grad_norm": 0.6283583045005798, "learning_rate": 9.999021265785221e-05, "loss": -0.0999, "step": 3360 }, { "epoch": 9.36111111111111, "grad_norm": 0.536971390247345, "learning_rate": 9.998965982698589e-05, "loss": -0.0971, "step": 3370 }, { "epoch": 9.38888888888889, "grad_norm": 0.4724583029747009, "learning_rate": 9.998909181058082e-05, "loss": -0.1122, "step": 3380 }, { "epoch": 9.416666666666666, "grad_norm": 0.7068848013877869, "learning_rate": 9.998850860880953e-05, "loss": -0.1015, "step": 3390 }, { "epoch": 9.444444444444445, "grad_norm": 0.8180752396583557, "learning_rate": 9.998791022184922e-05, "loss": -0.1019, "step": 3400 }, { "epoch": 9.472222222222221, "grad_norm": 0.9478771090507507, "learning_rate": 9.99872966498816e-05, "loss": -0.1048, "step": 3410 }, { "epoch": 9.5, "grad_norm": 0.5724837183952332, "learning_rate": 9.998666789309313e-05, "loss": -0.0982, "step": 3420 }, { "epoch": 9.527777777777779, "grad_norm": 0.6660880446434021, "learning_rate": 9.998602395167475e-05, "loss": -0.1019, "step": 3430 }, { "epoch": 9.555555555555555, "grad_norm": 0.6927004456520081, "learning_rate": 9.998536482582213e-05, "loss": -0.0941, "step": 3440 }, { "epoch": 9.583333333333334, "grad_norm": 0.6658582091331482, "learning_rate": 9.998469051573544e-05, "loss": -0.0999, "step": 3450 }, { "epoch": 9.61111111111111, "grad_norm": 0.7006404995918274, "learning_rate": 9.998400102161954e-05, "loss": -0.1063, "step": 3460 }, { "epoch": 9.63888888888889, "grad_norm": 0.7133253216743469, "learning_rate": 9.998329634368388e-05, "loss": -0.103, "step": 3470 }, { "epoch": 9.666666666666666, "grad_norm": 0.37817519903182983, "learning_rate": 9.998257648214253e-05, "loss": -0.1098, "step": 3480 }, { "epoch": 9.694444444444445, "grad_norm": 0.5944302082061768, "learning_rate": 9.998184143721417e-05, "loss": -0.0994, "step": 3490 }, { "epoch": 9.722222222222221, "grad_norm": 0.530181348323822, "learning_rate": 9.998109120912206e-05, "loss": -0.1073, "step": 3500 }, { "epoch": 9.75, "grad_norm": 0.5795698761940002, "learning_rate": 9.998032579809411e-05, "loss": -0.1063, "step": 3510 }, { "epoch": 9.777777777777779, "grad_norm": 0.6330494284629822, "learning_rate": 9.997954520436286e-05, "loss": -0.0978, "step": 3520 }, { "epoch": 9.805555555555555, "grad_norm": 0.6317851543426514, "learning_rate": 9.997874942816538e-05, "loss": -0.1134, "step": 3530 }, { "epoch": 9.833333333333334, "grad_norm": 0.5259532928466797, "learning_rate": 9.997793846974345e-05, "loss": -0.1034, "step": 3540 }, { "epoch": 9.86111111111111, "grad_norm": 0.6903944611549377, "learning_rate": 9.997711232934341e-05, "loss": -0.1052, "step": 3550 }, { "epoch": 9.88888888888889, "grad_norm": 0.5505329966545105, "learning_rate": 9.99762710072162e-05, "loss": -0.1078, "step": 3560 }, { "epoch": 9.916666666666666, "grad_norm": 0.7442355155944824, "learning_rate": 9.997541450361743e-05, "loss": -0.0971, "step": 3570 }, { "epoch": 9.944444444444445, "grad_norm": 0.6100121140480042, "learning_rate": 9.997454281880723e-05, "loss": -0.0978, "step": 3580 }, { "epoch": 9.972222222222221, "grad_norm": 0.6983389258384705, "learning_rate": 9.997365595305044e-05, "loss": -0.1027, "step": 3590 }, { "epoch": 10.0, "grad_norm": 0.6287941932678223, "learning_rate": 9.997275390661644e-05, "loss": -0.1043, "step": 3600 }, { "epoch": 10.027777777777779, "grad_norm": 0.5197926759719849, "learning_rate": 9.997183667977926e-05, "loss": -0.0978, "step": 3610 }, { "epoch": 10.055555555555555, "grad_norm": 0.6222547292709351, "learning_rate": 9.997090427281752e-05, "loss": -0.1059, "step": 3620 }, { "epoch": 10.083333333333334, "grad_norm": 0.602509081363678, "learning_rate": 9.996995668601448e-05, "loss": -0.1065, "step": 3630 }, { "epoch": 10.11111111111111, "grad_norm": 0.4975406229496002, "learning_rate": 9.996899391965798e-05, "loss": -0.116, "step": 3640 }, { "epoch": 10.13888888888889, "grad_norm": 0.5303145051002502, "learning_rate": 9.996801597404048e-05, "loss": -0.1036, "step": 3650 }, { "epoch": 10.166666666666666, "grad_norm": 0.7295588850975037, "learning_rate": 9.996702284945905e-05, "loss": -0.1158, "step": 3660 }, { "epoch": 10.194444444444445, "grad_norm": 0.5798919200897217, "learning_rate": 9.996601454621539e-05, "loss": -0.1034, "step": 3670 }, { "epoch": 10.222222222222221, "grad_norm": 0.5859751105308533, "learning_rate": 9.996499106461577e-05, "loss": -0.1166, "step": 3680 }, { "epoch": 10.25, "grad_norm": 0.48836854100227356, "learning_rate": 9.996395240497112e-05, "loss": -0.1158, "step": 3690 }, { "epoch": 10.277777777777779, "grad_norm": 0.5472903847694397, "learning_rate": 9.996289856759696e-05, "loss": -0.1181, "step": 3700 }, { "epoch": 10.305555555555555, "grad_norm": 0.5930604338645935, "learning_rate": 9.996182955281342e-05, "loss": -0.1087, "step": 3710 }, { "epoch": 10.333333333333334, "grad_norm": 1.0698508024215698, "learning_rate": 9.996074536094519e-05, "loss": -0.1026, "step": 3720 }, { "epoch": 10.36111111111111, "grad_norm": 0.42496204376220703, "learning_rate": 9.995964599232168e-05, "loss": -0.1112, "step": 3730 }, { "epoch": 10.38888888888889, "grad_norm": 0.640318751335144, "learning_rate": 9.995853144727683e-05, "loss": -0.1033, "step": 3740 }, { "epoch": 10.416666666666666, "grad_norm": 0.5876253843307495, "learning_rate": 9.99574017261492e-05, "loss": -0.1118, "step": 3750 }, { "epoch": 10.444444444444445, "grad_norm": 0.639844536781311, "learning_rate": 9.995625682928198e-05, "loss": -0.113, "step": 3760 }, { "epoch": 10.472222222222221, "grad_norm": 0.4998249411582947, "learning_rate": 9.995509675702295e-05, "loss": -0.116, "step": 3770 }, { "epoch": 10.5, "grad_norm": 0.4126482903957367, "learning_rate": 9.995392150972451e-05, "loss": -0.1149, "step": 3780 }, { "epoch": 10.527777777777779, "grad_norm": 0.5932841300964355, "learning_rate": 9.995273108774366e-05, "loss": -0.1, "step": 3790 }, { "epoch": 10.555555555555555, "grad_norm": 0.42811429500579834, "learning_rate": 9.995152549144205e-05, "loss": -0.1008, "step": 3800 }, { "epoch": 10.583333333333334, "grad_norm": 0.47091367840766907, "learning_rate": 9.995030472118587e-05, "loss": -0.1052, "step": 3810 }, { "epoch": 10.61111111111111, "grad_norm": 0.8209417462348938, "learning_rate": 9.9949068777346e-05, "loss": -0.1031, "step": 3820 }, { "epoch": 10.63888888888889, "grad_norm": 0.7360351085662842, "learning_rate": 9.994781766029786e-05, "loss": -0.1129, "step": 3830 }, { "epoch": 10.666666666666666, "grad_norm": 0.5039061903953552, "learning_rate": 9.994655137042151e-05, "loss": -0.1075, "step": 3840 }, { "epoch": 10.694444444444445, "grad_norm": 0.6326560974121094, "learning_rate": 9.99452699081016e-05, "loss": -0.1133, "step": 3850 }, { "epoch": 10.722222222222221, "grad_norm": 0.5246423482894897, "learning_rate": 9.994397327372743e-05, "loss": -0.0993, "step": 3860 }, { "epoch": 10.75, "grad_norm": 0.6557695269584656, "learning_rate": 9.994266146769286e-05, "loss": -0.1033, "step": 3870 }, { "epoch": 10.777777777777779, "grad_norm": 0.43626055121421814, "learning_rate": 9.994133449039642e-05, "loss": -0.109, "step": 3880 }, { "epoch": 10.805555555555555, "grad_norm": 0.538362979888916, "learning_rate": 9.993999234224118e-05, "loss": -0.1029, "step": 3890 }, { "epoch": 10.833333333333334, "grad_norm": 0.5343350172042847, "learning_rate": 9.993863502363485e-05, "loss": -0.1107, "step": 3900 }, { "epoch": 10.86111111111111, "grad_norm": 0.40724456310272217, "learning_rate": 9.993726253498976e-05, "loss": -0.1166, "step": 3910 }, { "epoch": 10.88888888888889, "grad_norm": 0.6878448128700256, "learning_rate": 9.993587487672282e-05, "loss": -0.1215, "step": 3920 }, { "epoch": 10.916666666666666, "grad_norm": 0.4779484272003174, "learning_rate": 9.993447204925558e-05, "loss": -0.1144, "step": 3930 }, { "epoch": 10.944444444444445, "grad_norm": 0.7894007563591003, "learning_rate": 9.993305405301416e-05, "loss": -0.1068, "step": 3940 }, { "epoch": 10.972222222222221, "grad_norm": 0.6225062012672424, "learning_rate": 9.993162088842935e-05, "loss": -0.1113, "step": 3950 }, { "epoch": 11.0, "grad_norm": 0.6981886625289917, "learning_rate": 9.993017255593646e-05, "loss": -0.095, "step": 3960 }, { "epoch": 11.027777777777779, "grad_norm": 0.7016492486000061, "learning_rate": 9.992870905597548e-05, "loss": -0.1126, "step": 3970 }, { "epoch": 11.055555555555555, "grad_norm": 0.47347357869148254, "learning_rate": 9.9927230388991e-05, "loss": -0.1192, "step": 3980 }, { "epoch": 11.083333333333334, "grad_norm": 0.5803160667419434, "learning_rate": 9.992573655543215e-05, "loss": -0.1182, "step": 3990 }, { "epoch": 11.11111111111111, "grad_norm": 0.7297594547271729, "learning_rate": 9.992422755575277e-05, "loss": -0.123, "step": 4000 }, { "epoch": 11.13888888888889, "grad_norm": 0.37463387846946716, "learning_rate": 9.992270339041123e-05, "loss": -0.1172, "step": 4010 }, { "epoch": 11.166666666666666, "grad_norm": 0.4964923858642578, "learning_rate": 9.992116405987053e-05, "loss": -0.1221, "step": 4020 }, { "epoch": 11.194444444444445, "grad_norm": 0.47075405716896057, "learning_rate": 9.991960956459828e-05, "loss": -0.1188, "step": 4030 }, { "epoch": 11.222222222222221, "grad_norm": 0.7090590000152588, "learning_rate": 9.991803990506669e-05, "loss": -0.1146, "step": 4040 }, { "epoch": 11.25, "grad_norm": 0.48406854271888733, "learning_rate": 9.991645508175258e-05, "loss": -0.1139, "step": 4050 }, { "epoch": 11.277777777777779, "grad_norm": 0.5257912278175354, "learning_rate": 9.99148550951374e-05, "loss": -0.1184, "step": 4060 }, { "epoch": 11.305555555555555, "grad_norm": 0.4500105381011963, "learning_rate": 9.991323994570716e-05, "loss": -0.1147, "step": 4070 }, { "epoch": 11.333333333333334, "grad_norm": 0.6699492931365967, "learning_rate": 9.99116096339525e-05, "loss": -0.1113, "step": 4080 }, { "epoch": 11.36111111111111, "grad_norm": 0.4863075613975525, "learning_rate": 9.990996416036869e-05, "loss": -0.1136, "step": 4090 }, { "epoch": 11.38888888888889, "grad_norm": 0.6549540162086487, "learning_rate": 9.990830352545555e-05, "loss": -0.1199, "step": 4100 }, { "epoch": 11.416666666666666, "grad_norm": 0.6956165432929993, "learning_rate": 9.990662772971756e-05, "loss": -0.1187, "step": 4110 }, { "epoch": 11.444444444444445, "grad_norm": 0.553525984287262, "learning_rate": 9.990493677366376e-05, "loss": -0.1134, "step": 4120 }, { "epoch": 11.472222222222221, "grad_norm": 0.41752779483795166, "learning_rate": 9.990323065780786e-05, "loss": -0.121, "step": 4130 }, { "epoch": 11.5, "grad_norm": 0.44535601139068604, "learning_rate": 9.990150938266808e-05, "loss": -0.1158, "step": 4140 }, { "epoch": 11.527777777777779, "grad_norm": 0.5236654877662659, "learning_rate": 9.989977294876733e-05, "loss": -0.1148, "step": 4150 }, { "epoch": 11.555555555555555, "grad_norm": 0.4530026912689209, "learning_rate": 9.989802135663308e-05, "loss": -0.1253, "step": 4160 }, { "epoch": 11.583333333333334, "grad_norm": 0.506323516368866, "learning_rate": 9.989625460679743e-05, "loss": -0.1244, "step": 4170 }, { "epoch": 11.61111111111111, "grad_norm": 0.4291667640209198, "learning_rate": 9.989447269979706e-05, "loss": -0.1218, "step": 4180 }, { "epoch": 11.63888888888889, "grad_norm": 0.41087469458580017, "learning_rate": 9.989267563617328e-05, "loss": -0.1186, "step": 4190 }, { "epoch": 11.666666666666666, "grad_norm": 0.44854041934013367, "learning_rate": 9.989086341647198e-05, "loss": -0.1261, "step": 4200 }, { "epoch": 11.694444444444445, "grad_norm": 0.5402235984802246, "learning_rate": 9.988903604124366e-05, "loss": -0.1219, "step": 4210 }, { "epoch": 11.722222222222221, "grad_norm": 0.3694511651992798, "learning_rate": 9.988719351104343e-05, "loss": -0.1219, "step": 4220 }, { "epoch": 11.75, "grad_norm": 0.2753910422325134, "learning_rate": 9.9885335826431e-05, "loss": -0.1273, "step": 4230 }, { "epoch": 11.777777777777779, "grad_norm": 0.5134520530700684, "learning_rate": 9.988346298797071e-05, "loss": -0.1288, "step": 4240 }, { "epoch": 11.805555555555555, "grad_norm": 0.6317590475082397, "learning_rate": 9.988157499623146e-05, "loss": -0.114, "step": 4250 }, { "epoch": 11.833333333333334, "grad_norm": 0.5103162527084351, "learning_rate": 9.987967185178677e-05, "loss": -0.1279, "step": 4260 }, { "epoch": 11.86111111111111, "grad_norm": 0.5108464956283569, "learning_rate": 9.987775355521476e-05, "loss": -0.1177, "step": 4270 }, { "epoch": 11.88888888888889, "grad_norm": 0.4854657053947449, "learning_rate": 9.987582010709817e-05, "loss": -0.12, "step": 4280 }, { "epoch": 11.916666666666666, "grad_norm": 0.43885597586631775, "learning_rate": 9.987387150802431e-05, "loss": -0.0931, "step": 4290 }, { "epoch": 11.944444444444445, "grad_norm": 0.33176305890083313, "learning_rate": 9.987190775858517e-05, "loss": -0.1287, "step": 4300 }, { "epoch": 11.972222222222221, "grad_norm": 0.6491981744766235, "learning_rate": 9.98699288593772e-05, "loss": -0.1195, "step": 4310 }, { "epoch": 12.0, "grad_norm": 0.6647459864616394, "learning_rate": 9.986793481100161e-05, "loss": -0.1219, "step": 4320 }, { "epoch": 12.027777777777779, "grad_norm": 0.6765521168708801, "learning_rate": 9.986592561406412e-05, "loss": -0.1211, "step": 4330 }, { "epoch": 12.055555555555555, "grad_norm": 0.44651028513908386, "learning_rate": 9.986390126917503e-05, "loss": -0.1286, "step": 4340 }, { "epoch": 12.083333333333334, "grad_norm": 0.6327084302902222, "learning_rate": 9.986186177694933e-05, "loss": -0.1249, "step": 4350 }, { "epoch": 12.11111111111111, "grad_norm": 0.47303828597068787, "learning_rate": 9.985980713800656e-05, "loss": -0.1086, "step": 4360 }, { "epoch": 12.13888888888889, "grad_norm": 0.5999845862388611, "learning_rate": 9.985773735297084e-05, "loss": -0.1255, "step": 4370 }, { "epoch": 12.166666666666666, "grad_norm": 0.43594375252723694, "learning_rate": 9.985565242247092e-05, "loss": -0.1221, "step": 4380 }, { "epoch": 12.194444444444445, "grad_norm": 0.5596786141395569, "learning_rate": 9.985355234714016e-05, "loss": -0.1167, "step": 4390 }, { "epoch": 12.222222222222221, "grad_norm": 0.4503979980945587, "learning_rate": 9.985143712761652e-05, "loss": -0.1291, "step": 4400 }, { "epoch": 12.25, "grad_norm": 0.5506424307823181, "learning_rate": 9.984930676454252e-05, "loss": -0.128, "step": 4410 }, { "epoch": 12.277777777777779, "grad_norm": 0.6324775218963623, "learning_rate": 9.984716125856532e-05, "loss": -0.1284, "step": 4420 }, { "epoch": 12.305555555555555, "grad_norm": 0.337259978055954, "learning_rate": 9.984500061033667e-05, "loss": -0.1348, "step": 4430 }, { "epoch": 12.333333333333334, "grad_norm": 0.3679768443107605, "learning_rate": 9.984282482051293e-05, "loss": -0.1252, "step": 4440 }, { "epoch": 12.36111111111111, "grad_norm": 0.5271323919296265, "learning_rate": 9.9840633889755e-05, "loss": -0.1251, "step": 4450 }, { "epoch": 12.38888888888889, "grad_norm": 0.5630580186843872, "learning_rate": 9.983842781872848e-05, "loss": -0.1214, "step": 4460 }, { "epoch": 12.416666666666666, "grad_norm": 0.40615320205688477, "learning_rate": 9.98362066081035e-05, "loss": -0.128, "step": 4470 }, { "epoch": 12.444444444444445, "grad_norm": 0.3234466016292572, "learning_rate": 9.983397025855479e-05, "loss": -0.1294, "step": 4480 }, { "epoch": 12.472222222222221, "grad_norm": 0.43933799862861633, "learning_rate": 9.983171877076171e-05, "loss": -0.1263, "step": 4490 }, { "epoch": 12.5, "grad_norm": 0.41943180561065674, "learning_rate": 9.98294521454082e-05, "loss": -0.1136, "step": 4500 }, { "epoch": 12.527777777777779, "grad_norm": 0.4436921179294586, "learning_rate": 9.98271703831828e-05, "loss": -0.1265, "step": 4510 }, { "epoch": 12.555555555555555, "grad_norm": 0.46341851353645325, "learning_rate": 9.982487348477865e-05, "loss": -0.1202, "step": 4520 }, { "epoch": 12.583333333333334, "grad_norm": 0.44143661856651306, "learning_rate": 9.982256145089347e-05, "loss": -0.1227, "step": 4530 }, { "epoch": 12.61111111111111, "grad_norm": 0.5824288129806519, "learning_rate": 9.982023428222962e-05, "loss": -0.1337, "step": 4540 }, { "epoch": 12.63888888888889, "grad_norm": 0.6291306018829346, "learning_rate": 9.981789197949403e-05, "loss": -0.1097, "step": 4550 }, { "epoch": 12.666666666666666, "grad_norm": 0.5472758412361145, "learning_rate": 9.98155345433982e-05, "loss": -0.1245, "step": 4560 }, { "epoch": 12.694444444444445, "grad_norm": 0.5827226042747498, "learning_rate": 9.981316197465831e-05, "loss": -0.1255, "step": 4570 }, { "epoch": 12.722222222222221, "grad_norm": 0.3978451192378998, "learning_rate": 9.981077427399504e-05, "loss": -0.1242, "step": 4580 }, { "epoch": 12.75, "grad_norm": 0.38386431336402893, "learning_rate": 9.980837144213371e-05, "loss": -0.136, "step": 4590 }, { "epoch": 12.777777777777779, "grad_norm": 0.40930354595184326, "learning_rate": 9.980595347980426e-05, "loss": -0.1319, "step": 4600 }, { "epoch": 12.805555555555555, "grad_norm": 0.552752673625946, "learning_rate": 9.980352038774119e-05, "loss": -0.1159, "step": 4610 }, { "epoch": 12.833333333333334, "grad_norm": 0.4440118074417114, "learning_rate": 9.98010721666836e-05, "loss": -0.1211, "step": 4620 }, { "epoch": 12.86111111111111, "grad_norm": 0.4516400396823883, "learning_rate": 9.979860881737523e-05, "loss": -0.134, "step": 4630 }, { "epoch": 12.88888888888889, "grad_norm": 0.43172216415405273, "learning_rate": 9.979613034056434e-05, "loss": -0.1259, "step": 4640 }, { "epoch": 12.916666666666666, "grad_norm": 0.48092854022979736, "learning_rate": 9.979363673700386e-05, "loss": -0.1124, "step": 4650 }, { "epoch": 12.944444444444445, "grad_norm": 0.4908120632171631, "learning_rate": 9.979112800745124e-05, "loss": -0.1154, "step": 4660 }, { "epoch": 12.972222222222221, "grad_norm": 0.5490595698356628, "learning_rate": 9.978860415266861e-05, "loss": -0.1193, "step": 4670 }, { "epoch": 13.0, "grad_norm": 0.5648190379142761, "learning_rate": 9.978606517342262e-05, "loss": -0.126, "step": 4680 }, { "epoch": 13.027777777777779, "grad_norm": 0.4429938793182373, "learning_rate": 9.978351107048456e-05, "loss": -0.1192, "step": 4690 }, { "epoch": 13.055555555555555, "grad_norm": 0.5294462442398071, "learning_rate": 9.978094184463029e-05, "loss": -0.1198, "step": 4700 }, { "epoch": 13.083333333333334, "grad_norm": 0.48171648383140564, "learning_rate": 9.977835749664029e-05, "loss": -0.1238, "step": 4710 }, { "epoch": 13.11111111111111, "grad_norm": 0.5057381987571716, "learning_rate": 9.97757580272996e-05, "loss": -0.1302, "step": 4720 }, { "epoch": 13.13888888888889, "grad_norm": 0.3842191994190216, "learning_rate": 9.977314343739786e-05, "loss": -0.1321, "step": 4730 }, { "epoch": 13.166666666666666, "grad_norm": 0.5926108956336975, "learning_rate": 9.977051372772934e-05, "loss": -0.1249, "step": 4740 }, { "epoch": 13.194444444444445, "grad_norm": 0.45047727227211, "learning_rate": 9.976786889909286e-05, "loss": -0.1347, "step": 4750 }, { "epoch": 13.222222222222221, "grad_norm": 0.6141397356987, "learning_rate": 9.976520895229185e-05, "loss": -0.1254, "step": 4760 }, { "epoch": 13.25, "grad_norm": 0.59272301197052, "learning_rate": 9.976253388813433e-05, "loss": -0.1229, "step": 4770 }, { "epoch": 13.277777777777779, "grad_norm": 0.5005635619163513, "learning_rate": 9.975984370743293e-05, "loss": -0.1227, "step": 4780 }, { "epoch": 13.305555555555555, "grad_norm": 0.3446647822856903, "learning_rate": 9.975713841100485e-05, "loss": -0.1171, "step": 4790 }, { "epoch": 13.333333333333334, "grad_norm": 0.514933168888092, "learning_rate": 9.975441799967187e-05, "loss": -0.1299, "step": 4800 }, { "epoch": 13.36111111111111, "grad_norm": 0.517482578754425, "learning_rate": 9.975168247426039e-05, "loss": -0.1247, "step": 4810 }, { "epoch": 13.38888888888889, "grad_norm": 0.6293244361877441, "learning_rate": 9.974893183560139e-05, "loss": -0.1278, "step": 4820 }, { "epoch": 13.416666666666666, "grad_norm": 0.4439452290534973, "learning_rate": 9.974616608453045e-05, "loss": -0.1255, "step": 4830 }, { "epoch": 13.444444444444445, "grad_norm": 0.47971463203430176, "learning_rate": 9.974338522188772e-05, "loss": -0.1289, "step": 4840 }, { "epoch": 13.472222222222221, "grad_norm": 0.48668771982192993, "learning_rate": 9.974058924851797e-05, "loss": -0.1309, "step": 4850 }, { "epoch": 13.5, "grad_norm": 0.6717028021812439, "learning_rate": 9.973777816527051e-05, "loss": -0.1208, "step": 4860 }, { "epoch": 13.527777777777779, "grad_norm": 0.6766639351844788, "learning_rate": 9.973495197299931e-05, "loss": -0.1133, "step": 4870 }, { "epoch": 13.555555555555555, "grad_norm": 0.5367018580436707, "learning_rate": 9.973211067256287e-05, "loss": -0.1173, "step": 4880 }, { "epoch": 13.583333333333334, "grad_norm": 0.6262452602386475, "learning_rate": 9.97292542648243e-05, "loss": -0.1162, "step": 4890 }, { "epoch": 13.61111111111111, "grad_norm": 0.5388481020927429, "learning_rate": 9.972638275065131e-05, "loss": -0.1203, "step": 4900 }, { "epoch": 13.63888888888889, "grad_norm": 0.6027723550796509, "learning_rate": 9.972349613091621e-05, "loss": -0.1158, "step": 4910 }, { "epoch": 13.666666666666666, "grad_norm": 0.43352001905441284, "learning_rate": 9.972059440649584e-05, "loss": -0.1215, "step": 4920 }, { "epoch": 13.694444444444445, "grad_norm": 0.45678845047950745, "learning_rate": 9.971767757827168e-05, "loss": -0.1257, "step": 4930 }, { "epoch": 13.722222222222221, "grad_norm": 0.4217965006828308, "learning_rate": 9.971474564712982e-05, "loss": -0.123, "step": 4940 }, { "epoch": 13.75, "grad_norm": 0.3742971420288086, "learning_rate": 9.971179861396084e-05, "loss": -0.1282, "step": 4950 }, { "epoch": 13.777777777777779, "grad_norm": 0.5471207499504089, "learning_rate": 9.970883647966003e-05, "loss": -0.1216, "step": 4960 }, { "epoch": 13.805555555555555, "grad_norm": 0.5101257562637329, "learning_rate": 9.970585924512717e-05, "loss": -0.1314, "step": 4970 }, { "epoch": 13.833333333333334, "grad_norm": 0.48422783613204956, "learning_rate": 9.970286691126669e-05, "loss": -0.1343, "step": 4980 }, { "epoch": 13.86111111111111, "grad_norm": 0.35367825627326965, "learning_rate": 9.969985947898756e-05, "loss": -0.1252, "step": 4990 }, { "epoch": 13.88888888888889, "grad_norm": 0.5191375017166138, "learning_rate": 9.969683694920337e-05, "loss": -0.1217, "step": 5000 }, { "epoch": 13.916666666666666, "grad_norm": 0.43086957931518555, "learning_rate": 9.969379932283228e-05, "loss": -0.1211, "step": 5010 }, { "epoch": 13.944444444444445, "grad_norm": 0.5061451196670532, "learning_rate": 9.969074660079704e-05, "loss": -0.1246, "step": 5020 }, { "epoch": 13.972222222222221, "grad_norm": 0.5424769520759583, "learning_rate": 9.968767878402501e-05, "loss": -0.1219, "step": 5030 }, { "epoch": 14.0, "grad_norm": 0.5833485126495361, "learning_rate": 9.968459587344808e-05, "loss": -0.1306, "step": 5040 }, { "epoch": 14.027777777777779, "grad_norm": 0.6181188225746155, "learning_rate": 9.968149787000278e-05, "loss": -0.1265, "step": 5050 }, { "epoch": 14.055555555555555, "grad_norm": 0.5004115104675293, "learning_rate": 9.967838477463018e-05, "loss": -0.1342, "step": 5060 }, { "epoch": 14.083333333333334, "grad_norm": 0.4783470332622528, "learning_rate": 9.967525658827597e-05, "loss": -0.1254, "step": 5070 }, { "epoch": 14.11111111111111, "grad_norm": 0.3687458634376526, "learning_rate": 9.967211331189042e-05, "loss": -0.1279, "step": 5080 }, { "epoch": 14.13888888888889, "grad_norm": 0.37028512358665466, "learning_rate": 9.966895494642834e-05, "loss": -0.1252, "step": 5090 }, { "epoch": 14.166666666666666, "grad_norm": 0.5844200849533081, "learning_rate": 9.96657814928492e-05, "loss": -0.1187, "step": 5100 }, { "epoch": 14.194444444444445, "grad_norm": 0.27186113595962524, "learning_rate": 9.966259295211697e-05, "loss": -0.1304, "step": 5110 }, { "epoch": 14.222222222222221, "grad_norm": 0.3409327268600464, "learning_rate": 9.965938932520028e-05, "loss": -0.1255, "step": 5120 }, { "epoch": 14.25, "grad_norm": 0.5472763776779175, "learning_rate": 9.965617061307229e-05, "loss": -0.1354, "step": 5130 }, { "epoch": 14.277777777777779, "grad_norm": 0.4737846255302429, "learning_rate": 9.965293681671077e-05, "loss": -0.1281, "step": 5140 }, { "epoch": 14.305555555555555, "grad_norm": 0.6007692217826843, "learning_rate": 9.964968793709804e-05, "loss": -0.1303, "step": 5150 }, { "epoch": 14.333333333333334, "grad_norm": 0.4768265187740326, "learning_rate": 9.964642397522106e-05, "loss": -0.133, "step": 5160 }, { "epoch": 14.36111111111111, "grad_norm": 0.4730849862098694, "learning_rate": 9.96431449320713e-05, "loss": -0.1317, "step": 5170 }, { "epoch": 14.38888888888889, "grad_norm": 0.44660112261772156, "learning_rate": 9.963985080864486e-05, "loss": -0.1288, "step": 5180 }, { "epoch": 14.416666666666666, "grad_norm": 0.43881914019584656, "learning_rate": 9.96365416059424e-05, "loss": -0.1323, "step": 5190 }, { "epoch": 14.444444444444445, "grad_norm": 0.3515077233314514, "learning_rate": 9.963321732496919e-05, "loss": -0.1306, "step": 5200 }, { "epoch": 14.472222222222221, "grad_norm": 0.47888749837875366, "learning_rate": 9.962987796673506e-05, "loss": -0.1289, "step": 5210 }, { "epoch": 14.5, "grad_norm": 0.5037688612937927, "learning_rate": 9.962652353225438e-05, "loss": -0.1384, "step": 5220 }, { "epoch": 14.527777777777779, "grad_norm": 0.6378742456436157, "learning_rate": 9.962315402254619e-05, "loss": -0.1223, "step": 5230 }, { "epoch": 14.555555555555555, "grad_norm": 0.3622201085090637, "learning_rate": 9.9619769438634e-05, "loss": -0.1334, "step": 5240 }, { "epoch": 14.583333333333334, "grad_norm": 0.43517184257507324, "learning_rate": 9.9616369781546e-05, "loss": -0.1395, "step": 5250 }, { "epoch": 14.61111111111111, "grad_norm": 0.39632517099380493, "learning_rate": 9.961295505231491e-05, "loss": -0.1347, "step": 5260 }, { "epoch": 14.63888888888889, "grad_norm": 0.6208352446556091, "learning_rate": 9.960952525197804e-05, "loss": -0.1332, "step": 5270 }, { "epoch": 14.666666666666666, "grad_norm": 0.5603777170181274, "learning_rate": 9.960608038157724e-05, "loss": -0.143, "step": 5280 }, { "epoch": 14.694444444444445, "grad_norm": 0.4725254476070404, "learning_rate": 9.960262044215901e-05, "loss": -0.1305, "step": 5290 }, { "epoch": 14.722222222222221, "grad_norm": 0.38174089789390564, "learning_rate": 9.959914543477435e-05, "loss": -0.1256, "step": 5300 }, { "epoch": 14.75, "grad_norm": 0.5838874578475952, "learning_rate": 9.959565536047892e-05, "loss": -0.1123, "step": 5310 }, { "epoch": 14.777777777777779, "grad_norm": 0.6901920437812805, "learning_rate": 9.959215022033288e-05, "loss": -0.1279, "step": 5320 }, { "epoch": 14.805555555555555, "grad_norm": 0.5985168814659119, "learning_rate": 9.9588630015401e-05, "loss": -0.12, "step": 5330 }, { "epoch": 14.833333333333334, "grad_norm": 0.4877694547176361, "learning_rate": 9.958509474675264e-05, "loss": -0.131, "step": 5340 }, { "epoch": 14.86111111111111, "grad_norm": 0.3975989818572998, "learning_rate": 9.958154441546171e-05, "loss": -0.1191, "step": 5350 }, { "epoch": 14.88888888888889, "grad_norm": 0.5359532237052917, "learning_rate": 9.957797902260673e-05, "loss": -0.1318, "step": 5360 }, { "epoch": 14.916666666666666, "grad_norm": 0.6772318482398987, "learning_rate": 9.957439856927073e-05, "loss": -0.1343, "step": 5370 }, { "epoch": 14.944444444444445, "grad_norm": 0.4812149405479431, "learning_rate": 9.957080305654139e-05, "loss": -0.1271, "step": 5380 }, { "epoch": 14.972222222222221, "grad_norm": 0.4714261591434479, "learning_rate": 9.956719248551092e-05, "loss": -0.1226, "step": 5390 }, { "epoch": 15.0, "grad_norm": 0.2996251881122589, "learning_rate": 9.956356685727612e-05, "loss": -0.1261, "step": 5400 }, { "epoch": 15.027777777777779, "grad_norm": 0.41562196612358093, "learning_rate": 9.955992617293836e-05, "loss": -0.1327, "step": 5410 }, { "epoch": 15.055555555555555, "grad_norm": 0.4774347245693207, "learning_rate": 9.955627043360358e-05, "loss": -0.1302, "step": 5420 }, { "epoch": 15.083333333333334, "grad_norm": 0.4904557466506958, "learning_rate": 9.955259964038231e-05, "loss": -0.1369, "step": 5430 }, { "epoch": 15.11111111111111, "grad_norm": 0.5336781740188599, "learning_rate": 9.954891379438962e-05, "loss": -0.1275, "step": 5440 }, { "epoch": 15.13888888888889, "grad_norm": 0.49999427795410156, "learning_rate": 9.954521289674519e-05, "loss": -0.122, "step": 5450 }, { "epoch": 15.166666666666666, "grad_norm": 0.4562298059463501, "learning_rate": 9.954149694857325e-05, "loss": -0.1257, "step": 5460 }, { "epoch": 15.194444444444445, "grad_norm": 0.4306125044822693, "learning_rate": 9.953776595100258e-05, "loss": -0.1195, "step": 5470 }, { "epoch": 15.222222222222221, "grad_norm": 0.5974892377853394, "learning_rate": 9.95340199051666e-05, "loss": -0.1309, "step": 5480 }, { "epoch": 15.25, "grad_norm": 0.31931719183921814, "learning_rate": 9.953025881220325e-05, "loss": -0.1352, "step": 5490 }, { "epoch": 15.277777777777779, "grad_norm": 0.4555600881576538, "learning_rate": 9.952648267325504e-05, "loss": -0.1328, "step": 5500 }, { "epoch": 15.305555555555555, "grad_norm": 0.5235995054244995, "learning_rate": 9.952269148946905e-05, "loss": -0.1355, "step": 5510 }, { "epoch": 15.333333333333334, "grad_norm": 0.47546157240867615, "learning_rate": 9.951888526199697e-05, "loss": -0.1319, "step": 5520 }, { "epoch": 15.36111111111111, "grad_norm": 0.5221265554428101, "learning_rate": 9.951506399199501e-05, "loss": -0.1308, "step": 5530 }, { "epoch": 15.38888888888889, "grad_norm": 0.43941229581832886, "learning_rate": 9.951122768062399e-05, "loss": -0.1362, "step": 5540 }, { "epoch": 15.416666666666666, "grad_norm": 0.4467078447341919, "learning_rate": 9.950737632904927e-05, "loss": -0.1342, "step": 5550 }, { "epoch": 15.444444444444445, "grad_norm": 0.47153759002685547, "learning_rate": 9.950350993844077e-05, "loss": -0.131, "step": 5560 }, { "epoch": 15.472222222222221, "grad_norm": 0.4099963903427124, "learning_rate": 9.949962850997303e-05, "loss": -0.1344, "step": 5570 }, { "epoch": 15.5, "grad_norm": 0.3872981071472168, "learning_rate": 9.949573204482512e-05, "loss": -0.1404, "step": 5580 }, { "epoch": 15.527777777777779, "grad_norm": 0.6276242733001709, "learning_rate": 9.949182054418064e-05, "loss": -0.1348, "step": 5590 }, { "epoch": 15.555555555555555, "grad_norm": 0.38127759099006653, "learning_rate": 9.948789400922787e-05, "loss": -0.1331, "step": 5600 }, { "epoch": 15.583333333333334, "grad_norm": 0.42733234167099, "learning_rate": 9.948395244115953e-05, "loss": -0.1348, "step": 5610 }, { "epoch": 15.61111111111111, "grad_norm": 0.44086727499961853, "learning_rate": 9.9479995841173e-05, "loss": -0.1324, "step": 5620 }, { "epoch": 15.63888888888889, "grad_norm": 0.35670581459999084, "learning_rate": 9.947602421047017e-05, "loss": -0.1461, "step": 5630 }, { "epoch": 15.666666666666666, "grad_norm": 0.5426246523857117, "learning_rate": 9.947203755025753e-05, "loss": -0.1379, "step": 5640 }, { "epoch": 15.694444444444445, "grad_norm": 0.44396817684173584, "learning_rate": 9.946803586174611e-05, "loss": -0.1349, "step": 5650 }, { "epoch": 15.722222222222221, "grad_norm": 0.4663013219833374, "learning_rate": 9.946401914615151e-05, "loss": -0.137, "step": 5660 }, { "epoch": 15.75, "grad_norm": 0.33873453736305237, "learning_rate": 9.945998740469394e-05, "loss": -0.1445, "step": 5670 }, { "epoch": 15.777777777777779, "grad_norm": 0.41419801115989685, "learning_rate": 9.945594063859809e-05, "loss": -0.1257, "step": 5680 }, { "epoch": 15.805555555555555, "grad_norm": 0.5133129358291626, "learning_rate": 9.94518788490933e-05, "loss": -0.1443, "step": 5690 }, { "epoch": 15.833333333333334, "grad_norm": 0.6071870923042297, "learning_rate": 9.944780203741341e-05, "loss": -0.1304, "step": 5700 }, { "epoch": 15.86111111111111, "grad_norm": 0.24243807792663574, "learning_rate": 9.944371020479686e-05, "loss": -0.1372, "step": 5710 }, { "epoch": 15.88888888888889, "grad_norm": 0.5185979008674622, "learning_rate": 9.943960335248662e-05, "loss": -0.1434, "step": 5720 }, { "epoch": 15.916666666666666, "grad_norm": 0.29396089911460876, "learning_rate": 9.943548148173027e-05, "loss": -0.1354, "step": 5730 }, { "epoch": 15.944444444444445, "grad_norm": 0.5652161240577698, "learning_rate": 9.943134459377992e-05, "loss": -0.1431, "step": 5740 }, { "epoch": 15.972222222222221, "grad_norm": 0.5221500992774963, "learning_rate": 9.942719268989222e-05, "loss": -0.1313, "step": 5750 }, { "epoch": 16.0, "grad_norm": 0.40446457266807556, "learning_rate": 9.942302577132844e-05, "loss": -0.1395, "step": 5760 }, { "epoch": 16.02777777777778, "grad_norm": 0.7018918395042419, "learning_rate": 9.941884383935438e-05, "loss": -0.1421, "step": 5770 }, { "epoch": 16.055555555555557, "grad_norm": 0.34028080105781555, "learning_rate": 9.941464689524039e-05, "loss": -0.1359, "step": 5780 }, { "epoch": 16.083333333333332, "grad_norm": 0.4003239870071411, "learning_rate": 9.941043494026139e-05, "loss": -0.1398, "step": 5790 }, { "epoch": 16.11111111111111, "grad_norm": 0.47954604029655457, "learning_rate": 9.940620797569685e-05, "loss": -0.139, "step": 5800 }, { "epoch": 16.13888888888889, "grad_norm": 0.5684925317764282, "learning_rate": 9.940196600283082e-05, "loss": -0.1169, "step": 5810 }, { "epoch": 16.166666666666668, "grad_norm": 0.3966541290283203, "learning_rate": 9.939770902295192e-05, "loss": -0.1251, "step": 5820 }, { "epoch": 16.194444444444443, "grad_norm": 0.41064414381980896, "learning_rate": 9.939343703735329e-05, "loss": -0.1345, "step": 5830 }, { "epoch": 16.22222222222222, "grad_norm": 0.43365800380706787, "learning_rate": 9.938915004733264e-05, "loss": -0.1358, "step": 5840 }, { "epoch": 16.25, "grad_norm": 0.5441865921020508, "learning_rate": 9.938484805419224e-05, "loss": -0.1355, "step": 5850 }, { "epoch": 16.27777777777778, "grad_norm": 0.34935006499290466, "learning_rate": 9.938053105923894e-05, "loss": -0.1371, "step": 5860 }, { "epoch": 16.305555555555557, "grad_norm": 0.5272371172904968, "learning_rate": 9.937619906378413e-05, "loss": -0.1406, "step": 5870 }, { "epoch": 16.333333333333332, "grad_norm": 0.5357279181480408, "learning_rate": 9.937185206914374e-05, "loss": -0.1331, "step": 5880 }, { "epoch": 16.36111111111111, "grad_norm": 0.43050140142440796, "learning_rate": 9.936749007663829e-05, "loss": -0.1395, "step": 5890 }, { "epoch": 16.38888888888889, "grad_norm": 0.833681046962738, "learning_rate": 9.93631130875928e-05, "loss": -0.1356, "step": 5900 }, { "epoch": 16.416666666666668, "grad_norm": 0.27266615629196167, "learning_rate": 9.935872110333692e-05, "loss": -0.1285, "step": 5910 }, { "epoch": 16.444444444444443, "grad_norm": 0.5499081015586853, "learning_rate": 9.935431412520484e-05, "loss": -0.1266, "step": 5920 }, { "epoch": 16.47222222222222, "grad_norm": 0.40392065048217773, "learning_rate": 9.934989215453523e-05, "loss": -0.1286, "step": 5930 }, { "epoch": 16.5, "grad_norm": 0.38093435764312744, "learning_rate": 9.934545519267139e-05, "loss": -0.1317, "step": 5940 }, { "epoch": 16.52777777777778, "grad_norm": 0.4575217366218567, "learning_rate": 9.934100324096117e-05, "loss": -0.1345, "step": 5950 }, { "epoch": 16.555555555555557, "grad_norm": 0.3249002993106842, "learning_rate": 9.933653630075692e-05, "loss": -0.1403, "step": 5960 }, { "epoch": 16.583333333333332, "grad_norm": 0.3946460783481598, "learning_rate": 9.93320543734156e-05, "loss": -0.1385, "step": 5970 }, { "epoch": 16.61111111111111, "grad_norm": 0.3964657187461853, "learning_rate": 9.932755746029871e-05, "loss": -0.144, "step": 5980 }, { "epoch": 16.63888888888889, "grad_norm": 0.41730445623397827, "learning_rate": 9.932304556277228e-05, "loss": -0.1328, "step": 5990 }, { "epoch": 16.666666666666668, "grad_norm": 0.5840293169021606, "learning_rate": 9.93185186822069e-05, "loss": -0.1353, "step": 6000 }, { "epoch": 16.694444444444443, "grad_norm": 0.48755934834480286, "learning_rate": 9.931397681997773e-05, "loss": -0.1356, "step": 6010 }, { "epoch": 16.72222222222222, "grad_norm": 0.35712528228759766, "learning_rate": 9.930941997746446e-05, "loss": -0.1364, "step": 6020 }, { "epoch": 16.75, "grad_norm": 0.3982155919075012, "learning_rate": 9.930484815605134e-05, "loss": -0.1351, "step": 6030 }, { "epoch": 16.77777777777778, "grad_norm": 0.5007644295692444, "learning_rate": 9.930026135712717e-05, "loss": -0.1287, "step": 6040 }, { "epoch": 16.805555555555557, "grad_norm": 0.33968040347099304, "learning_rate": 9.92956595820853e-05, "loss": -0.1404, "step": 6050 }, { "epoch": 16.833333333333332, "grad_norm": 0.4691331386566162, "learning_rate": 9.929104283232362e-05, "loss": -0.1422, "step": 6060 }, { "epoch": 16.86111111111111, "grad_norm": 0.37692317366600037, "learning_rate": 9.92864111092446e-05, "loss": -0.1409, "step": 6070 }, { "epoch": 16.88888888888889, "grad_norm": 0.46686163544654846, "learning_rate": 9.92817644142552e-05, "loss": -0.1473, "step": 6080 }, { "epoch": 16.916666666666668, "grad_norm": 0.5855314135551453, "learning_rate": 9.927710274876698e-05, "loss": -0.1356, "step": 6090 }, { "epoch": 16.944444444444443, "grad_norm": 0.4896209239959717, "learning_rate": 9.927242611419603e-05, "loss": -0.1353, "step": 6100 }, { "epoch": 16.97222222222222, "grad_norm": 0.4652169942855835, "learning_rate": 9.926773451196301e-05, "loss": -0.1331, "step": 6110 }, { "epoch": 17.0, "grad_norm": 0.3237535059452057, "learning_rate": 9.926302794349306e-05, "loss": -0.1386, "step": 6120 }, { "epoch": 17.02777777777778, "grad_norm": 0.5259121656417847, "learning_rate": 9.925830641021594e-05, "loss": -0.127, "step": 6130 }, { "epoch": 17.055555555555557, "grad_norm": 0.36816170811653137, "learning_rate": 9.925356991356593e-05, "loss": -0.141, "step": 6140 }, { "epoch": 17.083333333333332, "grad_norm": 0.4578109085559845, "learning_rate": 9.924881845498184e-05, "loss": -0.1418, "step": 6150 }, { "epoch": 17.11111111111111, "grad_norm": 0.5200081467628479, "learning_rate": 9.924405203590705e-05, "loss": -0.1365, "step": 6160 }, { "epoch": 17.13888888888889, "grad_norm": 0.3880941867828369, "learning_rate": 9.923927065778946e-05, "loss": -0.14, "step": 6170 }, { "epoch": 17.166666666666668, "grad_norm": 0.5012039542198181, "learning_rate": 9.923447432208154e-05, "loss": -0.1381, "step": 6180 }, { "epoch": 17.194444444444443, "grad_norm": 0.45803701877593994, "learning_rate": 9.922966303024027e-05, "loss": -0.1412, "step": 6190 }, { "epoch": 17.22222222222222, "grad_norm": 0.5198766589164734, "learning_rate": 9.922483678372721e-05, "loss": -0.1284, "step": 6200 }, { "epoch": 17.25, "grad_norm": 0.5361096858978271, "learning_rate": 9.921999558400845e-05, "loss": -0.1362, "step": 6210 }, { "epoch": 17.27777777777778, "grad_norm": 0.5273212194442749, "learning_rate": 9.92151394325546e-05, "loss": -0.1257, "step": 6220 }, { "epoch": 17.305555555555557, "grad_norm": 0.422460675239563, "learning_rate": 9.921026833084084e-05, "loss": -0.1377, "step": 6230 }, { "epoch": 17.333333333333332, "grad_norm": 0.42248332500457764, "learning_rate": 9.920538228034689e-05, "loss": -0.1444, "step": 6240 }, { "epoch": 17.36111111111111, "grad_norm": 0.39385494589805603, "learning_rate": 9.920048128255699e-05, "loss": -0.1381, "step": 6250 }, { "epoch": 17.38888888888889, "grad_norm": 0.6813430190086365, "learning_rate": 9.919556533895995e-05, "loss": -0.1278, "step": 6260 }, { "epoch": 17.416666666666668, "grad_norm": 0.3151148855686188, "learning_rate": 9.919063445104907e-05, "loss": -0.133, "step": 6270 }, { "epoch": 17.444444444444443, "grad_norm": 0.39923587441444397, "learning_rate": 9.918568862032227e-05, "loss": -0.1358, "step": 6280 }, { "epoch": 17.47222222222222, "grad_norm": 0.3692784607410431, "learning_rate": 9.918072784828194e-05, "loss": -0.1402, "step": 6290 }, { "epoch": 17.5, "grad_norm": 0.36497068405151367, "learning_rate": 9.917575213643501e-05, "loss": -0.1346, "step": 6300 }, { "epoch": 17.52777777777778, "grad_norm": 0.33223217725753784, "learning_rate": 9.917076148629302e-05, "loss": -0.142, "step": 6310 }, { "epoch": 17.555555555555557, "grad_norm": 0.44713208079338074, "learning_rate": 9.916575589937196e-05, "loss": -0.1381, "step": 6320 }, { "epoch": 17.583333333333332, "grad_norm": 0.6046751141548157, "learning_rate": 9.916073537719239e-05, "loss": -0.1357, "step": 6330 }, { "epoch": 17.61111111111111, "grad_norm": 0.4948788583278656, "learning_rate": 9.915569992127944e-05, "loss": -0.1367, "step": 6340 }, { "epoch": 17.63888888888889, "grad_norm": 0.3981380760669708, "learning_rate": 9.915064953316273e-05, "loss": -0.138, "step": 6350 }, { "epoch": 17.666666666666668, "grad_norm": 0.31315040588378906, "learning_rate": 9.914558421437645e-05, "loss": -0.1454, "step": 6360 }, { "epoch": 17.694444444444443, "grad_norm": 0.36841174960136414, "learning_rate": 9.914050396645929e-05, "loss": -0.1408, "step": 6370 }, { "epoch": 17.72222222222222, "grad_norm": 0.43821316957473755, "learning_rate": 9.913540879095452e-05, "loss": -0.1372, "step": 6380 }, { "epoch": 17.75, "grad_norm": 0.3066030740737915, "learning_rate": 9.913029868940987e-05, "loss": -0.1319, "step": 6390 }, { "epoch": 17.77777777777778, "grad_norm": 0.45994630455970764, "learning_rate": 9.912517366337772e-05, "loss": -0.1382, "step": 6400 }, { "epoch": 17.805555555555557, "grad_norm": 0.5813340544700623, "learning_rate": 9.912003371441487e-05, "loss": -0.1353, "step": 6410 }, { "epoch": 17.833333333333332, "grad_norm": 0.4117131531238556, "learning_rate": 9.911487884408271e-05, "loss": -0.1393, "step": 6420 }, { "epoch": 17.86111111111111, "grad_norm": 0.4558712840080261, "learning_rate": 9.910970905394719e-05, "loss": -0.1363, "step": 6430 }, { "epoch": 17.88888888888889, "grad_norm": 0.5122381448745728, "learning_rate": 9.91045243455787e-05, "loss": -0.1467, "step": 6440 }, { "epoch": 17.916666666666668, "grad_norm": 0.40141814947128296, "learning_rate": 9.909932472055225e-05, "loss": -0.1317, "step": 6450 }, { "epoch": 17.944444444444443, "grad_norm": 0.29721930623054504, "learning_rate": 9.909411018044734e-05, "loss": -0.1499, "step": 6460 }, { "epoch": 17.97222222222222, "grad_norm": 0.30673447251319885, "learning_rate": 9.908888072684802e-05, "loss": -0.1423, "step": 6470 }, { "epoch": 18.0, "grad_norm": 0.36537978053092957, "learning_rate": 9.908363636134285e-05, "loss": -0.1514, "step": 6480 }, { "epoch": 18.02777777777778, "grad_norm": 0.41709598898887634, "learning_rate": 9.907837708552493e-05, "loss": -0.1506, "step": 6490 }, { "epoch": 18.055555555555557, "grad_norm": 0.3581623435020447, "learning_rate": 9.90731029009919e-05, "loss": -0.1432, "step": 6500 }, { "epoch": 18.083333333333332, "grad_norm": 0.5597764253616333, "learning_rate": 9.906781380934589e-05, "loss": -0.128, "step": 6510 }, { "epoch": 18.11111111111111, "grad_norm": 0.46867576241493225, "learning_rate": 9.906250981219362e-05, "loss": -0.1361, "step": 6520 }, { "epoch": 18.13888888888889, "grad_norm": 0.5021114945411682, "learning_rate": 9.905719091114628e-05, "loss": -0.1457, "step": 6530 }, { "epoch": 18.166666666666668, "grad_norm": 0.6537953019142151, "learning_rate": 9.905185710781964e-05, "loss": -0.1347, "step": 6540 }, { "epoch": 18.194444444444443, "grad_norm": 0.2887960970401764, "learning_rate": 9.904650840383392e-05, "loss": -0.1399, "step": 6550 }, { "epoch": 18.22222222222222, "grad_norm": 0.30284398794174194, "learning_rate": 9.904114480081397e-05, "loss": -0.1464, "step": 6560 }, { "epoch": 18.25, "grad_norm": 0.5141924023628235, "learning_rate": 9.903576630038906e-05, "loss": -0.1319, "step": 6570 }, { "epoch": 18.27777777777778, "grad_norm": 0.31734079122543335, "learning_rate": 9.903037290419309e-05, "loss": -0.1432, "step": 6580 }, { "epoch": 18.305555555555557, "grad_norm": 0.3333049416542053, "learning_rate": 9.902496461386439e-05, "loss": -0.143, "step": 6590 }, { "epoch": 18.333333333333332, "grad_norm": 0.5346686840057373, "learning_rate": 9.901954143104588e-05, "loss": -0.1459, "step": 6600 }, { "epoch": 18.36111111111111, "grad_norm": 0.4380529224872589, "learning_rate": 9.901410335738496e-05, "loss": -0.1521, "step": 6610 }, { "epoch": 18.38888888888889, "grad_norm": 0.46393218636512756, "learning_rate": 9.900865039453358e-05, "loss": -0.1374, "step": 6620 }, { "epoch": 18.416666666666668, "grad_norm": 0.4175238013267517, "learning_rate": 9.900318254414821e-05, "loss": -0.1435, "step": 6630 }, { "epoch": 18.444444444444443, "grad_norm": 0.25873473286628723, "learning_rate": 9.899769980788985e-05, "loss": -0.1438, "step": 6640 }, { "epoch": 18.47222222222222, "grad_norm": 0.33409032225608826, "learning_rate": 9.899220218742398e-05, "loss": -0.1488, "step": 6650 }, { "epoch": 18.5, "grad_norm": 0.6090133190155029, "learning_rate": 9.898668968442066e-05, "loss": -0.1344, "step": 6660 }, { "epoch": 18.52777777777778, "grad_norm": 0.40431517362594604, "learning_rate": 9.898116230055443e-05, "loss": -0.1396, "step": 6670 }, { "epoch": 18.555555555555557, "grad_norm": 0.3312325179576874, "learning_rate": 9.897562003750437e-05, "loss": -0.1443, "step": 6680 }, { "epoch": 18.583333333333332, "grad_norm": 0.38539063930511475, "learning_rate": 9.897006289695407e-05, "loss": -0.1399, "step": 6690 }, { "epoch": 18.61111111111111, "grad_norm": 0.38787275552749634, "learning_rate": 9.896449088059164e-05, "loss": -0.1355, "step": 6700 }, { "epoch": 18.63888888888889, "grad_norm": 0.46342945098876953, "learning_rate": 9.89589039901097e-05, "loss": -0.1405, "step": 6710 }, { "epoch": 18.666666666666668, "grad_norm": 0.39082106947898865, "learning_rate": 9.895330222720542e-05, "loss": -0.1484, "step": 6720 }, { "epoch": 18.694444444444443, "grad_norm": 0.3803313374519348, "learning_rate": 9.894768559358047e-05, "loss": -0.1326, "step": 6730 }, { "epoch": 18.72222222222222, "grad_norm": 0.38885313272476196, "learning_rate": 9.894205409094101e-05, "loss": -0.1313, "step": 6740 }, { "epoch": 18.75, "grad_norm": 0.2805618643760681, "learning_rate": 9.893640772099777e-05, "loss": -0.1465, "step": 6750 }, { "epoch": 18.77777777777778, "grad_norm": 0.4753136932849884, "learning_rate": 9.893074648546595e-05, "loss": -0.1319, "step": 6760 }, { "epoch": 18.805555555555557, "grad_norm": 0.4807836711406708, "learning_rate": 9.892507038606528e-05, "loss": -0.1359, "step": 6770 }, { "epoch": 18.833333333333332, "grad_norm": 0.44244882464408875, "learning_rate": 9.891937942452003e-05, "loss": -0.1417, "step": 6780 }, { "epoch": 18.86111111111111, "grad_norm": 0.3863269090652466, "learning_rate": 9.891367360255895e-05, "loss": -0.1449, "step": 6790 }, { "epoch": 18.88888888888889, "grad_norm": 0.5837173461914062, "learning_rate": 9.890795292191532e-05, "loss": -0.1218, "step": 6800 }, { "epoch": 18.916666666666668, "grad_norm": 0.4620624780654907, "learning_rate": 9.890221738432694e-05, "loss": -0.1336, "step": 6810 }, { "epoch": 18.944444444444443, "grad_norm": 0.36331647634506226, "learning_rate": 9.88964669915361e-05, "loss": -0.1438, "step": 6820 }, { "epoch": 18.97222222222222, "grad_norm": 0.46349725127220154, "learning_rate": 9.889070174528963e-05, "loss": -0.1367, "step": 6830 }, { "epoch": 19.0, "grad_norm": 0.3852502405643463, "learning_rate": 9.888492164733883e-05, "loss": -0.1418, "step": 6840 }, { "epoch": 19.02777777777778, "grad_norm": 0.4819332957267761, "learning_rate": 9.88791266994396e-05, "loss": -0.1395, "step": 6850 }, { "epoch": 19.055555555555557, "grad_norm": 0.7132675051689148, "learning_rate": 9.887331690335223e-05, "loss": -0.1454, "step": 6860 }, { "epoch": 19.083333333333332, "grad_norm": 0.38891148567199707, "learning_rate": 9.886749226084163e-05, "loss": -0.15, "step": 6870 }, { "epoch": 19.11111111111111, "grad_norm": 0.5108673572540283, "learning_rate": 9.886165277367714e-05, "loss": -0.1361, "step": 6880 }, { "epoch": 19.13888888888889, "grad_norm": 0.45976129174232483, "learning_rate": 9.885579844363265e-05, "loss": -0.1465, "step": 6890 }, { "epoch": 19.166666666666668, "grad_norm": 0.49082911014556885, "learning_rate": 9.884992927248656e-05, "loss": -0.1411, "step": 6900 }, { "epoch": 19.194444444444443, "grad_norm": 0.48486700654029846, "learning_rate": 9.884404526202178e-05, "loss": -0.1373, "step": 6910 }, { "epoch": 19.22222222222222, "grad_norm": 0.42545291781425476, "learning_rate": 9.883814641402568e-05, "loss": -0.1308, "step": 6920 }, { "epoch": 19.25, "grad_norm": 0.2969660758972168, "learning_rate": 9.88322327302902e-05, "loss": -0.1462, "step": 6930 }, { "epoch": 19.27777777777778, "grad_norm": 0.1993880271911621, "learning_rate": 9.882630421261176e-05, "loss": -0.1444, "step": 6940 }, { "epoch": 19.305555555555557, "grad_norm": 0.30603787302970886, "learning_rate": 9.88203608627913e-05, "loss": -0.141, "step": 6950 }, { "epoch": 19.333333333333332, "grad_norm": 0.37342146039009094, "learning_rate": 9.881440268263422e-05, "loss": -0.1411, "step": 6960 }, { "epoch": 19.36111111111111, "grad_norm": 0.2570522725582123, "learning_rate": 9.880842967395048e-05, "loss": -0.1438, "step": 6970 }, { "epoch": 19.38888888888889, "grad_norm": 0.39438918232917786, "learning_rate": 9.880244183855452e-05, "loss": -0.1441, "step": 6980 }, { "epoch": 19.416666666666668, "grad_norm": 0.44482913613319397, "learning_rate": 9.879643917826527e-05, "loss": -0.1254, "step": 6990 }, { "epoch": 19.444444444444443, "grad_norm": 0.47894376516342163, "learning_rate": 9.87904216949062e-05, "loss": -0.1413, "step": 7000 }, { "epoch": 19.47222222222222, "grad_norm": 0.37498119473457336, "learning_rate": 9.878438939030526e-05, "loss": -0.1359, "step": 7010 }, { "epoch": 19.5, "grad_norm": 0.4097660779953003, "learning_rate": 9.877834226629489e-05, "loss": -0.1428, "step": 7020 }, { "epoch": 19.52777777777778, "grad_norm": 0.47634053230285645, "learning_rate": 9.877228032471206e-05, "loss": -0.1409, "step": 7030 }, { "epoch": 19.555555555555557, "grad_norm": 0.3068281412124634, "learning_rate": 9.876620356739823e-05, "loss": -0.1449, "step": 7040 }, { "epoch": 19.583333333333332, "grad_norm": 0.39558884501457214, "learning_rate": 9.876011199619935e-05, "loss": -0.1431, "step": 7050 }, { "epoch": 19.61111111111111, "grad_norm": 0.3595668375492096, "learning_rate": 9.875400561296589e-05, "loss": -0.1424, "step": 7060 }, { "epoch": 19.63888888888889, "grad_norm": 0.3943558931350708, "learning_rate": 9.874788441955278e-05, "loss": -0.135, "step": 7070 }, { "epoch": 19.666666666666668, "grad_norm": 0.39310476183891296, "learning_rate": 9.874174841781951e-05, "loss": -0.1364, "step": 7080 }, { "epoch": 19.694444444444443, "grad_norm": 0.37584370374679565, "learning_rate": 9.873559760963003e-05, "loss": -0.1418, "step": 7090 }, { "epoch": 19.72222222222222, "grad_norm": 0.3255833685398102, "learning_rate": 9.872943199685278e-05, "loss": -0.1444, "step": 7100 }, { "epoch": 19.75, "grad_norm": 0.3243601620197296, "learning_rate": 9.872325158136071e-05, "loss": -0.1437, "step": 7110 }, { "epoch": 19.77777777777778, "grad_norm": 0.6516162753105164, "learning_rate": 9.871705636503128e-05, "loss": -0.1478, "step": 7120 }, { "epoch": 19.805555555555557, "grad_norm": 0.37094247341156006, "learning_rate": 9.871084634974641e-05, "loss": -0.1425, "step": 7130 }, { "epoch": 19.833333333333332, "grad_norm": 0.49075552821159363, "learning_rate": 9.870462153739257e-05, "loss": -0.1393, "step": 7140 }, { "epoch": 19.86111111111111, "grad_norm": 0.412914514541626, "learning_rate": 9.869838192986067e-05, "loss": -0.1432, "step": 7150 }, { "epoch": 19.88888888888889, "grad_norm": 0.39772829413414, "learning_rate": 9.869212752904616e-05, "loss": -0.141, "step": 7160 }, { "epoch": 19.916666666666668, "grad_norm": 0.3147515058517456, "learning_rate": 9.868585833684894e-05, "loss": -0.1454, "step": 7170 }, { "epoch": 19.944444444444443, "grad_norm": 0.3377986550331116, "learning_rate": 9.867957435517342e-05, "loss": -0.1458, "step": 7180 }, { "epoch": 19.97222222222222, "grad_norm": 0.2569406032562256, "learning_rate": 9.867327558592854e-05, "loss": -0.1433, "step": 7190 }, { "epoch": 20.0, "grad_norm": 0.32388171553611755, "learning_rate": 9.866696203102766e-05, "loss": -0.1366, "step": 7200 }, { "epoch": 20.02777777777778, "grad_norm": 0.3306736648082733, "learning_rate": 9.86606336923887e-05, "loss": -0.1455, "step": 7210 }, { "epoch": 20.055555555555557, "grad_norm": 0.29443883895874023, "learning_rate": 9.865429057193403e-05, "loss": -0.1455, "step": 7220 }, { "epoch": 20.083333333333332, "grad_norm": 0.4117501974105835, "learning_rate": 9.864793267159053e-05, "loss": -0.1443, "step": 7230 }, { "epoch": 20.11111111111111, "grad_norm": 0.4651457965373993, "learning_rate": 9.864155999328957e-05, "loss": -0.1431, "step": 7240 }, { "epoch": 20.13888888888889, "grad_norm": 0.383597195148468, "learning_rate": 9.8635172538967e-05, "loss": -0.1371, "step": 7250 }, { "epoch": 20.166666666666668, "grad_norm": 0.2768787443637848, "learning_rate": 9.862877031056312e-05, "loss": -0.1427, "step": 7260 }, { "epoch": 20.194444444444443, "grad_norm": 0.39687851071357727, "learning_rate": 9.862235331002279e-05, "loss": -0.1333, "step": 7270 }, { "epoch": 20.22222222222222, "grad_norm": 0.7541289925575256, "learning_rate": 9.861592153929533e-05, "loss": -0.1398, "step": 7280 }, { "epoch": 20.25, "grad_norm": 0.5834373235702515, "learning_rate": 9.860947500033455e-05, "loss": -0.1418, "step": 7290 }, { "epoch": 20.27777777777778, "grad_norm": 0.44282224774360657, "learning_rate": 9.86030136950987e-05, "loss": -0.1342, "step": 7300 }, { "epoch": 20.305555555555557, "grad_norm": 0.36041495203971863, "learning_rate": 9.85965376255506e-05, "loss": -0.148, "step": 7310 }, { "epoch": 20.333333333333332, "grad_norm": 0.4277142286300659, "learning_rate": 9.859004679365747e-05, "loss": -0.1467, "step": 7320 }, { "epoch": 20.36111111111111, "grad_norm": 0.32487428188323975, "learning_rate": 9.858354120139108e-05, "loss": -0.1451, "step": 7330 }, { "epoch": 20.38888888888889, "grad_norm": 0.373870313167572, "learning_rate": 9.857702085072764e-05, "loss": -0.1449, "step": 7340 }, { "epoch": 20.416666666666668, "grad_norm": 0.48206478357315063, "learning_rate": 9.857048574364787e-05, "loss": -0.1422, "step": 7350 }, { "epoch": 20.444444444444443, "grad_norm": 0.36080238223075867, "learning_rate": 9.856393588213698e-05, "loss": -0.1352, "step": 7360 }, { "epoch": 20.47222222222222, "grad_norm": 0.40804144740104675, "learning_rate": 9.855737126818458e-05, "loss": -0.1401, "step": 7370 }, { "epoch": 20.5, "grad_norm": 0.4099501371383667, "learning_rate": 9.855079190378491e-05, "loss": -0.1401, "step": 7380 }, { "epoch": 20.52777777777778, "grad_norm": 0.2999636232852936, "learning_rate": 9.854419779093655e-05, "loss": -0.1431, "step": 7390 }, { "epoch": 20.555555555555557, "grad_norm": 0.44685104489326477, "learning_rate": 9.853758893164264e-05, "loss": -0.1466, "step": 7400 }, { "epoch": 20.583333333333332, "grad_norm": 0.34942975640296936, "learning_rate": 9.853096532791078e-05, "loss": -0.1483, "step": 7410 }, { "epoch": 20.61111111111111, "grad_norm": 0.31938233971595764, "learning_rate": 9.852432698175304e-05, "loss": -0.1328, "step": 7420 }, { "epoch": 20.63888888888889, "grad_norm": 0.5349838137626648, "learning_rate": 9.851767389518597e-05, "loss": -0.1358, "step": 7430 }, { "epoch": 20.666666666666668, "grad_norm": 0.3530218303203583, "learning_rate": 9.85110060702306e-05, "loss": -0.1482, "step": 7440 }, { "epoch": 20.694444444444443, "grad_norm": 0.41040822863578796, "learning_rate": 9.850432350891245e-05, "loss": -0.1478, "step": 7450 }, { "epoch": 20.72222222222222, "grad_norm": 0.40131616592407227, "learning_rate": 9.84976262132615e-05, "loss": -0.1469, "step": 7460 }, { "epoch": 20.75, "grad_norm": 0.39994707703590393, "learning_rate": 9.849091418531222e-05, "loss": -0.1416, "step": 7470 }, { "epoch": 20.77777777777778, "grad_norm": 0.5162163972854614, "learning_rate": 9.848418742710353e-05, "loss": -0.1378, "step": 7480 }, { "epoch": 20.805555555555557, "grad_norm": 0.31766968965530396, "learning_rate": 9.847744594067885e-05, "loss": -0.1447, "step": 7490 }, { "epoch": 20.833333333333332, "grad_norm": 0.3730025589466095, "learning_rate": 9.847068972808607e-05, "loss": -0.14, "step": 7500 }, { "epoch": 20.86111111111111, "grad_norm": 0.6289884448051453, "learning_rate": 9.846391879137756e-05, "loss": -0.1473, "step": 7510 }, { "epoch": 20.88888888888889, "grad_norm": 0.34475818276405334, "learning_rate": 9.845713313261012e-05, "loss": -0.1413, "step": 7520 }, { "epoch": 20.916666666666668, "grad_norm": 0.35013824701309204, "learning_rate": 9.845033275384505e-05, "loss": -0.1328, "step": 7530 }, { "epoch": 20.944444444444443, "grad_norm": 0.3263227939605713, "learning_rate": 9.844351765714818e-05, "loss": -0.1452, "step": 7540 }, { "epoch": 20.97222222222222, "grad_norm": 0.42239218950271606, "learning_rate": 9.843668784458971e-05, "loss": -0.1429, "step": 7550 }, { "epoch": 21.0, "grad_norm": 0.2860781252384186, "learning_rate": 9.842984331824437e-05, "loss": -0.149, "step": 7560 }, { "epoch": 21.02777777777778, "grad_norm": 0.33498725295066833, "learning_rate": 9.842298408019133e-05, "loss": -0.1572, "step": 7570 }, { "epoch": 21.055555555555557, "grad_norm": 0.4542574882507324, "learning_rate": 9.841611013251429e-05, "loss": -0.1444, "step": 7580 }, { "epoch": 21.083333333333332, "grad_norm": 0.37516412138938904, "learning_rate": 9.840922147730133e-05, "loss": -0.147, "step": 7590 }, { "epoch": 21.11111111111111, "grad_norm": 0.6852602362632751, "learning_rate": 9.840231811664506e-05, "loss": -0.1438, "step": 7600 }, { "epoch": 21.13888888888889, "grad_norm": 0.5078694820404053, "learning_rate": 9.839540005264252e-05, "loss": -0.1426, "step": 7610 }, { "epoch": 21.166666666666668, "grad_norm": 0.42627257108688354, "learning_rate": 9.838846728739527e-05, "loss": -0.1507, "step": 7620 }, { "epoch": 21.194444444444443, "grad_norm": 0.3333900272846222, "learning_rate": 9.838151982300927e-05, "loss": -0.1344, "step": 7630 }, { "epoch": 21.22222222222222, "grad_norm": 0.30451977252960205, "learning_rate": 9.8374557661595e-05, "loss": -0.1457, "step": 7640 }, { "epoch": 21.25, "grad_norm": 0.33878225088119507, "learning_rate": 9.836758080526735e-05, "loss": -0.1394, "step": 7650 }, { "epoch": 21.27777777777778, "grad_norm": 0.4311405420303345, "learning_rate": 9.836058925614575e-05, "loss": -0.1469, "step": 7660 }, { "epoch": 21.305555555555557, "grad_norm": 0.2539895176887512, "learning_rate": 9.8353583016354e-05, "loss": -0.1493, "step": 7670 }, { "epoch": 21.333333333333332, "grad_norm": 0.2929133474826813, "learning_rate": 9.834656208802044e-05, "loss": -0.1478, "step": 7680 }, { "epoch": 21.36111111111111, "grad_norm": 0.2731199562549591, "learning_rate": 9.833952647327784e-05, "loss": -0.1462, "step": 7690 }, { "epoch": 21.38888888888889, "grad_norm": 0.3968902826309204, "learning_rate": 9.833247617426342e-05, "loss": -0.1483, "step": 7700 }, { "epoch": 21.416666666666668, "grad_norm": 0.39756155014038086, "learning_rate": 9.832541119311889e-05, "loss": -0.1471, "step": 7710 }, { "epoch": 21.444444444444443, "grad_norm": 0.389934778213501, "learning_rate": 9.83183315319904e-05, "loss": -0.136, "step": 7720 }, { "epoch": 21.47222222222222, "grad_norm": 0.3449733257293701, "learning_rate": 9.831123719302855e-05, "loss": -0.1425, "step": 7730 }, { "epoch": 21.5, "grad_norm": 0.4005852937698364, "learning_rate": 9.830412817838842e-05, "loss": -0.1464, "step": 7740 }, { "epoch": 21.52777777777778, "grad_norm": 0.364165335893631, "learning_rate": 9.829700449022956e-05, "loss": -0.1406, "step": 7750 }, { "epoch": 21.555555555555557, "grad_norm": 0.37194177508354187, "learning_rate": 9.828986613071593e-05, "loss": -0.1499, "step": 7760 }, { "epoch": 21.583333333333332, "grad_norm": 0.41089484095573425, "learning_rate": 9.828271310201601e-05, "loss": -0.149, "step": 7770 }, { "epoch": 21.61111111111111, "grad_norm": 0.417978972196579, "learning_rate": 9.827554540630268e-05, "loss": -0.1455, "step": 7780 }, { "epoch": 21.63888888888889, "grad_norm": 0.39050453901290894, "learning_rate": 9.826836304575329e-05, "loss": -0.1359, "step": 7790 }, { "epoch": 21.666666666666668, "grad_norm": 0.28262096643447876, "learning_rate": 9.826116602254966e-05, "loss": -0.153, "step": 7800 }, { "epoch": 21.694444444444443, "grad_norm": 0.44464802742004395, "learning_rate": 9.825395433887805e-05, "loss": -0.1444, "step": 7810 }, { "epoch": 21.72222222222222, "grad_norm": 0.3527679443359375, "learning_rate": 9.824672799692917e-05, "loss": -0.1479, "step": 7820 }, { "epoch": 21.75, "grad_norm": 0.6067623496055603, "learning_rate": 9.823948699889823e-05, "loss": -0.1437, "step": 7830 }, { "epoch": 21.77777777777778, "grad_norm": 0.3939252495765686, "learning_rate": 9.823223134698483e-05, "loss": -0.149, "step": 7840 }, { "epoch": 21.805555555555557, "grad_norm": 0.5206489562988281, "learning_rate": 9.822496104339303e-05, "loss": -0.1441, "step": 7850 }, { "epoch": 21.833333333333332, "grad_norm": 0.49517151713371277, "learning_rate": 9.821767609033138e-05, "loss": -0.1438, "step": 7860 }, { "epoch": 21.86111111111111, "grad_norm": 0.45759856700897217, "learning_rate": 9.821037649001284e-05, "loss": -0.1399, "step": 7870 }, { "epoch": 21.88888888888889, "grad_norm": 0.4043305814266205, "learning_rate": 9.820306224465486e-05, "loss": -0.1386, "step": 7880 }, { "epoch": 21.916666666666668, "grad_norm": 0.48860520124435425, "learning_rate": 9.819573335647928e-05, "loss": -0.1459, "step": 7890 }, { "epoch": 21.944444444444443, "grad_norm": 0.4669260084629059, "learning_rate": 9.818838982771246e-05, "loss": -0.1347, "step": 7900 }, { "epoch": 21.97222222222222, "grad_norm": 0.3433322608470917, "learning_rate": 9.818103166058514e-05, "loss": -0.1473, "step": 7910 }, { "epoch": 22.0, "grad_norm": 0.32116326689720154, "learning_rate": 9.817365885733254e-05, "loss": -0.1455, "step": 7920 }, { "epoch": 22.02777777777778, "grad_norm": 0.352649986743927, "learning_rate": 9.816627142019434e-05, "loss": -0.1451, "step": 7930 }, { "epoch": 22.055555555555557, "grad_norm": 0.369086354970932, "learning_rate": 9.815886935141463e-05, "loss": -0.139, "step": 7940 }, { "epoch": 22.083333333333332, "grad_norm": 0.316665917634964, "learning_rate": 9.8151452653242e-05, "loss": -0.1527, "step": 7950 }, { "epoch": 22.11111111111111, "grad_norm": 0.5910707116127014, "learning_rate": 9.814402132792939e-05, "loss": -0.1457, "step": 7960 }, { "epoch": 22.13888888888889, "grad_norm": 0.3818620443344116, "learning_rate": 9.813657537773428e-05, "loss": -0.1432, "step": 7970 }, { "epoch": 22.166666666666668, "grad_norm": 0.37091705203056335, "learning_rate": 9.812911480491854e-05, "loss": -0.1513, "step": 7980 }, { "epoch": 22.194444444444443, "grad_norm": 0.35713258385658264, "learning_rate": 9.81216396117485e-05, "loss": -0.1474, "step": 7990 }, { "epoch": 22.22222222222222, "grad_norm": 0.34822532534599304, "learning_rate": 9.811414980049491e-05, "loss": -0.1467, "step": 8000 }, { "epoch": 22.25, "grad_norm": 0.40326449275016785, "learning_rate": 9.810664537343301e-05, "loss": -0.1501, "step": 8010 }, { "epoch": 22.27777777777778, "grad_norm": 0.4854445159435272, "learning_rate": 9.809912633284243e-05, "loss": -0.148, "step": 8020 }, { "epoch": 22.305555555555557, "grad_norm": 0.36553147435188293, "learning_rate": 9.809159268100725e-05, "loss": -0.1477, "step": 8030 }, { "epoch": 22.333333333333332, "grad_norm": 0.33933037519454956, "learning_rate": 9.808404442021599e-05, "loss": -0.1493, "step": 8040 }, { "epoch": 22.36111111111111, "grad_norm": 0.42129963636398315, "learning_rate": 9.807648155276163e-05, "loss": -0.1483, "step": 8050 }, { "epoch": 22.38888888888889, "grad_norm": 0.48523926734924316, "learning_rate": 9.806890408094156e-05, "loss": -0.1436, "step": 8060 }, { "epoch": 22.416666666666668, "grad_norm": 0.3713196814060211, "learning_rate": 9.806131200705761e-05, "loss": -0.1472, "step": 8070 }, { "epoch": 22.444444444444443, "grad_norm": 0.33642578125, "learning_rate": 9.805370533341605e-05, "loss": -0.1414, "step": 8080 }, { "epoch": 22.47222222222222, "grad_norm": 0.29362815618515015, "learning_rate": 9.804608406232762e-05, "loss": -0.1518, "step": 8090 }, { "epoch": 22.5, "grad_norm": 0.4087931513786316, "learning_rate": 9.803844819610741e-05, "loss": -0.1474, "step": 8100 }, { "epoch": 22.52777777777778, "grad_norm": 0.30327022075653076, "learning_rate": 9.803079773707504e-05, "loss": -0.1422, "step": 8110 }, { "epoch": 22.555555555555557, "grad_norm": 0.4739992022514343, "learning_rate": 9.802313268755447e-05, "loss": -0.1363, "step": 8120 }, { "epoch": 22.583333333333332, "grad_norm": 0.5584814548492432, "learning_rate": 9.801545304987419e-05, "loss": -0.1438, "step": 8130 }, { "epoch": 22.61111111111111, "grad_norm": 0.4012373387813568, "learning_rate": 9.800775882636704e-05, "loss": -0.1452, "step": 8140 }, { "epoch": 22.63888888888889, "grad_norm": 0.5267181992530823, "learning_rate": 9.800005001937034e-05, "loss": -0.1475, "step": 8150 }, { "epoch": 22.666666666666668, "grad_norm": 0.3163875341415405, "learning_rate": 9.79923266312258e-05, "loss": -0.1478, "step": 8160 }, { "epoch": 22.694444444444443, "grad_norm": 0.49427518248558044, "learning_rate": 9.79845886642796e-05, "loss": -0.1429, "step": 8170 }, { "epoch": 22.72222222222222, "grad_norm": 0.4403149485588074, "learning_rate": 9.797683612088233e-05, "loss": -0.1507, "step": 8180 }, { "epoch": 22.75, "grad_norm": 0.28307992219924927, "learning_rate": 9.796906900338898e-05, "loss": -0.1427, "step": 8190 }, { "epoch": 22.77777777777778, "grad_norm": 0.5075427293777466, "learning_rate": 9.796128731415903e-05, "loss": -0.1405, "step": 8200 }, { "epoch": 22.805555555555557, "grad_norm": 0.3217777907848358, "learning_rate": 9.795349105555634e-05, "loss": -0.1466, "step": 8210 }, { "epoch": 22.833333333333332, "grad_norm": 0.28487855195999146, "learning_rate": 9.794568022994922e-05, "loss": -0.1458, "step": 8220 }, { "epoch": 22.86111111111111, "grad_norm": 0.5153539776802063, "learning_rate": 9.793785483971034e-05, "loss": -0.1475, "step": 8230 }, { "epoch": 22.88888888888889, "grad_norm": 0.35787081718444824, "learning_rate": 9.793001488721691e-05, "loss": -0.146, "step": 8240 }, { "epoch": 22.916666666666668, "grad_norm": 0.5184857845306396, "learning_rate": 9.792216037485047e-05, "loss": -0.1462, "step": 8250 }, { "epoch": 22.944444444444443, "grad_norm": 0.24898751080036163, "learning_rate": 9.791429130499704e-05, "loss": -0.1488, "step": 8260 }, { "epoch": 22.97222222222222, "grad_norm": 0.38097915053367615, "learning_rate": 9.790640768004698e-05, "loss": -0.1454, "step": 8270 }, { "epoch": 23.0, "grad_norm": 0.3609219789505005, "learning_rate": 9.789850950239518e-05, "loss": -0.1389, "step": 8280 }, { "epoch": 23.02777777777778, "grad_norm": 0.3721885681152344, "learning_rate": 9.789059677444089e-05, "loss": -0.1434, "step": 8290 }, { "epoch": 23.055555555555557, "grad_norm": 0.32281678915023804, "learning_rate": 9.788266949858776e-05, "loss": -0.1566, "step": 8300 }, { "epoch": 23.083333333333332, "grad_norm": 0.44867488741874695, "learning_rate": 9.787472767724392e-05, "loss": -0.1603, "step": 8310 }, { "epoch": 23.11111111111111, "grad_norm": 0.4402677118778229, "learning_rate": 9.786677131282185e-05, "loss": -0.1526, "step": 8320 }, { "epoch": 23.13888888888889, "grad_norm": 0.39589422941207886, "learning_rate": 9.785880040773853e-05, "loss": -0.1475, "step": 8330 }, { "epoch": 23.166666666666668, "grad_norm": 0.624754786491394, "learning_rate": 9.785081496441527e-05, "loss": -0.1411, "step": 8340 }, { "epoch": 23.194444444444443, "grad_norm": 0.3760470151901245, "learning_rate": 9.784281498527785e-05, "loss": -0.1479, "step": 8350 }, { "epoch": 23.22222222222222, "grad_norm": 0.5037923455238342, "learning_rate": 9.783480047275646e-05, "loss": -0.1406, "step": 8360 }, { "epoch": 23.25, "grad_norm": 0.3075854182243347, "learning_rate": 9.78267714292857e-05, "loss": -0.1475, "step": 8370 }, { "epoch": 23.27777777777778, "grad_norm": 0.3455106317996979, "learning_rate": 9.781872785730454e-05, "loss": -0.1471, "step": 8380 }, { "epoch": 23.305555555555557, "grad_norm": 0.45987266302108765, "learning_rate": 9.781066975925646e-05, "loss": -0.1486, "step": 8390 }, { "epoch": 23.333333333333332, "grad_norm": 0.3775578439235687, "learning_rate": 9.780259713758928e-05, "loss": -0.1534, "step": 8400 }, { "epoch": 23.36111111111111, "grad_norm": 0.4246465861797333, "learning_rate": 9.779450999475524e-05, "loss": -0.1545, "step": 8410 }, { "epoch": 23.38888888888889, "grad_norm": 0.41401514410972595, "learning_rate": 9.7786408333211e-05, "loss": -0.155, "step": 8420 }, { "epoch": 23.416666666666668, "grad_norm": 0.3321317732334137, "learning_rate": 9.777829215541764e-05, "loss": -0.1482, "step": 8430 }, { "epoch": 23.444444444444443, "grad_norm": 0.3710100054740906, "learning_rate": 9.777016146384064e-05, "loss": -0.1541, "step": 8440 }, { "epoch": 23.47222222222222, "grad_norm": 0.576160728931427, "learning_rate": 9.776201626094988e-05, "loss": -0.1491, "step": 8450 }, { "epoch": 23.5, "grad_norm": 0.49136796593666077, "learning_rate": 9.775385654921965e-05, "loss": -0.151, "step": 8460 }, { "epoch": 23.52777777777778, "grad_norm": 0.29044508934020996, "learning_rate": 9.774568233112868e-05, "loss": -0.1462, "step": 8470 }, { "epoch": 23.555555555555557, "grad_norm": 0.3756619095802307, "learning_rate": 9.773749360916007e-05, "loss": -0.146, "step": 8480 }, { "epoch": 23.583333333333332, "grad_norm": 0.30979976058006287, "learning_rate": 9.772929038580134e-05, "loss": -0.1491, "step": 8490 }, { "epoch": 23.61111111111111, "grad_norm": 0.5724631547927856, "learning_rate": 9.772107266354439e-05, "loss": -0.1456, "step": 8500 }, { "epoch": 23.63888888888889, "grad_norm": 0.3852592706680298, "learning_rate": 9.77128404448856e-05, "loss": -0.1358, "step": 8510 }, { "epoch": 23.666666666666668, "grad_norm": 0.42944639921188354, "learning_rate": 9.770459373232565e-05, "loss": -0.1565, "step": 8520 }, { "epoch": 23.694444444444443, "grad_norm": 0.464107483625412, "learning_rate": 9.769633252836969e-05, "loss": -0.1428, "step": 8530 }, { "epoch": 23.72222222222222, "grad_norm": 0.5003796815872192, "learning_rate": 9.768805683552724e-05, "loss": -0.1533, "step": 8540 }, { "epoch": 23.75, "grad_norm": 0.36414337158203125, "learning_rate": 9.767976665631228e-05, "loss": -0.1506, "step": 8550 }, { "epoch": 23.77777777777778, "grad_norm": 0.40157464146614075, "learning_rate": 9.767146199324311e-05, "loss": -0.1497, "step": 8560 }, { "epoch": 23.805555555555557, "grad_norm": 0.4052000343799591, "learning_rate": 9.766314284884249e-05, "loss": -0.1572, "step": 8570 }, { "epoch": 23.833333333333332, "grad_norm": 0.3136754035949707, "learning_rate": 9.765480922563752e-05, "loss": -0.1496, "step": 8580 }, { "epoch": 23.86111111111111, "grad_norm": 0.40961602330207825, "learning_rate": 9.764646112615978e-05, "loss": -0.1507, "step": 8590 }, { "epoch": 23.88888888888889, "grad_norm": 0.3105032742023468, "learning_rate": 9.763809855294517e-05, "loss": -0.1536, "step": 8600 }, { "epoch": 23.916666666666668, "grad_norm": 0.3791199028491974, "learning_rate": 9.762972150853404e-05, "loss": -0.1536, "step": 8610 }, { "epoch": 23.944444444444443, "grad_norm": 0.42362353205680847, "learning_rate": 9.762132999547111e-05, "loss": -0.1535, "step": 8620 }, { "epoch": 23.97222222222222, "grad_norm": 0.5086018443107605, "learning_rate": 9.761292401630549e-05, "loss": -0.1428, "step": 8630 }, { "epoch": 24.0, "grad_norm": 0.2711368501186371, "learning_rate": 9.76045035735907e-05, "loss": -0.1553, "step": 8640 }, { "epoch": 24.02777777777778, "grad_norm": 0.24859407544136047, "learning_rate": 9.759606866988464e-05, "loss": -0.1499, "step": 8650 }, { "epoch": 24.055555555555557, "grad_norm": 0.2815381586551666, "learning_rate": 9.758761930774963e-05, "loss": -0.1569, "step": 8660 }, { "epoch": 24.083333333333332, "grad_norm": 0.24295075237751007, "learning_rate": 9.757915548975235e-05, "loss": -0.1574, "step": 8670 }, { "epoch": 24.11111111111111, "grad_norm": 0.4359634220600128, "learning_rate": 9.757067721846389e-05, "loss": -0.1532, "step": 8680 }, { "epoch": 24.13888888888889, "grad_norm": 0.33819594979286194, "learning_rate": 9.756218449645971e-05, "loss": -0.1522, "step": 8690 }, { "epoch": 24.166666666666668, "grad_norm": 0.4104031026363373, "learning_rate": 9.75536773263197e-05, "loss": -0.1508, "step": 8700 }, { "epoch": 24.194444444444443, "grad_norm": 0.24778637290000916, "learning_rate": 9.75451557106281e-05, "loss": -0.1552, "step": 8710 }, { "epoch": 24.22222222222222, "grad_norm": 0.4328976571559906, "learning_rate": 9.753661965197354e-05, "loss": -0.1522, "step": 8720 }, { "epoch": 24.25, "grad_norm": 0.3909602761268616, "learning_rate": 9.752806915294908e-05, "loss": -0.1529, "step": 8730 }, { "epoch": 24.27777777777778, "grad_norm": 0.3495517671108246, "learning_rate": 9.75195042161521e-05, "loss": -0.1538, "step": 8740 }, { "epoch": 24.305555555555557, "grad_norm": 0.46148669719696045, "learning_rate": 9.751092484418442e-05, "loss": -0.1576, "step": 8750 }, { "epoch": 24.333333333333332, "grad_norm": 0.4304441213607788, "learning_rate": 9.750233103965224e-05, "loss": -0.1488, "step": 8760 }, { "epoch": 24.36111111111111, "grad_norm": 0.4332529604434967, "learning_rate": 9.749372280516611e-05, "loss": -0.147, "step": 8770 }, { "epoch": 24.38888888888889, "grad_norm": 0.4765974283218384, "learning_rate": 9.748510014334097e-05, "loss": -0.1418, "step": 8780 }, { "epoch": 24.416666666666668, "grad_norm": 0.3767548203468323, "learning_rate": 9.747646305679621e-05, "loss": -0.1442, "step": 8790 }, { "epoch": 24.444444444444443, "grad_norm": 0.3512078523635864, "learning_rate": 9.74678115481555e-05, "loss": -0.1345, "step": 8800 }, { "epoch": 24.47222222222222, "grad_norm": 0.49064767360687256, "learning_rate": 9.745914562004696e-05, "loss": -0.1443, "step": 8810 }, { "epoch": 24.5, "grad_norm": 0.41971686482429504, "learning_rate": 9.745046527510307e-05, "loss": -0.1498, "step": 8820 }, { "epoch": 24.52777777777778, "grad_norm": 0.3214225769042969, "learning_rate": 9.744177051596068e-05, "loss": -0.1442, "step": 8830 }, { "epoch": 24.555555555555557, "grad_norm": 0.4597470462322235, "learning_rate": 9.743306134526105e-05, "loss": -0.1446, "step": 8840 }, { "epoch": 24.583333333333332, "grad_norm": 0.4510306119918823, "learning_rate": 9.742433776564977e-05, "loss": -0.1491, "step": 8850 }, { "epoch": 24.61111111111111, "grad_norm": 0.3733977973461151, "learning_rate": 9.741559977977683e-05, "loss": -0.1467, "step": 8860 }, { "epoch": 24.63888888888889, "grad_norm": 0.5124517679214478, "learning_rate": 9.740684739029661e-05, "loss": -0.1495, "step": 8870 }, { "epoch": 24.666666666666668, "grad_norm": 0.5279785990715027, "learning_rate": 9.739808059986789e-05, "loss": -0.1423, "step": 8880 }, { "epoch": 24.694444444444443, "grad_norm": 0.4995977580547333, "learning_rate": 9.738929941115373e-05, "loss": -0.1446, "step": 8890 }, { "epoch": 24.72222222222222, "grad_norm": 0.3921814560890198, "learning_rate": 9.738050382682167e-05, "loss": -0.1569, "step": 8900 }, { "epoch": 24.75, "grad_norm": 0.42043089866638184, "learning_rate": 9.737169384954355e-05, "loss": -0.1457, "step": 8910 }, { "epoch": 24.77777777777778, "grad_norm": 0.3276885747909546, "learning_rate": 9.736286948199562e-05, "loss": -0.1507, "step": 8920 }, { "epoch": 24.805555555555557, "grad_norm": 0.3894919455051422, "learning_rate": 9.735403072685848e-05, "loss": -0.1518, "step": 8930 }, { "epoch": 24.833333333333332, "grad_norm": 0.5697619915008545, "learning_rate": 9.734517758681712e-05, "loss": -0.1514, "step": 8940 }, { "epoch": 24.86111111111111, "grad_norm": 0.44825130701065063, "learning_rate": 9.733631006456088e-05, "loss": -0.1526, "step": 8950 }, { "epoch": 24.88888888888889, "grad_norm": 0.4179113805294037, "learning_rate": 9.732742816278348e-05, "loss": -0.1517, "step": 8960 }, { "epoch": 24.916666666666668, "grad_norm": 0.5169837474822998, "learning_rate": 9.731853188418302e-05, "loss": -0.1524, "step": 8970 }, { "epoch": 24.944444444444443, "grad_norm": 0.35534608364105225, "learning_rate": 9.730962123146194e-05, "loss": -0.145, "step": 8980 }, { "epoch": 24.97222222222222, "grad_norm": 0.43961742520332336, "learning_rate": 9.730069620732709e-05, "loss": -0.1512, "step": 8990 }, { "epoch": 25.0, "grad_norm": 0.32562151551246643, "learning_rate": 9.72917568144896e-05, "loss": -0.153, "step": 9000 }, { "epoch": 25.02777777777778, "grad_norm": 0.2835281789302826, "learning_rate": 9.728280305566509e-05, "loss": -0.1422, "step": 9010 }, { "epoch": 25.055555555555557, "grad_norm": 0.3372002840042114, "learning_rate": 9.727383493357343e-05, "loss": -0.1443, "step": 9020 }, { "epoch": 25.083333333333332, "grad_norm": 0.44782814383506775, "learning_rate": 9.726485245093891e-05, "loss": -0.1476, "step": 9030 }, { "epoch": 25.11111111111111, "grad_norm": 0.5292194485664368, "learning_rate": 9.725585561049018e-05, "loss": -0.1516, "step": 9040 }, { "epoch": 25.13888888888889, "grad_norm": 0.4162791967391968, "learning_rate": 9.724684441496022e-05, "loss": -0.1432, "step": 9050 }, { "epoch": 25.166666666666668, "grad_norm": 0.3842003345489502, "learning_rate": 9.72378188670864e-05, "loss": -0.1513, "step": 9060 }, { "epoch": 25.194444444444443, "grad_norm": 0.272505521774292, "learning_rate": 9.722877896961047e-05, "loss": -0.1544, "step": 9070 }, { "epoch": 25.22222222222222, "grad_norm": 0.3557271957397461, "learning_rate": 9.721972472527848e-05, "loss": -0.1521, "step": 9080 }, { "epoch": 25.25, "grad_norm": 0.2987447679042816, "learning_rate": 9.721065613684089e-05, "loss": -0.1559, "step": 9090 }, { "epoch": 25.27777777777778, "grad_norm": 0.28910037875175476, "learning_rate": 9.72015732070525e-05, "loss": -0.1544, "step": 9100 }, { "epoch": 25.305555555555557, "grad_norm": 0.33208173513412476, "learning_rate": 9.719247593867244e-05, "loss": -0.1589, "step": 9110 }, { "epoch": 25.333333333333332, "grad_norm": 0.37975963950157166, "learning_rate": 9.718336433446423e-05, "loss": -0.1551, "step": 9120 }, { "epoch": 25.36111111111111, "grad_norm": 0.3486824035644531, "learning_rate": 9.717423839719574e-05, "loss": -0.1511, "step": 9130 }, { "epoch": 25.38888888888889, "grad_norm": 0.6060886383056641, "learning_rate": 9.71650981296392e-05, "loss": -0.1487, "step": 9140 }, { "epoch": 25.416666666666668, "grad_norm": 0.3957447409629822, "learning_rate": 9.715594353457118e-05, "loss": -0.152, "step": 9150 }, { "epoch": 25.444444444444443, "grad_norm": 0.4619891345500946, "learning_rate": 9.714677461477257e-05, "loss": -0.1471, "step": 9160 }, { "epoch": 25.47222222222222, "grad_norm": 0.3696295917034149, "learning_rate": 9.713759137302869e-05, "loss": -0.147, "step": 9170 }, { "epoch": 25.5, "grad_norm": 0.2760431170463562, "learning_rate": 9.712839381212914e-05, "loss": -0.1498, "step": 9180 }, { "epoch": 25.52777777777778, "grad_norm": 0.3418677747249603, "learning_rate": 9.71191819348679e-05, "loss": -0.1488, "step": 9190 }, { "epoch": 25.555555555555557, "grad_norm": 0.34300872683525085, "learning_rate": 9.710995574404331e-05, "loss": -0.1501, "step": 9200 }, { "epoch": 25.583333333333332, "grad_norm": 0.2504650354385376, "learning_rate": 9.710071524245802e-05, "loss": -0.1452, "step": 9210 }, { "epoch": 25.61111111111111, "grad_norm": 0.3717433214187622, "learning_rate": 9.709146043291906e-05, "loss": -0.1576, "step": 9220 }, { "epoch": 25.63888888888889, "grad_norm": 0.30197593569755554, "learning_rate": 9.70821913182378e-05, "loss": -0.1499, "step": 9230 }, { "epoch": 25.666666666666668, "grad_norm": 0.3656774163246155, "learning_rate": 9.707290790122995e-05, "loss": -0.1519, "step": 9240 }, { "epoch": 25.694444444444443, "grad_norm": 0.20422972738742828, "learning_rate": 9.706361018471557e-05, "loss": -0.1511, "step": 9250 }, { "epoch": 25.72222222222222, "grad_norm": 0.3493691086769104, "learning_rate": 9.705429817151906e-05, "loss": -0.1534, "step": 9260 }, { "epoch": 25.75, "grad_norm": 0.3532660901546478, "learning_rate": 9.704497186446917e-05, "loss": -0.1557, "step": 9270 }, { "epoch": 25.77777777777778, "grad_norm": 0.2400609850883484, "learning_rate": 9.703563126639896e-05, "loss": -0.1479, "step": 9280 }, { "epoch": 25.805555555555557, "grad_norm": 0.540220320224762, "learning_rate": 9.70262763801459e-05, "loss": -0.1555, "step": 9290 }, { "epoch": 25.833333333333332, "grad_norm": 0.2864193618297577, "learning_rate": 9.701690720855171e-05, "loss": -0.1498, "step": 9300 }, { "epoch": 25.86111111111111, "grad_norm": 0.4426887631416321, "learning_rate": 9.700752375446253e-05, "loss": -0.1499, "step": 9310 }, { "epoch": 25.88888888888889, "grad_norm": 0.37749940156936646, "learning_rate": 9.69981260207288e-05, "loss": -0.1359, "step": 9320 }, { "epoch": 25.916666666666668, "grad_norm": 0.4544353187084198, "learning_rate": 9.698871401020529e-05, "loss": -0.1492, "step": 9330 }, { "epoch": 25.944444444444443, "grad_norm": 0.36506399512290955, "learning_rate": 9.697928772575112e-05, "loss": -0.1541, "step": 9340 }, { "epoch": 25.97222222222222, "grad_norm": 0.35455620288848877, "learning_rate": 9.696984717022976e-05, "loss": -0.1604, "step": 9350 }, { "epoch": 26.0, "grad_norm": 0.21675680577754974, "learning_rate": 9.6960392346509e-05, "loss": -0.16, "step": 9360 }, { "epoch": 26.02777777777778, "grad_norm": 0.32673078775405884, "learning_rate": 9.695092325746097e-05, "loss": -0.1406, "step": 9370 }, { "epoch": 26.055555555555557, "grad_norm": 0.2871890962123871, "learning_rate": 9.694143990596211e-05, "loss": -0.1558, "step": 9380 }, { "epoch": 26.083333333333332, "grad_norm": 0.3571845293045044, "learning_rate": 9.693194229489325e-05, "loss": -0.1405, "step": 9390 }, { "epoch": 26.11111111111111, "grad_norm": 0.3138675391674042, "learning_rate": 9.692243042713944e-05, "loss": -0.1567, "step": 9400 }, { "epoch": 26.13888888888889, "grad_norm": 0.3573773205280304, "learning_rate": 9.691290430559022e-05, "loss": -0.1527, "step": 9410 }, { "epoch": 26.166666666666668, "grad_norm": 0.38128015398979187, "learning_rate": 9.690336393313932e-05, "loss": -0.1553, "step": 9420 }, { "epoch": 26.194444444444443, "grad_norm": 0.31965184211730957, "learning_rate": 9.689380931268487e-05, "loss": -0.1491, "step": 9430 }, { "epoch": 26.22222222222222, "grad_norm": 0.23051461577415466, "learning_rate": 9.688424044712932e-05, "loss": -0.1523, "step": 9440 }, { "epoch": 26.25, "grad_norm": 0.3522700369358063, "learning_rate": 9.687465733937942e-05, "loss": -0.145, "step": 9450 }, { "epoch": 26.27777777777778, "grad_norm": 0.4564496874809265, "learning_rate": 9.686505999234627e-05, "loss": -0.1542, "step": 9460 }, { "epoch": 26.305555555555557, "grad_norm": 0.21627460420131683, "learning_rate": 9.685544840894529e-05, "loss": -0.1543, "step": 9470 }, { "epoch": 26.333333333333332, "grad_norm": 0.285600870847702, "learning_rate": 9.684582259209624e-05, "loss": -0.1561, "step": 9480 }, { "epoch": 26.36111111111111, "grad_norm": 0.3584376871585846, "learning_rate": 9.683618254472317e-05, "loss": -0.1514, "step": 9490 }, { "epoch": 26.38888888888889, "grad_norm": 0.46872976422309875, "learning_rate": 9.682652826975449e-05, "loss": -0.1506, "step": 9500 }, { "epoch": 26.416666666666668, "grad_norm": 0.2660931944847107, "learning_rate": 9.681685977012291e-05, "loss": -0.1532, "step": 9510 }, { "epoch": 26.444444444444443, "grad_norm": 0.32704129815101624, "learning_rate": 9.680717704876546e-05, "loss": -0.1511, "step": 9520 }, { "epoch": 26.47222222222222, "grad_norm": 0.38913848996162415, "learning_rate": 9.679748010862349e-05, "loss": -0.1506, "step": 9530 }, { "epoch": 26.5, "grad_norm": 0.38001418113708496, "learning_rate": 9.678776895264267e-05, "loss": -0.154, "step": 9540 }, { "epoch": 26.52777777777778, "grad_norm": 0.26019811630249023, "learning_rate": 9.6778043583773e-05, "loss": -0.1431, "step": 9550 }, { "epoch": 26.555555555555557, "grad_norm": 0.20188280940055847, "learning_rate": 9.67683040049688e-05, "loss": -0.1566, "step": 9560 }, { "epoch": 26.583333333333332, "grad_norm": 0.24092860519886017, "learning_rate": 9.675855021918869e-05, "loss": -0.1519, "step": 9570 }, { "epoch": 26.61111111111111, "grad_norm": 0.3915291726589203, "learning_rate": 9.674878222939561e-05, "loss": -0.1586, "step": 9580 }, { "epoch": 26.63888888888889, "grad_norm": 0.3583378791809082, "learning_rate": 9.673900003855681e-05, "loss": -0.1574, "step": 9590 }, { "epoch": 26.666666666666668, "grad_norm": 0.2861388921737671, "learning_rate": 9.672920364964389e-05, "loss": -0.1473, "step": 9600 }, { "epoch": 26.694444444444443, "grad_norm": 0.5605649948120117, "learning_rate": 9.671939306563269e-05, "loss": -0.1559, "step": 9610 }, { "epoch": 26.72222222222222, "grad_norm": 0.3888830840587616, "learning_rate": 9.670956828950345e-05, "loss": -0.1504, "step": 9620 }, { "epoch": 26.75, "grad_norm": 0.3298644423484802, "learning_rate": 9.669972932424065e-05, "loss": -0.1549, "step": 9630 }, { "epoch": 26.77777777777778, "grad_norm": 0.27522727847099304, "learning_rate": 9.668987617283312e-05, "loss": -0.1571, "step": 9640 }, { "epoch": 26.805555555555557, "grad_norm": 0.39532312750816345, "learning_rate": 9.668000883827397e-05, "loss": -0.1457, "step": 9650 }, { "epoch": 26.833333333333332, "grad_norm": 0.3977895677089691, "learning_rate": 9.667012732356067e-05, "loss": -0.1499, "step": 9660 }, { "epoch": 26.86111111111111, "grad_norm": 0.47975584864616394, "learning_rate": 9.666023163169493e-05, "loss": -0.1323, "step": 9670 }, { "epoch": 26.88888888888889, "grad_norm": 0.3001899719238281, "learning_rate": 9.665032176568281e-05, "loss": -0.1533, "step": 9680 }, { "epoch": 26.916666666666668, "grad_norm": 0.5330898761749268, "learning_rate": 9.664039772853469e-05, "loss": -0.1513, "step": 9690 }, { "epoch": 26.944444444444443, "grad_norm": 0.25037458539009094, "learning_rate": 9.663045952326518e-05, "loss": -0.1508, "step": 9700 }, { "epoch": 26.97222222222222, "grad_norm": 0.35338732600212097, "learning_rate": 9.662050715289328e-05, "loss": -0.1575, "step": 9710 }, { "epoch": 27.0, "grad_norm": 0.39142200350761414, "learning_rate": 9.661054062044226e-05, "loss": -0.1464, "step": 9720 }, { "epoch": 27.02777777777778, "grad_norm": 0.2835354804992676, "learning_rate": 9.660055992893968e-05, "loss": -0.1548, "step": 9730 }, { "epoch": 27.055555555555557, "grad_norm": 0.4294823110103607, "learning_rate": 9.659056508141739e-05, "loss": -0.1526, "step": 9740 }, { "epoch": 27.083333333333332, "grad_norm": 0.3826455771923065, "learning_rate": 9.658055608091161e-05, "loss": -0.1467, "step": 9750 }, { "epoch": 27.11111111111111, "grad_norm": 0.3137735426425934, "learning_rate": 9.657053293046276e-05, "loss": -0.1489, "step": 9760 }, { "epoch": 27.13888888888889, "grad_norm": 0.328193336725235, "learning_rate": 9.656049563311564e-05, "loss": -0.1527, "step": 9770 }, { "epoch": 27.166666666666668, "grad_norm": 0.424761563539505, "learning_rate": 9.655044419191929e-05, "loss": -0.1461, "step": 9780 }, { "epoch": 27.194444444444443, "grad_norm": 0.28153905272483826, "learning_rate": 9.654037860992711e-05, "loss": -0.1516, "step": 9790 }, { "epoch": 27.22222222222222, "grad_norm": 0.5571033954620361, "learning_rate": 9.653029889019672e-05, "loss": -0.1492, "step": 9800 }, { "epoch": 27.25, "grad_norm": 0.5051412582397461, "learning_rate": 9.65202050357901e-05, "loss": -0.153, "step": 9810 }, { "epoch": 27.27777777777778, "grad_norm": 0.3450796902179718, "learning_rate": 9.651009704977347e-05, "loss": -0.1559, "step": 9820 }, { "epoch": 27.305555555555557, "grad_norm": 0.30056247115135193, "learning_rate": 9.649997493521738e-05, "loss": -0.156, "step": 9830 }, { "epoch": 27.333333333333332, "grad_norm": 0.37888777256011963, "learning_rate": 9.64898386951967e-05, "loss": -0.1585, "step": 9840 }, { "epoch": 27.36111111111111, "grad_norm": 0.31867602467536926, "learning_rate": 9.647968833279049e-05, "loss": -0.1544, "step": 9850 }, { "epoch": 27.38888888888889, "grad_norm": 0.4982568621635437, "learning_rate": 9.646952385108218e-05, "loss": -0.148, "step": 9860 }, { "epoch": 27.416666666666668, "grad_norm": 0.32491442561149597, "learning_rate": 9.645934525315951e-05, "loss": -0.157, "step": 9870 }, { "epoch": 27.444444444444443, "grad_norm": 0.3200719952583313, "learning_rate": 9.644915254211442e-05, "loss": -0.1516, "step": 9880 }, { "epoch": 27.47222222222222, "grad_norm": 0.30698683857917786, "learning_rate": 9.643894572104321e-05, "loss": -0.1531, "step": 9890 }, { "epoch": 27.5, "grad_norm": 0.3628738522529602, "learning_rate": 9.642872479304644e-05, "loss": -0.1513, "step": 9900 }, { "epoch": 27.52777777777778, "grad_norm": 0.2934640944004059, "learning_rate": 9.641848976122895e-05, "loss": -0.1523, "step": 9910 }, { "epoch": 27.555555555555557, "grad_norm": 0.34396594762802124, "learning_rate": 9.64082406286999e-05, "loss": -0.1624, "step": 9920 }, { "epoch": 27.583333333333332, "grad_norm": 0.30109867453575134, "learning_rate": 9.639797739857269e-05, "loss": -0.1425, "step": 9930 }, { "epoch": 27.61111111111111, "grad_norm": 0.2531625032424927, "learning_rate": 9.638770007396498e-05, "loss": -0.1532, "step": 9940 }, { "epoch": 27.63888888888889, "grad_norm": 0.34572967886924744, "learning_rate": 9.63774086579988e-05, "loss": -0.1431, "step": 9950 }, { "epoch": 27.666666666666668, "grad_norm": 0.33876892924308777, "learning_rate": 9.63671031538004e-05, "loss": -0.1571, "step": 9960 }, { "epoch": 27.694444444444443, "grad_norm": 0.38283026218414307, "learning_rate": 9.635678356450031e-05, "loss": -0.1496, "step": 9970 }, { "epoch": 27.72222222222222, "grad_norm": 0.2683168053627014, "learning_rate": 9.634644989323336e-05, "loss": -0.1555, "step": 9980 }, { "epoch": 27.75, "grad_norm": 0.4105791449546814, "learning_rate": 9.633610214313861e-05, "loss": -0.1532, "step": 9990 }, { "epoch": 27.77777777777778, "grad_norm": 0.45063546299934387, "learning_rate": 9.632574031735951e-05, "loss": -0.1517, "step": 10000 }, { "epoch": 27.805555555555557, "grad_norm": 0.4413215219974518, "learning_rate": 9.631536441904364e-05, "loss": -0.8985, "step": 10010 }, { "epoch": 27.833333333333332, "grad_norm": 0.3667204976081848, "learning_rate": 9.630497445134293e-05, "loss": -0.8905, "step": 10020 }, { "epoch": 27.86111111111111, "grad_norm": 0.3181016445159912, "learning_rate": 9.62945704174136e-05, "loss": -0.8986, "step": 10030 }, { "epoch": 27.88888888888889, "grad_norm": 0.3498016893863678, "learning_rate": 9.628415232041612e-05, "loss": -0.8951, "step": 10040 }, { "epoch": 27.916666666666668, "grad_norm": 0.397929847240448, "learning_rate": 9.627372016351524e-05, "loss": -0.8944, "step": 10050 }, { "epoch": 27.944444444444443, "grad_norm": 0.35696080327033997, "learning_rate": 9.626327394987995e-05, "loss": -0.8979, "step": 10060 }, { "epoch": 27.97222222222222, "grad_norm": 0.45300713181495667, "learning_rate": 9.625281368268355e-05, "loss": -0.8964, "step": 10070 }, { "epoch": 28.0, "grad_norm": 0.267633855342865, "learning_rate": 9.624233936510357e-05, "loss": -0.8935, "step": 10080 }, { "epoch": 28.02777777777778, "grad_norm": 0.5066948533058167, "learning_rate": 9.623185100032187e-05, "loss": -0.8927, "step": 10090 }, { "epoch": 28.055555555555557, "grad_norm": 0.4411720633506775, "learning_rate": 9.62213485915245e-05, "loss": -0.8962, "step": 10100 }, { "epoch": 28.083333333333332, "grad_norm": 0.2944292426109314, "learning_rate": 9.621083214190186e-05, "loss": -0.9011, "step": 10110 }, { "epoch": 28.11111111111111, "grad_norm": 0.3792889714241028, "learning_rate": 9.62003016546485e-05, "loss": -0.8901, "step": 10120 }, { "epoch": 28.13888888888889, "grad_norm": 0.34925344586372375, "learning_rate": 9.618975713296339e-05, "loss": -0.8998, "step": 10130 }, { "epoch": 28.166666666666668, "grad_norm": 0.32733288407325745, "learning_rate": 9.61791985800496e-05, "loss": -0.9023, "step": 10140 }, { "epoch": 28.194444444444443, "grad_norm": 0.328337699174881, "learning_rate": 9.616862599911458e-05, "loss": -0.8946, "step": 10150 }, { "epoch": 28.22222222222222, "grad_norm": 0.3169555366039276, "learning_rate": 9.615803939337e-05, "loss": -0.8969, "step": 10160 }, { "epoch": 28.25, "grad_norm": 0.43647488951683044, "learning_rate": 9.614743876603178e-05, "loss": -0.8968, "step": 10170 }, { "epoch": 28.27777777777778, "grad_norm": 0.38485461473464966, "learning_rate": 9.613682412032013e-05, "loss": -0.8911, "step": 10180 }, { "epoch": 28.305555555555557, "grad_norm": 0.47337520122528076, "learning_rate": 9.612619545945947e-05, "loss": -0.8863, "step": 10190 }, { "epoch": 28.333333333333332, "grad_norm": 0.6039876937866211, "learning_rate": 9.611555278667852e-05, "loss": -0.8962, "step": 10200 }, { "epoch": 28.36111111111111, "grad_norm": 0.3283575773239136, "learning_rate": 9.610489610521024e-05, "loss": -0.9006, "step": 10210 }, { "epoch": 28.38888888888889, "grad_norm": 0.423964262008667, "learning_rate": 9.609422541829187e-05, "loss": -0.8957, "step": 10220 }, { "epoch": 28.416666666666668, "grad_norm": 0.26758304238319397, "learning_rate": 9.608354072916486e-05, "loss": -0.8856, "step": 10230 }, { "epoch": 28.444444444444443, "grad_norm": 0.30135059356689453, "learning_rate": 9.607284204107493e-05, "loss": -0.8925, "step": 10240 }, { "epoch": 28.47222222222222, "grad_norm": 0.3141939640045166, "learning_rate": 9.606212935727208e-05, "loss": -0.8996, "step": 10250 }, { "epoch": 28.5, "grad_norm": 0.4265463948249817, "learning_rate": 9.605140268101052e-05, "loss": -0.8978, "step": 10260 }, { "epoch": 28.52777777777778, "grad_norm": 0.3648260235786438, "learning_rate": 9.604066201554875e-05, "loss": -0.8946, "step": 10270 }, { "epoch": 28.555555555555557, "grad_norm": 0.4304715692996979, "learning_rate": 9.60299073641495e-05, "loss": -0.9023, "step": 10280 }, { "epoch": 28.583333333333332, "grad_norm": 0.3241228461265564, "learning_rate": 9.601913873007974e-05, "loss": -0.9017, "step": 10290 }, { "epoch": 28.61111111111111, "grad_norm": 0.3692132532596588, "learning_rate": 9.60083561166107e-05, "loss": -0.8964, "step": 10300 }, { "epoch": 28.63888888888889, "grad_norm": 0.32700785994529724, "learning_rate": 9.599755952701783e-05, "loss": -0.8937, "step": 10310 }, { "epoch": 28.666666666666668, "grad_norm": 0.29991111159324646, "learning_rate": 9.598674896458089e-05, "loss": -0.8994, "step": 10320 }, { "epoch": 28.694444444444443, "grad_norm": 0.23473036289215088, "learning_rate": 9.597592443258383e-05, "loss": -0.907, "step": 10330 }, { "epoch": 28.72222222222222, "grad_norm": 0.2687413990497589, "learning_rate": 9.596508593431483e-05, "loss": -0.9058, "step": 10340 }, { "epoch": 28.75, "grad_norm": 0.29161959886550903, "learning_rate": 9.59542334730664e-05, "loss": -0.8986, "step": 10350 }, { "epoch": 28.77777777777778, "grad_norm": 0.33524996042251587, "learning_rate": 9.594336705213516e-05, "loss": -0.9031, "step": 10360 }, { "epoch": 28.805555555555557, "grad_norm": 0.2407703697681427, "learning_rate": 9.593248667482208e-05, "loss": -0.9027, "step": 10370 }, { "epoch": 28.833333333333332, "grad_norm": 0.39678120613098145, "learning_rate": 9.592159234443233e-05, "loss": -0.8961, "step": 10380 }, { "epoch": 28.86111111111111, "grad_norm": 0.3446725904941559, "learning_rate": 9.59106840642753e-05, "loss": -0.891, "step": 10390 }, { "epoch": 28.88888888888889, "grad_norm": 0.42379501461982727, "learning_rate": 9.589976183766467e-05, "loss": -0.8969, "step": 10400 }, { "epoch": 28.916666666666668, "grad_norm": 0.47293800115585327, "learning_rate": 9.58888256679183e-05, "loss": -0.8958, "step": 10410 }, { "epoch": 28.944444444444443, "grad_norm": 0.3363898992538452, "learning_rate": 9.587787555835832e-05, "loss": -0.9011, "step": 10420 }, { "epoch": 28.97222222222222, "grad_norm": 0.40296781063079834, "learning_rate": 9.586691151231107e-05, "loss": -0.8927, "step": 10430 }, { "epoch": 29.0, "grad_norm": 0.3040519058704376, "learning_rate": 9.585593353310715e-05, "loss": -0.8988, "step": 10440 }, { "epoch": 29.02777777777778, "grad_norm": 0.3872525691986084, "learning_rate": 9.58449416240814e-05, "loss": -0.898, "step": 10450 }, { "epoch": 29.055555555555557, "grad_norm": 0.4394565224647522, "learning_rate": 9.583393578857283e-05, "loss": -0.8989, "step": 10460 }, { "epoch": 29.083333333333332, "grad_norm": 0.32084885239601135, "learning_rate": 9.582291602992474e-05, "loss": -0.9044, "step": 10470 }, { "epoch": 29.11111111111111, "grad_norm": 0.4502035677433014, "learning_rate": 9.581188235148466e-05, "loss": -0.899, "step": 10480 }, { "epoch": 29.13888888888889, "grad_norm": 0.331834614276886, "learning_rate": 9.58008347566043e-05, "loss": -0.904, "step": 10490 }, { "epoch": 29.166666666666668, "grad_norm": 0.46099981665611267, "learning_rate": 9.578977324863965e-05, "loss": -0.9008, "step": 10500 }, { "epoch": 29.194444444444443, "grad_norm": 0.43328383564949036, "learning_rate": 9.577869783095089e-05, "loss": -0.8965, "step": 10510 }, { "epoch": 29.22222222222222, "grad_norm": 0.5422468781471252, "learning_rate": 9.576760850690245e-05, "loss": -0.8958, "step": 10520 }, { "epoch": 29.25, "grad_norm": 0.43738463521003723, "learning_rate": 9.575650527986298e-05, "loss": -0.8943, "step": 10530 }, { "epoch": 29.27777777777778, "grad_norm": 0.5358120799064636, "learning_rate": 9.574538815320531e-05, "loss": -0.9014, "step": 10540 }, { "epoch": 29.305555555555557, "grad_norm": 0.30259788036346436, "learning_rate": 9.573425713030656e-05, "loss": -0.8935, "step": 10550 }, { "epoch": 29.333333333333332, "grad_norm": 0.2719493508338928, "learning_rate": 9.572311221454806e-05, "loss": -0.9008, "step": 10560 }, { "epoch": 29.36111111111111, "grad_norm": 0.23243238031864166, "learning_rate": 9.57119534093153e-05, "loss": -0.8984, "step": 10570 }, { "epoch": 29.38888888888889, "grad_norm": 0.403144896030426, "learning_rate": 9.570078071799806e-05, "loss": -0.8989, "step": 10580 }, { "epoch": 29.416666666666668, "grad_norm": 0.2973700761795044, "learning_rate": 9.568959414399028e-05, "loss": -0.9018, "step": 10590 }, { "epoch": 29.444444444444443, "grad_norm": 0.28306588530540466, "learning_rate": 9.567839369069018e-05, "loss": -0.8928, "step": 10600 }, { "epoch": 29.47222222222222, "grad_norm": 0.4713895320892334, "learning_rate": 9.566717936150013e-05, "loss": -0.8926, "step": 10610 }, { "epoch": 29.5, "grad_norm": 0.41729646921157837, "learning_rate": 9.565595115982678e-05, "loss": -0.9027, "step": 10620 }, { "epoch": 29.52777777777778, "grad_norm": 0.3758026361465454, "learning_rate": 9.564470908908094e-05, "loss": -0.8952, "step": 10630 }, { "epoch": 29.555555555555557, "grad_norm": 0.4203490912914276, "learning_rate": 9.563345315267764e-05, "loss": -0.8988, "step": 10640 }, { "epoch": 29.583333333333332, "grad_norm": 0.3504927456378937, "learning_rate": 9.562218335403616e-05, "loss": -0.9004, "step": 10650 }, { "epoch": 29.61111111111111, "grad_norm": 0.2997572422027588, "learning_rate": 9.561089969657999e-05, "loss": -0.9008, "step": 10660 }, { "epoch": 29.63888888888889, "grad_norm": 0.2532881498336792, "learning_rate": 9.559960218373673e-05, "loss": -0.8951, "step": 10670 }, { "epoch": 29.666666666666668, "grad_norm": 0.3655722737312317, "learning_rate": 9.558829081893836e-05, "loss": -0.9024, "step": 10680 }, { "epoch": 29.694444444444443, "grad_norm": 0.31882545351982117, "learning_rate": 9.55769656056209e-05, "loss": -0.9013, "step": 10690 }, { "epoch": 29.72222222222222, "grad_norm": 0.3934582471847534, "learning_rate": 9.556562654722469e-05, "loss": -0.8967, "step": 10700 }, { "epoch": 29.75, "grad_norm": 0.4517625570297241, "learning_rate": 9.555427364719422e-05, "loss": -0.8935, "step": 10710 }, { "epoch": 29.77777777777778, "grad_norm": 0.28436702489852905, "learning_rate": 9.55429069089782e-05, "loss": -0.8959, "step": 10720 }, { "epoch": 29.805555555555557, "grad_norm": 0.4410730004310608, "learning_rate": 9.553152633602956e-05, "loss": -0.8945, "step": 10730 }, { "epoch": 29.833333333333332, "grad_norm": 0.3375900685787201, "learning_rate": 9.552013193180543e-05, "loss": -0.893, "step": 10740 }, { "epoch": 29.86111111111111, "grad_norm": 0.49431079626083374, "learning_rate": 9.550872369976707e-05, "loss": -0.8981, "step": 10750 }, { "epoch": 29.88888888888889, "grad_norm": 0.3486853837966919, "learning_rate": 9.549730164338007e-05, "loss": -0.8981, "step": 10760 }, { "epoch": 29.916666666666668, "grad_norm": 0.3384708762168884, "learning_rate": 9.548586576611408e-05, "loss": -0.9017, "step": 10770 }, { "epoch": 29.944444444444443, "grad_norm": 0.32802844047546387, "learning_rate": 9.54744160714431e-05, "loss": -0.9017, "step": 10780 }, { "epoch": 29.97222222222222, "grad_norm": 0.3697637915611267, "learning_rate": 9.546295256284516e-05, "loss": -0.9033, "step": 10790 }, { "epoch": 30.0, "grad_norm": 0.28220242261886597, "learning_rate": 9.545147524380265e-05, "loss": -0.8939, "step": 10800 }, { "epoch": 30.02777777777778, "grad_norm": 0.3634759187698364, "learning_rate": 9.543998411780201e-05, "loss": -0.8982, "step": 10810 }, { "epoch": 30.055555555555557, "grad_norm": 0.2876187860965729, "learning_rate": 9.542847918833397e-05, "loss": -0.8998, "step": 10820 }, { "epoch": 30.083333333333332, "grad_norm": 0.3297697901725769, "learning_rate": 9.541696045889343e-05, "loss": -0.9045, "step": 10830 }, { "epoch": 30.11111111111111, "grad_norm": 0.38367903232574463, "learning_rate": 9.540542793297947e-05, "loss": -0.9065, "step": 10840 }, { "epoch": 30.13888888888889, "grad_norm": 0.3807423412799835, "learning_rate": 9.539388161409537e-05, "loss": -0.9017, "step": 10850 }, { "epoch": 30.166666666666668, "grad_norm": 0.413211852312088, "learning_rate": 9.538232150574857e-05, "loss": -0.8995, "step": 10860 }, { "epoch": 30.194444444444443, "grad_norm": 0.24802270531654358, "learning_rate": 9.537074761145076e-05, "loss": -0.9068, "step": 10870 }, { "epoch": 30.22222222222222, "grad_norm": 0.4426515996456146, "learning_rate": 9.535915993471778e-05, "loss": -0.8995, "step": 10880 }, { "epoch": 30.25, "grad_norm": 0.4292397201061249, "learning_rate": 9.534755847906964e-05, "loss": -0.8939, "step": 10890 }, { "epoch": 30.27777777777778, "grad_norm": 0.46077367663383484, "learning_rate": 9.533594324803057e-05, "loss": -0.9022, "step": 10900 }, { "epoch": 30.305555555555557, "grad_norm": 0.34180718660354614, "learning_rate": 9.532431424512895e-05, "loss": -0.9071, "step": 10910 }, { "epoch": 30.333333333333332, "grad_norm": 0.3864966332912445, "learning_rate": 9.531267147389741e-05, "loss": -0.8974, "step": 10920 }, { "epoch": 30.36111111111111, "grad_norm": 0.4453238844871521, "learning_rate": 9.530101493787266e-05, "loss": -0.8986, "step": 10930 }, { "epoch": 30.38888888888889, "grad_norm": 0.3757612109184265, "learning_rate": 9.528934464059571e-05, "loss": -0.9033, "step": 10940 }, { "epoch": 30.416666666666668, "grad_norm": 0.4471813142299652, "learning_rate": 9.527766058561163e-05, "loss": -0.9042, "step": 10950 }, { "epoch": 30.444444444444443, "grad_norm": 0.2898789942264557, "learning_rate": 9.526596277646976e-05, "loss": -0.8953, "step": 10960 }, { "epoch": 30.47222222222222, "grad_norm": 0.4270773231983185, "learning_rate": 9.525425121672358e-05, "loss": -0.9023, "step": 10970 }, { "epoch": 30.5, "grad_norm": 0.4534004330635071, "learning_rate": 9.524252590993074e-05, "loss": -0.9097, "step": 10980 }, { "epoch": 30.52777777777778, "grad_norm": 0.3385164737701416, "learning_rate": 9.523078685965309e-05, "loss": -0.9018, "step": 10990 }, { "epoch": 30.555555555555557, "grad_norm": 0.24076952040195465, "learning_rate": 9.521903406945664e-05, "loss": -0.9103, "step": 11000 }, { "epoch": 30.583333333333332, "grad_norm": 0.3796299993991852, "learning_rate": 9.520726754291158e-05, "loss": -0.9028, "step": 11010 }, { "epoch": 30.61111111111111, "grad_norm": 0.4641033113002777, "learning_rate": 9.519548728359227e-05, "loss": -0.9034, "step": 11020 }, { "epoch": 30.63888888888889, "grad_norm": 0.36638912558555603, "learning_rate": 9.518369329507726e-05, "loss": -0.9065, "step": 11030 }, { "epoch": 30.666666666666668, "grad_norm": 0.3842536509037018, "learning_rate": 9.51718855809492e-05, "loss": -0.9117, "step": 11040 }, { "epoch": 30.694444444444443, "grad_norm": 0.3473201096057892, "learning_rate": 9.516006414479502e-05, "loss": -0.8999, "step": 11050 }, { "epoch": 30.72222222222222, "grad_norm": 0.33963248133659363, "learning_rate": 9.514822899020572e-05, "loss": -0.8995, "step": 11060 }, { "epoch": 30.75, "grad_norm": 0.2542429566383362, "learning_rate": 9.513638012077654e-05, "loss": -0.9053, "step": 11070 }, { "epoch": 30.77777777777778, "grad_norm": 0.22078482806682587, "learning_rate": 9.512451754010683e-05, "loss": -0.9103, "step": 11080 }, { "epoch": 30.805555555555557, "grad_norm": 0.267119437456131, "learning_rate": 9.511264125180013e-05, "loss": -0.9127, "step": 11090 }, { "epoch": 30.833333333333332, "grad_norm": 0.36275187134742737, "learning_rate": 9.510075125946414e-05, "loss": -0.9075, "step": 11100 }, { "epoch": 30.86111111111111, "grad_norm": 0.2611982226371765, "learning_rate": 9.508884756671075e-05, "loss": -0.9049, "step": 11110 }, { "epoch": 30.88888888888889, "grad_norm": 0.35331907868385315, "learning_rate": 9.507693017715596e-05, "loss": -0.9101, "step": 11120 }, { "epoch": 30.916666666666668, "grad_norm": 0.3473776578903198, "learning_rate": 9.506499909441997e-05, "loss": -0.9067, "step": 11130 }, { "epoch": 30.944444444444443, "grad_norm": 0.20211371779441833, "learning_rate": 9.505305432212713e-05, "loss": -0.911, "step": 11140 }, { "epoch": 30.97222222222222, "grad_norm": 0.39164698123931885, "learning_rate": 9.504109586390595e-05, "loss": -0.9073, "step": 11150 }, { "epoch": 31.0, "grad_norm": 0.36393290758132935, "learning_rate": 9.502912372338908e-05, "loss": -0.9057, "step": 11160 }, { "epoch": 31.02777777777778, "grad_norm": 0.32965582609176636, "learning_rate": 9.501713790421335e-05, "loss": -0.9072, "step": 11170 }, { "epoch": 31.055555555555557, "grad_norm": 0.35052618384361267, "learning_rate": 9.500513841001974e-05, "loss": -0.9085, "step": 11180 }, { "epoch": 31.083333333333332, "grad_norm": 0.21491776406764984, "learning_rate": 9.499312524445336e-05, "loss": -0.9043, "step": 11190 }, { "epoch": 31.11111111111111, "grad_norm": 0.35496726632118225, "learning_rate": 9.498109841116351e-05, "loss": -0.9121, "step": 11200 }, { "epoch": 31.13888888888889, "grad_norm": 0.39971140027046204, "learning_rate": 9.496905791380363e-05, "loss": -0.8933, "step": 11210 }, { "epoch": 31.166666666666668, "grad_norm": 0.3024987280368805, "learning_rate": 9.495700375603129e-05, "loss": -0.9119, "step": 11220 }, { "epoch": 31.194444444444443, "grad_norm": 0.47013670206069946, "learning_rate": 9.494493594150822e-05, "loss": -0.9019, "step": 11230 }, { "epoch": 31.22222222222222, "grad_norm": 0.2639388144016266, "learning_rate": 9.493285447390032e-05, "loss": -0.9075, "step": 11240 }, { "epoch": 31.25, "grad_norm": 0.36191096901893616, "learning_rate": 9.492075935687761e-05, "loss": -0.9085, "step": 11250 }, { "epoch": 31.27777777777778, "grad_norm": 0.2738473415374756, "learning_rate": 9.490865059411427e-05, "loss": -0.9112, "step": 11260 }, { "epoch": 31.305555555555557, "grad_norm": 0.3414211869239807, "learning_rate": 9.489652818928863e-05, "loss": -0.9074, "step": 11270 }, { "epoch": 31.333333333333332, "grad_norm": 0.2284814417362213, "learning_rate": 9.488439214608315e-05, "loss": -0.9038, "step": 11280 }, { "epoch": 31.36111111111111, "grad_norm": 0.2812272608280182, "learning_rate": 9.487224246818444e-05, "loss": -0.9043, "step": 11290 }, { "epoch": 31.38888888888889, "grad_norm": 0.33170706033706665, "learning_rate": 9.486007915928325e-05, "loss": -0.9131, "step": 11300 }, { "epoch": 31.416666666666668, "grad_norm": 0.3008541762828827, "learning_rate": 9.484790222307448e-05, "loss": -0.9107, "step": 11310 }, { "epoch": 31.444444444444443, "grad_norm": 0.30985623598098755, "learning_rate": 9.483571166325716e-05, "loss": -0.9013, "step": 11320 }, { "epoch": 31.47222222222222, "grad_norm": 0.3566613495349884, "learning_rate": 9.482350748353444e-05, "loss": -0.9044, "step": 11330 }, { "epoch": 31.5, "grad_norm": 0.23595254123210907, "learning_rate": 9.481128968761363e-05, "loss": -0.9051, "step": 11340 }, { "epoch": 31.52777777777778, "grad_norm": 0.3103790283203125, "learning_rate": 9.479905827920621e-05, "loss": -0.9036, "step": 11350 }, { "epoch": 31.555555555555557, "grad_norm": 0.31492412090301514, "learning_rate": 9.478681326202773e-05, "loss": -0.9054, "step": 11360 }, { "epoch": 31.583333333333332, "grad_norm": 0.45255446434020996, "learning_rate": 9.477455463979791e-05, "loss": -0.9091, "step": 11370 }, { "epoch": 31.61111111111111, "grad_norm": 0.44914814829826355, "learning_rate": 9.476228241624059e-05, "loss": -0.8943, "step": 11380 }, { "epoch": 31.63888888888889, "grad_norm": 0.4430846571922302, "learning_rate": 9.474999659508374e-05, "loss": -0.9072, "step": 11390 }, { "epoch": 31.666666666666668, "grad_norm": 0.41077399253845215, "learning_rate": 9.47376971800595e-05, "loss": -0.9149, "step": 11400 }, { "epoch": 31.694444444444443, "grad_norm": 0.361754447221756, "learning_rate": 9.472538417490409e-05, "loss": -0.9013, "step": 11410 }, { "epoch": 31.72222222222222, "grad_norm": 0.6094354391098022, "learning_rate": 9.471305758335784e-05, "loss": -0.9056, "step": 11420 }, { "epoch": 31.75, "grad_norm": 0.307307630777359, "learning_rate": 9.47007174091653e-05, "loss": -0.9021, "step": 11430 }, { "epoch": 31.77777777777778, "grad_norm": 0.39102470874786377, "learning_rate": 9.468836365607507e-05, "loss": -0.8989, "step": 11440 }, { "epoch": 31.805555555555557, "grad_norm": 0.35613545775413513, "learning_rate": 9.467599632783988e-05, "loss": -0.9, "step": 11450 }, { "epoch": 31.833333333333332, "grad_norm": 0.2521402835845947, "learning_rate": 9.466361542821662e-05, "loss": -0.8995, "step": 11460 }, { "epoch": 31.86111111111111, "grad_norm": 0.3276854455471039, "learning_rate": 9.465122096096625e-05, "loss": -0.9092, "step": 11470 }, { "epoch": 31.88888888888889, "grad_norm": 0.36480987071990967, "learning_rate": 9.463881292985391e-05, "loss": -0.9068, "step": 11480 }, { "epoch": 31.916666666666668, "grad_norm": 0.6075291633605957, "learning_rate": 9.462639133864881e-05, "loss": -0.9074, "step": 11490 }, { "epoch": 31.944444444444443, "grad_norm": 0.2808277904987335, "learning_rate": 9.461395619112432e-05, "loss": -0.9071, "step": 11500 }, { "epoch": 31.97222222222222, "grad_norm": 0.30482590198516846, "learning_rate": 9.460150749105791e-05, "loss": -0.9081, "step": 11510 }, { "epoch": 32.0, "grad_norm": 0.284229576587677, "learning_rate": 9.458904524223116e-05, "loss": -0.9054, "step": 11520 }, { "epoch": 32.02777777777778, "grad_norm": 0.4659735858440399, "learning_rate": 9.457656944842976e-05, "loss": -0.9011, "step": 11530 }, { "epoch": 32.05555555555556, "grad_norm": 0.43607282638549805, "learning_rate": 9.456408011344353e-05, "loss": -0.9112, "step": 11540 }, { "epoch": 32.083333333333336, "grad_norm": 0.23083098232746124, "learning_rate": 9.455157724106643e-05, "loss": -0.9042, "step": 11550 }, { "epoch": 32.111111111111114, "grad_norm": 0.34786587953567505, "learning_rate": 9.453906083509647e-05, "loss": -0.8974, "step": 11560 }, { "epoch": 32.138888888888886, "grad_norm": 0.36340901255607605, "learning_rate": 9.45265308993358e-05, "loss": -0.9086, "step": 11570 }, { "epoch": 32.166666666666664, "grad_norm": 0.4006851017475128, "learning_rate": 9.451398743759071e-05, "loss": -0.902, "step": 11580 }, { "epoch": 32.19444444444444, "grad_norm": 0.2349085658788681, "learning_rate": 9.450143045367156e-05, "loss": -0.91, "step": 11590 }, { "epoch": 32.22222222222222, "grad_norm": 0.3073784410953522, "learning_rate": 9.448885995139283e-05, "loss": -0.9077, "step": 11600 }, { "epoch": 32.25, "grad_norm": 0.3956170082092285, "learning_rate": 9.44762759345731e-05, "loss": -0.9136, "step": 11610 }, { "epoch": 32.27777777777778, "grad_norm": 0.381334125995636, "learning_rate": 9.446367840703509e-05, "loss": -0.9079, "step": 11620 }, { "epoch": 32.30555555555556, "grad_norm": 0.3894904851913452, "learning_rate": 9.445106737260556e-05, "loss": -0.9117, "step": 11630 }, { "epoch": 32.333333333333336, "grad_norm": 0.36747264862060547, "learning_rate": 9.443844283511543e-05, "loss": -0.9078, "step": 11640 }, { "epoch": 32.361111111111114, "grad_norm": 0.27697035670280457, "learning_rate": 9.442580479839968e-05, "loss": -0.905, "step": 11650 }, { "epoch": 32.388888888888886, "grad_norm": 0.36446619033813477, "learning_rate": 9.441315326629745e-05, "loss": -0.9057, "step": 11660 }, { "epoch": 32.416666666666664, "grad_norm": 0.35630008578300476, "learning_rate": 9.44004882426519e-05, "loss": -0.9054, "step": 11670 }, { "epoch": 32.44444444444444, "grad_norm": 0.27435973286628723, "learning_rate": 9.438780973131037e-05, "loss": -0.9059, "step": 11680 }, { "epoch": 32.47222222222222, "grad_norm": 0.41273197531700134, "learning_rate": 9.437511773612423e-05, "loss": -0.9022, "step": 11690 }, { "epoch": 32.5, "grad_norm": 0.3516541123390198, "learning_rate": 9.436241226094896e-05, "loss": -0.8963, "step": 11700 }, { "epoch": 32.52777777777778, "grad_norm": 0.2866617441177368, "learning_rate": 9.434969330964418e-05, "loss": -0.9049, "step": 11710 }, { "epoch": 32.55555555555556, "grad_norm": 0.3727327287197113, "learning_rate": 9.433696088607356e-05, "loss": -0.9082, "step": 11720 }, { "epoch": 32.583333333333336, "grad_norm": 0.2530583441257477, "learning_rate": 9.432421499410486e-05, "loss": -0.9096, "step": 11730 }, { "epoch": 32.611111111111114, "grad_norm": 0.29855087399482727, "learning_rate": 9.431145563760998e-05, "loss": -0.9114, "step": 11740 }, { "epoch": 32.638888888888886, "grad_norm": 0.31220096349716187, "learning_rate": 9.429868282046484e-05, "loss": -0.9013, "step": 11750 }, { "epoch": 32.666666666666664, "grad_norm": 0.3270179331302643, "learning_rate": 9.428589654654951e-05, "loss": -0.9002, "step": 11760 }, { "epoch": 32.69444444444444, "grad_norm": 0.3411288559436798, "learning_rate": 9.42730968197481e-05, "loss": -0.9082, "step": 11770 }, { "epoch": 32.72222222222222, "grad_norm": 0.3866243362426758, "learning_rate": 9.426028364394883e-05, "loss": -0.9111, "step": 11780 }, { "epoch": 32.75, "grad_norm": 0.36706796288490295, "learning_rate": 9.424745702304402e-05, "loss": -0.9131, "step": 11790 }, { "epoch": 32.77777777777778, "grad_norm": 0.2736358046531677, "learning_rate": 9.423461696093006e-05, "loss": -0.9059, "step": 11800 }, { "epoch": 32.80555555555556, "grad_norm": 0.32337686419487, "learning_rate": 9.422176346150741e-05, "loss": -0.9087, "step": 11810 }, { "epoch": 32.833333333333336, "grad_norm": 0.320103257894516, "learning_rate": 9.420889652868063e-05, "loss": -0.9073, "step": 11820 }, { "epoch": 32.861111111111114, "grad_norm": 0.28670603036880493, "learning_rate": 9.419601616635836e-05, "loss": -0.9088, "step": 11830 }, { "epoch": 32.888888888888886, "grad_norm": 0.33385559916496277, "learning_rate": 9.418312237845331e-05, "loss": -0.9064, "step": 11840 }, { "epoch": 32.916666666666664, "grad_norm": 0.35711193084716797, "learning_rate": 9.417021516888225e-05, "loss": -0.9095, "step": 11850 }, { "epoch": 32.94444444444444, "grad_norm": 0.35193657875061035, "learning_rate": 9.415729454156608e-05, "loss": -0.9106, "step": 11860 }, { "epoch": 32.97222222222222, "grad_norm": 0.2985500693321228, "learning_rate": 9.414436050042973e-05, "loss": -0.9073, "step": 11870 }, { "epoch": 33.0, "grad_norm": 0.3026544749736786, "learning_rate": 9.413141304940223e-05, "loss": -0.9102, "step": 11880 }, { "epoch": 33.02777777777778, "grad_norm": 0.33914488554000854, "learning_rate": 9.411845219241666e-05, "loss": -0.9122, "step": 11890 }, { "epoch": 33.05555555555556, "grad_norm": 0.3244518041610718, "learning_rate": 9.410547793341021e-05, "loss": -0.9086, "step": 11900 }, { "epoch": 33.083333333333336, "grad_norm": 0.2889268398284912, "learning_rate": 9.409249027632408e-05, "loss": -0.9143, "step": 11910 }, { "epoch": 33.111111111111114, "grad_norm": 0.3485279977321625, "learning_rate": 9.407948922510362e-05, "loss": -0.9136, "step": 11920 }, { "epoch": 33.138888888888886, "grad_norm": 0.3871753513813019, "learning_rate": 9.406647478369817e-05, "loss": -0.9124, "step": 11930 }, { "epoch": 33.166666666666664, "grad_norm": 0.3154606521129608, "learning_rate": 9.405344695606118e-05, "loss": -0.9124, "step": 11940 }, { "epoch": 33.19444444444444, "grad_norm": 0.38019949197769165, "learning_rate": 9.404040574615018e-05, "loss": -0.91, "step": 11950 }, { "epoch": 33.22222222222222, "grad_norm": 0.28843772411346436, "learning_rate": 9.402735115792674e-05, "loss": -0.9062, "step": 11960 }, { "epoch": 33.25, "grad_norm": 0.3560175597667694, "learning_rate": 9.401428319535649e-05, "loss": -0.9159, "step": 11970 }, { "epoch": 33.27777777777778, "grad_norm": 0.25800082087516785, "learning_rate": 9.400120186240912e-05, "loss": -0.9077, "step": 11980 }, { "epoch": 33.30555555555556, "grad_norm": 0.40746352076530457, "learning_rate": 9.398810716305844e-05, "loss": -0.9111, "step": 11990 }, { "epoch": 33.333333333333336, "grad_norm": 0.3389308452606201, "learning_rate": 9.397499910128222e-05, "loss": -0.9106, "step": 12000 }, { "epoch": 33.361111111111114, "grad_norm": 0.3444628119468689, "learning_rate": 9.396187768106237e-05, "loss": -0.9137, "step": 12010 }, { "epoch": 33.388888888888886, "grad_norm": 0.30042701959609985, "learning_rate": 9.394874290638482e-05, "loss": -0.9109, "step": 12020 }, { "epoch": 33.416666666666664, "grad_norm": 0.3393172025680542, "learning_rate": 9.393559478123959e-05, "loss": -0.9101, "step": 12030 }, { "epoch": 33.44444444444444, "grad_norm": 0.3636147677898407, "learning_rate": 9.39224333096207e-05, "loss": -0.9088, "step": 12040 }, { "epoch": 33.47222222222222, "grad_norm": 0.25161921977996826, "learning_rate": 9.390925849552629e-05, "loss": -0.909, "step": 12050 }, { "epoch": 33.5, "grad_norm": 0.30666014552116394, "learning_rate": 9.389607034295849e-05, "loss": -0.9145, "step": 12060 }, { "epoch": 33.52777777777778, "grad_norm": 0.2945348918437958, "learning_rate": 9.388286885592355e-05, "loss": -0.8956, "step": 12070 }, { "epoch": 33.55555555555556, "grad_norm": 0.22672443091869354, "learning_rate": 9.386965403843168e-05, "loss": -0.9109, "step": 12080 }, { "epoch": 33.583333333333336, "grad_norm": 0.40622803568840027, "learning_rate": 9.385642589449726e-05, "loss": -0.9124, "step": 12090 }, { "epoch": 33.611111111111114, "grad_norm": 0.36037391424179077, "learning_rate": 9.38431844281386e-05, "loss": -0.9155, "step": 12100 }, { "epoch": 33.638888888888886, "grad_norm": 0.3500380516052246, "learning_rate": 9.38299296433781e-05, "loss": -0.9076, "step": 12110 }, { "epoch": 33.666666666666664, "grad_norm": 0.32032087445259094, "learning_rate": 9.381666154424226e-05, "loss": -0.9108, "step": 12120 }, { "epoch": 33.69444444444444, "grad_norm": 0.37031638622283936, "learning_rate": 9.380338013476157e-05, "loss": -0.9128, "step": 12130 }, { "epoch": 33.72222222222222, "grad_norm": 0.505394697189331, "learning_rate": 9.379008541897054e-05, "loss": -0.9098, "step": 12140 }, { "epoch": 33.75, "grad_norm": 0.2374287247657776, "learning_rate": 9.377677740090777e-05, "loss": -0.9001, "step": 12150 }, { "epoch": 33.77777777777778, "grad_norm": 0.3417874276638031, "learning_rate": 9.376345608461588e-05, "loss": -0.9095, "step": 12160 }, { "epoch": 33.80555555555556, "grad_norm": 0.3228709101676941, "learning_rate": 9.375012147414155e-05, "loss": -0.915, "step": 12170 }, { "epoch": 33.833333333333336, "grad_norm": 0.3675225079059601, "learning_rate": 9.373677357353545e-05, "loss": -0.9089, "step": 12180 }, { "epoch": 33.861111111111114, "grad_norm": 0.30849847197532654, "learning_rate": 9.372341238685237e-05, "loss": -0.9151, "step": 12190 }, { "epoch": 33.888888888888886, "grad_norm": 0.29448261857032776, "learning_rate": 9.371003791815102e-05, "loss": -0.9161, "step": 12200 }, { "epoch": 33.916666666666664, "grad_norm": 0.3357759118080139, "learning_rate": 9.369665017149429e-05, "loss": -0.9128, "step": 12210 }, { "epoch": 33.94444444444444, "grad_norm": 0.2750263512134552, "learning_rate": 9.368324915094895e-05, "loss": -0.9143, "step": 12220 }, { "epoch": 33.97222222222222, "grad_norm": 0.4284840226173401, "learning_rate": 9.366983486058591e-05, "loss": -0.9064, "step": 12230 }, { "epoch": 34.0, "grad_norm": 0.46664902567863464, "learning_rate": 9.365640730448009e-05, "loss": -0.9074, "step": 12240 }, { "epoch": 34.02777777777778, "grad_norm": 0.4433521628379822, "learning_rate": 9.36429664867104e-05, "loss": -0.9098, "step": 12250 }, { "epoch": 34.05555555555556, "grad_norm": 0.3212355077266693, "learning_rate": 9.362951241135982e-05, "loss": -0.9097, "step": 12260 }, { "epoch": 34.083333333333336, "grad_norm": 0.36400696635246277, "learning_rate": 9.361604508251534e-05, "loss": -0.9158, "step": 12270 }, { "epoch": 34.111111111111114, "grad_norm": 0.2376376837491989, "learning_rate": 9.360256450426799e-05, "loss": -0.9123, "step": 12280 }, { "epoch": 34.138888888888886, "grad_norm": 0.35819560289382935, "learning_rate": 9.358907068071279e-05, "loss": -0.9143, "step": 12290 }, { "epoch": 34.166666666666664, "grad_norm": 0.3060341775417328, "learning_rate": 9.357556361594882e-05, "loss": -0.9176, "step": 12300 }, { "epoch": 34.19444444444444, "grad_norm": 0.2456623762845993, "learning_rate": 9.356204331407917e-05, "loss": -0.9117, "step": 12310 }, { "epoch": 34.22222222222222, "grad_norm": 0.290454626083374, "learning_rate": 9.354850977921094e-05, "loss": -0.9075, "step": 12320 }, { "epoch": 34.25, "grad_norm": 0.26395514607429504, "learning_rate": 9.353496301545529e-05, "loss": -0.9104, "step": 12330 }, { "epoch": 34.27777777777778, "grad_norm": 0.35411617159843445, "learning_rate": 9.352140302692733e-05, "loss": -0.909, "step": 12340 }, { "epoch": 34.30555555555556, "grad_norm": 0.41047346591949463, "learning_rate": 9.350782981774627e-05, "loss": -0.9074, "step": 12350 }, { "epoch": 34.333333333333336, "grad_norm": 0.33574768900871277, "learning_rate": 9.349424339203526e-05, "loss": -0.9076, "step": 12360 }, { "epoch": 34.361111111111114, "grad_norm": 0.37497255206108093, "learning_rate": 9.34806437539215e-05, "loss": -0.9064, "step": 12370 }, { "epoch": 34.388888888888886, "grad_norm": 0.2621542513370514, "learning_rate": 9.346703090753622e-05, "loss": -0.9035, "step": 12380 }, { "epoch": 34.416666666666664, "grad_norm": 0.48540395498275757, "learning_rate": 9.345340485701461e-05, "loss": -0.907, "step": 12390 }, { "epoch": 34.44444444444444, "grad_norm": 0.24383220076560974, "learning_rate": 9.343976560649595e-05, "loss": -0.9137, "step": 12400 }, { "epoch": 34.47222222222222, "grad_norm": 0.3404589295387268, "learning_rate": 9.342611316012344e-05, "loss": -0.9107, "step": 12410 }, { "epoch": 34.5, "grad_norm": 0.27908986806869507, "learning_rate": 9.341244752204437e-05, "loss": -0.9114, "step": 12420 }, { "epoch": 34.52777777777778, "grad_norm": 0.30476558208465576, "learning_rate": 9.339876869640995e-05, "loss": -0.9099, "step": 12430 }, { "epoch": 34.55555555555556, "grad_norm": 0.33251065015792847, "learning_rate": 9.33850766873755e-05, "loss": -0.9102, "step": 12440 }, { "epoch": 34.583333333333336, "grad_norm": 0.32673880457878113, "learning_rate": 9.337137149910028e-05, "loss": -0.9091, "step": 12450 }, { "epoch": 34.611111111111114, "grad_norm": 0.36111801862716675, "learning_rate": 9.335765313574753e-05, "loss": -0.9154, "step": 12460 }, { "epoch": 34.638888888888886, "grad_norm": 0.4011201858520508, "learning_rate": 9.334392160148457e-05, "loss": -0.9135, "step": 12470 }, { "epoch": 34.666666666666664, "grad_norm": 0.3015933036804199, "learning_rate": 9.333017690048264e-05, "loss": -0.9181, "step": 12480 }, { "epoch": 34.69444444444444, "grad_norm": 0.37072476744651794, "learning_rate": 9.331641903691706e-05, "loss": -0.914, "step": 12490 }, { "epoch": 34.72222222222222, "grad_norm": 0.26071399450302124, "learning_rate": 9.330264801496707e-05, "loss": -0.9115, "step": 12500 }, { "epoch": 34.75, "grad_norm": 0.2998230457305908, "learning_rate": 9.328886383881594e-05, "loss": -0.9084, "step": 12510 }, { "epoch": 34.77777777777778, "grad_norm": 0.3013496994972229, "learning_rate": 9.327506651265095e-05, "loss": -0.9097, "step": 12520 }, { "epoch": 34.80555555555556, "grad_norm": 0.31424441933631897, "learning_rate": 9.326125604066338e-05, "loss": -0.9118, "step": 12530 }, { "epoch": 34.833333333333336, "grad_norm": 0.36066317558288574, "learning_rate": 9.324743242704847e-05, "loss": -0.9149, "step": 12540 }, { "epoch": 34.861111111111114, "grad_norm": 0.40320637822151184, "learning_rate": 9.323359567600546e-05, "loss": -0.9091, "step": 12550 }, { "epoch": 34.888888888888886, "grad_norm": 0.2595938742160797, "learning_rate": 9.321974579173761e-05, "loss": -0.9126, "step": 12560 }, { "epoch": 34.916666666666664, "grad_norm": 0.3145144581794739, "learning_rate": 9.320588277845213e-05, "loss": -0.9045, "step": 12570 }, { "epoch": 34.94444444444444, "grad_norm": 0.4254692494869232, "learning_rate": 9.319200664036026e-05, "loss": -0.9088, "step": 12580 }, { "epoch": 34.97222222222222, "grad_norm": 0.28021150827407837, "learning_rate": 9.31781173816772e-05, "loss": -0.9152, "step": 12590 }, { "epoch": 35.0, "grad_norm": 0.3114517629146576, "learning_rate": 9.316421500662212e-05, "loss": -0.9125, "step": 12600 }, { "epoch": 35.02777777777778, "grad_norm": 0.3077056109905243, "learning_rate": 9.31502995194182e-05, "loss": -0.9123, "step": 12610 }, { "epoch": 35.05555555555556, "grad_norm": 0.39213892817497253, "learning_rate": 9.31363709242926e-05, "loss": -0.9096, "step": 12620 }, { "epoch": 35.083333333333336, "grad_norm": 0.3651985228061676, "learning_rate": 9.312242922547647e-05, "loss": -0.9168, "step": 12630 }, { "epoch": 35.111111111111114, "grad_norm": 0.39827316999435425, "learning_rate": 9.310847442720492e-05, "loss": -0.9113, "step": 12640 }, { "epoch": 35.138888888888886, "grad_norm": 0.34706610441207886, "learning_rate": 9.309450653371706e-05, "loss": -0.9118, "step": 12650 }, { "epoch": 35.166666666666664, "grad_norm": 0.2612818777561188, "learning_rate": 9.308052554925595e-05, "loss": -0.9155, "step": 12660 }, { "epoch": 35.19444444444444, "grad_norm": 0.34610307216644287, "learning_rate": 9.306653147806867e-05, "loss": -0.9089, "step": 12670 }, { "epoch": 35.22222222222222, "grad_norm": 0.2930184602737427, "learning_rate": 9.305252432440622e-05, "loss": -0.9179, "step": 12680 }, { "epoch": 35.25, "grad_norm": 0.29817408323287964, "learning_rate": 9.303850409252361e-05, "loss": -0.9144, "step": 12690 }, { "epoch": 35.27777777777778, "grad_norm": 0.32573044300079346, "learning_rate": 9.302447078667985e-05, "loss": -0.9182, "step": 12700 }, { "epoch": 35.30555555555556, "grad_norm": 0.30197885632514954, "learning_rate": 9.301042441113783e-05, "loss": -0.9182, "step": 12710 }, { "epoch": 35.333333333333336, "grad_norm": 0.24117401242256165, "learning_rate": 9.299636497016451e-05, "loss": -0.9107, "step": 12720 }, { "epoch": 35.361111111111114, "grad_norm": 0.3582025468349457, "learning_rate": 9.298229246803076e-05, "loss": -0.9125, "step": 12730 }, { "epoch": 35.388888888888886, "grad_norm": 0.22731620073318481, "learning_rate": 9.296820690901144e-05, "loss": -0.9155, "step": 12740 }, { "epoch": 35.416666666666664, "grad_norm": 0.3347285985946655, "learning_rate": 9.295410829738539e-05, "loss": -0.9152, "step": 12750 }, { "epoch": 35.44444444444444, "grad_norm": 0.2805345058441162, "learning_rate": 9.293999663743535e-05, "loss": -0.9123, "step": 12760 }, { "epoch": 35.47222222222222, "grad_norm": 0.3335971534252167, "learning_rate": 9.292587193344813e-05, "loss": -0.9113, "step": 12770 }, { "epoch": 35.5, "grad_norm": 0.42344072461128235, "learning_rate": 9.291173418971437e-05, "loss": -0.9105, "step": 12780 }, { "epoch": 35.52777777777778, "grad_norm": 0.3776038885116577, "learning_rate": 9.28975834105288e-05, "loss": -0.9063, "step": 12790 }, { "epoch": 35.55555555555556, "grad_norm": 0.29630643129348755, "learning_rate": 9.288341960019004e-05, "loss": -0.9143, "step": 12800 }, { "epoch": 35.583333333333336, "grad_norm": 0.3131060302257538, "learning_rate": 9.286924276300067e-05, "loss": -0.908, "step": 12810 }, { "epoch": 35.611111111111114, "grad_norm": 0.2822323739528656, "learning_rate": 9.285505290326726e-05, "loss": -0.9145, "step": 12820 }, { "epoch": 35.638888888888886, "grad_norm": 0.3017115592956543, "learning_rate": 9.284085002530027e-05, "loss": -0.9156, "step": 12830 }, { "epoch": 35.666666666666664, "grad_norm": 0.2885758578777313, "learning_rate": 9.282663413341422e-05, "loss": -0.9125, "step": 12840 }, { "epoch": 35.69444444444444, "grad_norm": 0.3074117600917816, "learning_rate": 9.281240523192747e-05, "loss": -0.917, "step": 12850 }, { "epoch": 35.72222222222222, "grad_norm": 0.23985116183757782, "learning_rate": 9.279816332516242e-05, "loss": -0.9167, "step": 12860 }, { "epoch": 35.75, "grad_norm": 0.3508278727531433, "learning_rate": 9.278390841744536e-05, "loss": -0.9134, "step": 12870 }, { "epoch": 35.77777777777778, "grad_norm": 0.29545801877975464, "learning_rate": 9.276964051310658e-05, "loss": -0.9157, "step": 12880 }, { "epoch": 35.80555555555556, "grad_norm": 0.281697154045105, "learning_rate": 9.275535961648027e-05, "loss": -0.9191, "step": 12890 }, { "epoch": 35.833333333333336, "grad_norm": 0.552752673625946, "learning_rate": 9.274106573190459e-05, "loss": -0.9139, "step": 12900 }, { "epoch": 35.861111111111114, "grad_norm": 0.30558717250823975, "learning_rate": 9.272675886372168e-05, "loss": -0.9169, "step": 12910 }, { "epoch": 35.888888888888886, "grad_norm": 0.2612910270690918, "learning_rate": 9.271243901627754e-05, "loss": -0.9114, "step": 12920 }, { "epoch": 35.916666666666664, "grad_norm": 0.42055177688598633, "learning_rate": 9.269810619392219e-05, "loss": -0.9137, "step": 12930 }, { "epoch": 35.94444444444444, "grad_norm": 0.3807121515274048, "learning_rate": 9.268376040100955e-05, "loss": -0.9053, "step": 12940 }, { "epoch": 35.97222222222222, "grad_norm": 0.29230111837387085, "learning_rate": 9.266940164189752e-05, "loss": -0.9169, "step": 12950 }, { "epoch": 36.0, "grad_norm": 0.5006816387176514, "learning_rate": 9.265502992094787e-05, "loss": -0.9112, "step": 12960 }, { "epoch": 36.02777777777778, "grad_norm": 0.3721257150173187, "learning_rate": 9.264064524252638e-05, "loss": -0.9172, "step": 12970 }, { "epoch": 36.05555555555556, "grad_norm": 0.31159308552742004, "learning_rate": 9.262624761100271e-05, "loss": -0.9101, "step": 12980 }, { "epoch": 36.083333333333336, "grad_norm": 0.34760868549346924, "learning_rate": 9.261183703075051e-05, "loss": -0.9156, "step": 12990 }, { "epoch": 36.111111111111114, "grad_norm": 0.22744765877723694, "learning_rate": 9.259741350614733e-05, "loss": -0.9171, "step": 13000 }, { "epoch": 36.138888888888886, "grad_norm": 0.3395998179912567, "learning_rate": 9.258297704157464e-05, "loss": -0.9122, "step": 13010 }, { "epoch": 36.166666666666664, "grad_norm": 0.2896188199520111, "learning_rate": 9.256852764141786e-05, "loss": -0.913, "step": 13020 }, { "epoch": 36.19444444444444, "grad_norm": 0.32500797510147095, "learning_rate": 9.255406531006634e-05, "loss": -0.9151, "step": 13030 }, { "epoch": 36.22222222222222, "grad_norm": 0.2814231812953949, "learning_rate": 9.253959005191335e-05, "loss": -0.907, "step": 13040 }, { "epoch": 36.25, "grad_norm": 0.344762921333313, "learning_rate": 9.25251018713561e-05, "loss": -0.9105, "step": 13050 }, { "epoch": 36.27777777777778, "grad_norm": 0.35681483149528503, "learning_rate": 9.251060077279571e-05, "loss": -0.9159, "step": 13060 }, { "epoch": 36.30555555555556, "grad_norm": 0.23547464609146118, "learning_rate": 9.249608676063724e-05, "loss": -0.9137, "step": 13070 }, { "epoch": 36.333333333333336, "grad_norm": 0.25514477491378784, "learning_rate": 9.248155983928964e-05, "loss": -0.9141, "step": 13080 }, { "epoch": 36.361111111111114, "grad_norm": 0.3154398202896118, "learning_rate": 9.246702001316583e-05, "loss": -0.9142, "step": 13090 }, { "epoch": 36.388888888888886, "grad_norm": 0.2489330917596817, "learning_rate": 9.245246728668262e-05, "loss": -0.9159, "step": 13100 }, { "epoch": 36.416666666666664, "grad_norm": 0.37640380859375, "learning_rate": 9.243790166426073e-05, "loss": -0.9122, "step": 13110 }, { "epoch": 36.44444444444444, "grad_norm": 0.2665908932685852, "learning_rate": 9.242332315032484e-05, "loss": -0.9157, "step": 13120 }, { "epoch": 36.47222222222222, "grad_norm": 0.27385199069976807, "learning_rate": 9.240873174930349e-05, "loss": -0.9067, "step": 13130 }, { "epoch": 36.5, "grad_norm": 0.3004947006702423, "learning_rate": 9.239412746562917e-05, "loss": -0.9111, "step": 13140 }, { "epoch": 36.52777777777778, "grad_norm": 0.3792250156402588, "learning_rate": 9.237951030373828e-05, "loss": -0.8993, "step": 13150 }, { "epoch": 36.55555555555556, "grad_norm": 0.31671762466430664, "learning_rate": 9.236488026807113e-05, "loss": -0.9084, "step": 13160 }, { "epoch": 36.583333333333336, "grad_norm": 0.3766680657863617, "learning_rate": 9.235023736307193e-05, "loss": -0.9142, "step": 13170 }, { "epoch": 36.611111111111114, "grad_norm": 0.29552093148231506, "learning_rate": 9.233558159318881e-05, "loss": -0.9128, "step": 13180 }, { "epoch": 36.638888888888886, "grad_norm": 0.3701368272304535, "learning_rate": 9.232091296287382e-05, "loss": -0.9172, "step": 13190 }, { "epoch": 36.666666666666664, "grad_norm": 0.2898363769054413, "learning_rate": 9.230623147658288e-05, "loss": -0.9123, "step": 13200 }, { "epoch": 36.69444444444444, "grad_norm": 0.2787690758705139, "learning_rate": 9.229153713877586e-05, "loss": -0.9194, "step": 13210 }, { "epoch": 36.72222222222222, "grad_norm": 0.243441641330719, "learning_rate": 9.227682995391649e-05, "loss": -0.9115, "step": 13220 }, { "epoch": 36.75, "grad_norm": 0.44143593311309814, "learning_rate": 9.226210992647243e-05, "loss": -0.9148, "step": 13230 }, { "epoch": 36.77777777777778, "grad_norm": 0.36113840341567993, "learning_rate": 9.224737706091525e-05, "loss": -0.9193, "step": 13240 }, { "epoch": 36.80555555555556, "grad_norm": 0.20181851089000702, "learning_rate": 9.223263136172039e-05, "loss": -0.9121, "step": 13250 }, { "epoch": 36.833333333333336, "grad_norm": 0.2822332978248596, "learning_rate": 9.22178728333672e-05, "loss": -0.9029, "step": 13260 }, { "epoch": 36.861111111111114, "grad_norm": 0.4083636701107025, "learning_rate": 9.220310148033897e-05, "loss": -0.9126, "step": 13270 }, { "epoch": 36.888888888888886, "grad_norm": 0.4271197021007538, "learning_rate": 9.21883173071228e-05, "loss": -0.9112, "step": 13280 }, { "epoch": 36.916666666666664, "grad_norm": 0.31892839074134827, "learning_rate": 9.217352031820976e-05, "loss": -0.9195, "step": 13290 }, { "epoch": 36.94444444444444, "grad_norm": 0.3270905017852783, "learning_rate": 9.215871051809477e-05, "loss": -0.9143, "step": 13300 }, { "epoch": 36.97222222222222, "grad_norm": 0.2746763825416565, "learning_rate": 9.214388791127666e-05, "loss": -0.9123, "step": 13310 }, { "epoch": 37.0, "grad_norm": 0.22305230796337128, "learning_rate": 9.212905250225814e-05, "loss": -0.9159, "step": 13320 }, { "epoch": 37.02777777777778, "grad_norm": 0.47921088337898254, "learning_rate": 9.211420429554583e-05, "loss": -0.9149, "step": 13330 }, { "epoch": 37.05555555555556, "grad_norm": 0.36873605847358704, "learning_rate": 9.209934329565022e-05, "loss": -0.9177, "step": 13340 }, { "epoch": 37.083333333333336, "grad_norm": 0.32629430294036865, "learning_rate": 9.208446950708568e-05, "loss": -0.9095, "step": 13350 }, { "epoch": 37.111111111111114, "grad_norm": 0.34520453214645386, "learning_rate": 9.20695829343705e-05, "loss": -0.9151, "step": 13360 }, { "epoch": 37.138888888888886, "grad_norm": 0.31179845333099365, "learning_rate": 9.205468358202678e-05, "loss": -0.9204, "step": 13370 }, { "epoch": 37.166666666666664, "grad_norm": 0.33112478256225586, "learning_rate": 9.203977145458059e-05, "loss": -0.9115, "step": 13380 }, { "epoch": 37.19444444444444, "grad_norm": 0.2584744393825531, "learning_rate": 9.202484655656182e-05, "loss": -0.918, "step": 13390 }, { "epoch": 37.22222222222222, "grad_norm": 0.2694011330604553, "learning_rate": 9.200990889250427e-05, "loss": -0.9169, "step": 13400 }, { "epoch": 37.25, "grad_norm": 0.27248403429985046, "learning_rate": 9.19949584669456e-05, "loss": -0.9196, "step": 13410 }, { "epoch": 37.27777777777778, "grad_norm": 0.2970707416534424, "learning_rate": 9.197999528442738e-05, "loss": -0.9127, "step": 13420 }, { "epoch": 37.30555555555556, "grad_norm": 0.2102946788072586, "learning_rate": 9.196501934949499e-05, "loss": -0.9193, "step": 13430 }, { "epoch": 37.333333333333336, "grad_norm": 0.3895426392555237, "learning_rate": 9.195003066669776e-05, "loss": -0.917, "step": 13440 }, { "epoch": 37.361111111111114, "grad_norm": 0.35627204179763794, "learning_rate": 9.193502924058884e-05, "loss": -0.9205, "step": 13450 }, { "epoch": 37.388888888888886, "grad_norm": 0.33768659830093384, "learning_rate": 9.192001507572526e-05, "loss": -0.9129, "step": 13460 }, { "epoch": 37.416666666666664, "grad_norm": 0.33109596371650696, "learning_rate": 9.190498817666793e-05, "loss": -0.9088, "step": 13470 }, { "epoch": 37.44444444444444, "grad_norm": 0.2762235701084137, "learning_rate": 9.188994854798163e-05, "loss": -0.911, "step": 13480 }, { "epoch": 37.47222222222222, "grad_norm": 0.21904484927654266, "learning_rate": 9.187489619423499e-05, "loss": -0.9143, "step": 13490 }, { "epoch": 37.5, "grad_norm": 0.2598377466201782, "learning_rate": 9.185983112000056e-05, "loss": -0.9172, "step": 13500 }, { "epoch": 37.52777777777778, "grad_norm": 0.3395910859107971, "learning_rate": 9.184475332985464e-05, "loss": -0.9179, "step": 13510 }, { "epoch": 37.55555555555556, "grad_norm": 0.31327518820762634, "learning_rate": 9.182966282837754e-05, "loss": -0.9158, "step": 13520 }, { "epoch": 37.583333333333336, "grad_norm": 0.3508719801902771, "learning_rate": 9.18145596201533e-05, "loss": -0.9217, "step": 13530 }, { "epoch": 37.611111111111114, "grad_norm": 0.3016316294670105, "learning_rate": 9.179944370976991e-05, "loss": -0.917, "step": 13540 }, { "epoch": 37.638888888888886, "grad_norm": 0.30830511450767517, "learning_rate": 9.178431510181918e-05, "loss": -0.9193, "step": 13550 }, { "epoch": 37.666666666666664, "grad_norm": 0.3602955639362335, "learning_rate": 9.176917380089675e-05, "loss": -0.9157, "step": 13560 }, { "epoch": 37.69444444444444, "grad_norm": 0.4985174834728241, "learning_rate": 9.175401981160219e-05, "loss": -0.9117, "step": 13570 }, { "epoch": 37.72222222222222, "grad_norm": 0.28938332200050354, "learning_rate": 9.173885313853885e-05, "loss": -0.9183, "step": 13580 }, { "epoch": 37.75, "grad_norm": 0.41485828161239624, "learning_rate": 9.172367378631398e-05, "loss": -0.911, "step": 13590 }, { "epoch": 37.77777777777778, "grad_norm": 0.2991109788417816, "learning_rate": 9.170848175953866e-05, "loss": -0.9087, "step": 13600 }, { "epoch": 37.80555555555556, "grad_norm": 0.25728997588157654, "learning_rate": 9.169327706282784e-05, "loss": -0.9213, "step": 13610 }, { "epoch": 37.833333333333336, "grad_norm": 0.30892840027809143, "learning_rate": 9.167805970080029e-05, "loss": -0.9188, "step": 13620 }, { "epoch": 37.861111111111114, "grad_norm": 0.34693580865859985, "learning_rate": 9.166282967807864e-05, "loss": -0.9154, "step": 13630 }, { "epoch": 37.888888888888886, "grad_norm": 0.32414403557777405, "learning_rate": 9.16475869992894e-05, "loss": -0.9231, "step": 13640 }, { "epoch": 37.916666666666664, "grad_norm": 0.27460652589797974, "learning_rate": 9.163233166906284e-05, "loss": -0.9237, "step": 13650 }, { "epoch": 37.94444444444444, "grad_norm": 0.3205818235874176, "learning_rate": 9.161706369203317e-05, "loss": -0.9167, "step": 13660 }, { "epoch": 37.97222222222222, "grad_norm": 0.2023601233959198, "learning_rate": 9.16017830728384e-05, "loss": -0.9237, "step": 13670 }, { "epoch": 38.0, "grad_norm": 0.2665761709213257, "learning_rate": 9.158648981612035e-05, "loss": -0.9105, "step": 13680 }, { "epoch": 38.02777777777778, "grad_norm": 0.26181215047836304, "learning_rate": 9.157118392652472e-05, "loss": -0.9195, "step": 13690 }, { "epoch": 38.05555555555556, "grad_norm": 0.21442709863185883, "learning_rate": 9.155586540870104e-05, "loss": -0.9219, "step": 13700 }, { "epoch": 38.083333333333336, "grad_norm": 0.26013875007629395, "learning_rate": 9.154053426730267e-05, "loss": -0.9175, "step": 13710 }, { "epoch": 38.111111111111114, "grad_norm": 0.4448208212852478, "learning_rate": 9.15251905069868e-05, "loss": -0.9125, "step": 13720 }, { "epoch": 38.138888888888886, "grad_norm": 0.30889636278152466, "learning_rate": 9.150983413241446e-05, "loss": -0.9147, "step": 13730 }, { "epoch": 38.166666666666664, "grad_norm": 0.5222804546356201, "learning_rate": 9.149446514825051e-05, "loss": -0.9067, "step": 13740 }, { "epoch": 38.19444444444444, "grad_norm": 0.45922064781188965, "learning_rate": 9.147908355916365e-05, "loss": -0.907, "step": 13750 }, { "epoch": 38.22222222222222, "grad_norm": 0.4861834943294525, "learning_rate": 9.146368936982642e-05, "loss": -0.9076, "step": 13760 }, { "epoch": 38.25, "grad_norm": 0.21716146171092987, "learning_rate": 9.144828258491511e-05, "loss": -0.9127, "step": 13770 }, { "epoch": 38.27777777777778, "grad_norm": 0.21688804030418396, "learning_rate": 9.143286320910996e-05, "loss": -0.9136, "step": 13780 }, { "epoch": 38.30555555555556, "grad_norm": 0.22857673466205597, "learning_rate": 9.141743124709491e-05, "loss": -0.9185, "step": 13790 }, { "epoch": 38.333333333333336, "grad_norm": 0.3477899730205536, "learning_rate": 9.140198670355784e-05, "loss": -0.9178, "step": 13800 }, { "epoch": 38.361111111111114, "grad_norm": 0.33231163024902344, "learning_rate": 9.138652958319034e-05, "loss": -0.9088, "step": 13810 }, { "epoch": 38.388888888888886, "grad_norm": 0.3473755717277527, "learning_rate": 9.137105989068791e-05, "loss": -0.9174, "step": 13820 }, { "epoch": 38.416666666666664, "grad_norm": 0.346589595079422, "learning_rate": 9.135557763074983e-05, "loss": -0.903, "step": 13830 }, { "epoch": 38.44444444444444, "grad_norm": 0.27839672565460205, "learning_rate": 9.13400828080792e-05, "loss": -0.9112, "step": 13840 }, { "epoch": 38.47222222222222, "grad_norm": 0.352924644947052, "learning_rate": 9.132457542738292e-05, "loss": -0.9141, "step": 13850 }, { "epoch": 38.5, "grad_norm": 0.24050435423851013, "learning_rate": 9.130905549337174e-05, "loss": -0.9224, "step": 13860 }, { "epoch": 38.52777777777778, "grad_norm": 0.42993834614753723, "learning_rate": 9.129352301076021e-05, "loss": -0.9181, "step": 13870 }, { "epoch": 38.55555555555556, "grad_norm": 0.2315978705883026, "learning_rate": 9.127797798426668e-05, "loss": -0.9193, "step": 13880 }, { "epoch": 38.583333333333336, "grad_norm": 0.3617701530456543, "learning_rate": 9.126242041861333e-05, "loss": -0.9126, "step": 13890 }, { "epoch": 38.611111111111114, "grad_norm": 0.39275816082954407, "learning_rate": 9.124685031852611e-05, "loss": -0.9129, "step": 13900 }, { "epoch": 38.638888888888886, "grad_norm": 0.3902490437030792, "learning_rate": 9.123126768873482e-05, "loss": -0.9157, "step": 13910 }, { "epoch": 38.666666666666664, "grad_norm": 0.5153685212135315, "learning_rate": 9.121567253397308e-05, "loss": -0.9165, "step": 13920 }, { "epoch": 38.69444444444444, "grad_norm": 0.34036633372306824, "learning_rate": 9.120006485897824e-05, "loss": -0.9212, "step": 13930 }, { "epoch": 38.72222222222222, "grad_norm": 0.3687053918838501, "learning_rate": 9.118444466849152e-05, "loss": -0.9157, "step": 13940 }, { "epoch": 38.75, "grad_norm": 0.36928069591522217, "learning_rate": 9.116881196725793e-05, "loss": -0.9199, "step": 13950 }, { "epoch": 38.77777777777778, "grad_norm": 0.28883838653564453, "learning_rate": 9.115316676002627e-05, "loss": -0.9189, "step": 13960 }, { "epoch": 38.80555555555556, "grad_norm": 0.30303868651390076, "learning_rate": 9.113750905154911e-05, "loss": -0.9164, "step": 13970 }, { "epoch": 38.833333333333336, "grad_norm": 0.2636045515537262, "learning_rate": 9.112183884658289e-05, "loss": -0.9135, "step": 13980 }, { "epoch": 38.861111111111114, "grad_norm": 0.27258044481277466, "learning_rate": 9.11061561498878e-05, "loss": -0.9227, "step": 13990 }, { "epoch": 38.888888888888886, "grad_norm": 0.3408144414424896, "learning_rate": 9.109046096622779e-05, "loss": -0.9172, "step": 14000 }, { "epoch": 38.916666666666664, "grad_norm": 0.2873440682888031, "learning_rate": 9.107475330037069e-05, "loss": -0.9207, "step": 14010 }, { "epoch": 38.94444444444444, "grad_norm": 0.27279335260391235, "learning_rate": 9.105903315708806e-05, "loss": -0.9166, "step": 14020 }, { "epoch": 38.97222222222222, "grad_norm": 0.3133898377418518, "learning_rate": 9.104330054115524e-05, "loss": -0.9228, "step": 14030 }, { "epoch": 39.0, "grad_norm": 0.2157442271709442, "learning_rate": 9.102755545735141e-05, "loss": -0.9219, "step": 14040 }, { "epoch": 39.02777777777778, "grad_norm": 0.24964667856693268, "learning_rate": 9.10117979104595e-05, "loss": -0.9181, "step": 14050 }, { "epoch": 39.05555555555556, "grad_norm": 0.25160786509513855, "learning_rate": 9.099602790526624e-05, "loss": -0.919, "step": 14060 }, { "epoch": 39.083333333333336, "grad_norm": 0.2534812390804291, "learning_rate": 9.098024544656212e-05, "loss": -0.922, "step": 14070 }, { "epoch": 39.111111111111114, "grad_norm": 0.2858244776725769, "learning_rate": 9.096445053914148e-05, "loss": -0.9123, "step": 14080 }, { "epoch": 39.138888888888886, "grad_norm": 0.39347168803215027, "learning_rate": 9.094864318780236e-05, "loss": -0.916, "step": 14090 }, { "epoch": 39.166666666666664, "grad_norm": 0.3226303160190582, "learning_rate": 9.093282339734663e-05, "loss": -0.905, "step": 14100 }, { "epoch": 39.19444444444444, "grad_norm": 0.3391883373260498, "learning_rate": 9.091699117257992e-05, "loss": -0.917, "step": 14110 }, { "epoch": 39.22222222222222, "grad_norm": 0.2501485049724579, "learning_rate": 9.090114651831163e-05, "loss": -0.9205, "step": 14120 }, { "epoch": 39.25, "grad_norm": 0.3847505450248718, "learning_rate": 9.088528943935497e-05, "loss": -0.9101, "step": 14130 }, { "epoch": 39.27777777777778, "grad_norm": 0.24896836280822754, "learning_rate": 9.086941994052689e-05, "loss": -0.9189, "step": 14140 }, { "epoch": 39.30555555555556, "grad_norm": 0.3567883372306824, "learning_rate": 9.085353802664813e-05, "loss": -0.9239, "step": 14150 }, { "epoch": 39.333333333333336, "grad_norm": 0.29460224509239197, "learning_rate": 9.08376437025432e-05, "loss": -0.9201, "step": 14160 }, { "epoch": 39.361111111111114, "grad_norm": 0.19481511414051056, "learning_rate": 9.082173697304035e-05, "loss": -0.9171, "step": 14170 }, { "epoch": 39.388888888888886, "grad_norm": 0.25159093737602234, "learning_rate": 9.080581784297166e-05, "loss": -0.9243, "step": 14180 }, { "epoch": 39.416666666666664, "grad_norm": 0.32387393712997437, "learning_rate": 9.078988631717291e-05, "loss": -0.9184, "step": 14190 }, { "epoch": 39.44444444444444, "grad_norm": 0.24880164861679077, "learning_rate": 9.077394240048369e-05, "loss": -0.9176, "step": 14200 }, { "epoch": 39.47222222222222, "grad_norm": 0.3293898105621338, "learning_rate": 9.075798609774736e-05, "loss": -0.9134, "step": 14210 }, { "epoch": 39.5, "grad_norm": 0.2728685736656189, "learning_rate": 9.0742017413811e-05, "loss": -0.9204, "step": 14220 }, { "epoch": 39.52777777777778, "grad_norm": 0.36454322934150696, "learning_rate": 9.072603635352548e-05, "loss": -0.9177, "step": 14230 }, { "epoch": 39.55555555555556, "grad_norm": 0.31262362003326416, "learning_rate": 9.071004292174541e-05, "loss": -0.9169, "step": 14240 }, { "epoch": 39.583333333333336, "grad_norm": 0.25124114751815796, "learning_rate": 9.06940371233292e-05, "loss": -0.9236, "step": 14250 }, { "epoch": 39.611111111111114, "grad_norm": 0.25343894958496094, "learning_rate": 9.067801896313898e-05, "loss": -0.9185, "step": 14260 }, { "epoch": 39.638888888888886, "grad_norm": 0.21212303638458252, "learning_rate": 9.066198844604064e-05, "loss": -0.9179, "step": 14270 }, { "epoch": 39.666666666666664, "grad_norm": 0.2741559147834778, "learning_rate": 9.06459455769038e-05, "loss": -0.9191, "step": 14280 }, { "epoch": 39.69444444444444, "grad_norm": 0.2947413921356201, "learning_rate": 9.062989036060193e-05, "loss": -0.9092, "step": 14290 }, { "epoch": 39.72222222222222, "grad_norm": 0.3810998797416687, "learning_rate": 9.061382280201212e-05, "loss": -0.9205, "step": 14300 }, { "epoch": 39.75, "grad_norm": 0.3429548144340515, "learning_rate": 9.059774290601528e-05, "loss": -0.9104, "step": 14310 }, { "epoch": 39.77777777777778, "grad_norm": 0.28068554401397705, "learning_rate": 9.058165067749606e-05, "loss": -0.9182, "step": 14320 }, { "epoch": 39.80555555555556, "grad_norm": 0.3967474699020386, "learning_rate": 9.056554612134288e-05, "loss": -0.9156, "step": 14330 }, { "epoch": 39.833333333333336, "grad_norm": 0.1992427557706833, "learning_rate": 9.054942924244785e-05, "loss": -0.9178, "step": 14340 }, { "epoch": 39.861111111111114, "grad_norm": 0.2482256442308426, "learning_rate": 9.053330004570686e-05, "loss": -0.9164, "step": 14350 }, { "epoch": 39.888888888888886, "grad_norm": 0.17783796787261963, "learning_rate": 9.051715853601955e-05, "loss": -0.918, "step": 14360 }, { "epoch": 39.916666666666664, "grad_norm": 0.3083334267139435, "learning_rate": 9.050100471828926e-05, "loss": -0.9065, "step": 14370 }, { "epoch": 39.94444444444444, "grad_norm": 0.29910531640052795, "learning_rate": 9.048483859742311e-05, "loss": -0.9128, "step": 14380 }, { "epoch": 39.97222222222222, "grad_norm": 0.4085061550140381, "learning_rate": 9.046866017833193e-05, "loss": -0.9186, "step": 14390 }, { "epoch": 40.0, "grad_norm": 0.2564908266067505, "learning_rate": 9.045246946593029e-05, "loss": -0.9173, "step": 14400 }, { "epoch": 40.02777777777778, "grad_norm": 0.3437108099460602, "learning_rate": 9.043626646513652e-05, "loss": -0.9213, "step": 14410 }, { "epoch": 40.05555555555556, "grad_norm": 0.2444336712360382, "learning_rate": 9.042005118087267e-05, "loss": -0.9254, "step": 14420 }, { "epoch": 40.083333333333336, "grad_norm": 0.30051976442337036, "learning_rate": 9.040382361806448e-05, "loss": -0.916, "step": 14430 }, { "epoch": 40.111111111111114, "grad_norm": 0.355300635099411, "learning_rate": 9.038758378164148e-05, "loss": -0.9159, "step": 14440 }, { "epoch": 40.138888888888886, "grad_norm": 0.2756991982460022, "learning_rate": 9.037133167653691e-05, "loss": -0.9187, "step": 14450 }, { "epoch": 40.166666666666664, "grad_norm": 0.4379718601703644, "learning_rate": 9.035506730768771e-05, "loss": -0.9174, "step": 14460 }, { "epoch": 40.19444444444444, "grad_norm": 0.4012654721736908, "learning_rate": 9.033879068003458e-05, "loss": -0.9194, "step": 14470 }, { "epoch": 40.22222222222222, "grad_norm": 0.3397239148616791, "learning_rate": 9.032250179852193e-05, "loss": -0.9155, "step": 14480 }, { "epoch": 40.25, "grad_norm": 0.25133833289146423, "learning_rate": 9.030620066809787e-05, "loss": -0.9208, "step": 14490 }, { "epoch": 40.27777777777778, "grad_norm": 0.4465566575527191, "learning_rate": 9.028988729371428e-05, "loss": -0.9195, "step": 14500 }, { "epoch": 40.30555555555556, "grad_norm": 0.41300514340400696, "learning_rate": 9.027356168032673e-05, "loss": -0.915, "step": 14510 }, { "epoch": 40.333333333333336, "grad_norm": 0.5027793049812317, "learning_rate": 9.02572238328945e-05, "loss": -0.9148, "step": 14520 }, { "epoch": 40.361111111111114, "grad_norm": 0.38567468523979187, "learning_rate": 9.02408737563806e-05, "loss": -0.919, "step": 14530 }, { "epoch": 40.388888888888886, "grad_norm": 0.42711448669433594, "learning_rate": 9.022451145575174e-05, "loss": -0.9175, "step": 14540 }, { "epoch": 40.416666666666664, "grad_norm": 0.2613467872142792, "learning_rate": 9.02081369359784e-05, "loss": -0.9215, "step": 14550 }, { "epoch": 40.44444444444444, "grad_norm": 0.1807118058204651, "learning_rate": 9.019175020203465e-05, "loss": -0.9224, "step": 14560 }, { "epoch": 40.47222222222222, "grad_norm": 0.31785455346107483, "learning_rate": 9.017535125889842e-05, "loss": -0.9218, "step": 14570 }, { "epoch": 40.5, "grad_norm": 0.22281183302402496, "learning_rate": 9.015894011155124e-05, "loss": -0.9163, "step": 14580 }, { "epoch": 40.52777777777778, "grad_norm": 0.3236309289932251, "learning_rate": 9.014251676497838e-05, "loss": -0.9109, "step": 14590 }, { "epoch": 40.55555555555556, "grad_norm": 0.34454628825187683, "learning_rate": 9.012608122416884e-05, "loss": -0.9197, "step": 14600 }, { "epoch": 40.583333333333336, "grad_norm": 0.24976104497909546, "learning_rate": 9.010963349411529e-05, "loss": -0.917, "step": 14610 }, { "epoch": 40.611111111111114, "grad_norm": 0.2641493082046509, "learning_rate": 9.00931735798141e-05, "loss": -0.9176, "step": 14620 }, { "epoch": 40.638888888888886, "grad_norm": 0.308040589094162, "learning_rate": 9.00767014862654e-05, "loss": -0.9183, "step": 14630 }, { "epoch": 40.666666666666664, "grad_norm": 0.2532641589641571, "learning_rate": 9.006021721847295e-05, "loss": -0.9226, "step": 14640 }, { "epoch": 40.69444444444444, "grad_norm": 0.3536502718925476, "learning_rate": 9.004372078144423e-05, "loss": -0.917, "step": 14650 }, { "epoch": 40.72222222222222, "grad_norm": 0.44871488213539124, "learning_rate": 9.002721218019043e-05, "loss": -0.9134, "step": 14660 }, { "epoch": 40.75, "grad_norm": 0.33136317133903503, "learning_rate": 9.001069141972642e-05, "loss": -0.9226, "step": 14670 }, { "epoch": 40.77777777777778, "grad_norm": 0.45473024249076843, "learning_rate": 8.99941585050708e-05, "loss": -0.9189, "step": 14680 }, { "epoch": 40.80555555555556, "grad_norm": 0.3726162016391754, "learning_rate": 8.997761344124578e-05, "loss": -0.919, "step": 14690 }, { "epoch": 40.833333333333336, "grad_norm": 0.27905112504959106, "learning_rate": 8.996105623327737e-05, "loss": -0.9217, "step": 14700 }, { "epoch": 40.861111111111114, "grad_norm": 0.3328411877155304, "learning_rate": 8.994448688619517e-05, "loss": -0.9222, "step": 14710 }, { "epoch": 40.888888888888886, "grad_norm": 0.30852562189102173, "learning_rate": 8.992790540503253e-05, "loss": -0.9186, "step": 14720 }, { "epoch": 40.916666666666664, "grad_norm": 0.2350669801235199, "learning_rate": 8.991131179482648e-05, "loss": -0.9216, "step": 14730 }, { "epoch": 40.94444444444444, "grad_norm": 0.30569636821746826, "learning_rate": 8.989470606061768e-05, "loss": -0.9146, "step": 14740 }, { "epoch": 40.97222222222222, "grad_norm": 0.4504517614841461, "learning_rate": 8.987808820745056e-05, "loss": -0.9118, "step": 14750 }, { "epoch": 41.0, "grad_norm": 0.3828029930591583, "learning_rate": 8.986145824037315e-05, "loss": -0.9228, "step": 14760 }, { "epoch": 41.02777777777778, "grad_norm": 0.3807852566242218, "learning_rate": 8.984481616443721e-05, "loss": -0.9119, "step": 14770 }, { "epoch": 41.05555555555556, "grad_norm": 0.25612837076187134, "learning_rate": 8.982816198469815e-05, "loss": -0.9231, "step": 14780 }, { "epoch": 41.083333333333336, "grad_norm": 0.3188915550708771, "learning_rate": 8.98114957062151e-05, "loss": -0.9178, "step": 14790 }, { "epoch": 41.111111111111114, "grad_norm": 0.33469846844673157, "learning_rate": 8.97948173340508e-05, "loss": -0.9237, "step": 14800 }, { "epoch": 41.138888888888886, "grad_norm": 0.33856654167175293, "learning_rate": 8.977812687327172e-05, "loss": -0.9249, "step": 14810 }, { "epoch": 41.166666666666664, "grad_norm": 0.2395068258047104, "learning_rate": 8.976142432894798e-05, "loss": -0.9205, "step": 14820 }, { "epoch": 41.19444444444444, "grad_norm": 0.2769401967525482, "learning_rate": 8.974470970615336e-05, "loss": -0.915, "step": 14830 }, { "epoch": 41.22222222222222, "grad_norm": 0.32891449332237244, "learning_rate": 8.972798300996534e-05, "loss": -0.9163, "step": 14840 }, { "epoch": 41.25, "grad_norm": 0.40988174080848694, "learning_rate": 8.971124424546504e-05, "loss": -0.9208, "step": 14850 }, { "epoch": 41.27777777777778, "grad_norm": 0.2923906743526459, "learning_rate": 8.969449341773724e-05, "loss": -0.9154, "step": 14860 }, { "epoch": 41.30555555555556, "grad_norm": 0.3429287075996399, "learning_rate": 8.967773053187042e-05, "loss": -0.9143, "step": 14870 }, { "epoch": 41.333333333333336, "grad_norm": 0.25415289402008057, "learning_rate": 8.966095559295668e-05, "loss": -0.9202, "step": 14880 }, { "epoch": 41.361111111111114, "grad_norm": 0.4272677004337311, "learning_rate": 8.964416860609184e-05, "loss": -0.9171, "step": 14890 }, { "epoch": 41.388888888888886, "grad_norm": 0.35558709502220154, "learning_rate": 8.962736957637532e-05, "loss": -0.9246, "step": 14900 }, { "epoch": 41.416666666666664, "grad_norm": 0.23569965362548828, "learning_rate": 8.96105585089102e-05, "loss": -0.9248, "step": 14910 }, { "epoch": 41.44444444444444, "grad_norm": 0.37532398104667664, "learning_rate": 8.959373540880329e-05, "loss": -0.9079, "step": 14920 }, { "epoch": 41.47222222222222, "grad_norm": 0.35284698009490967, "learning_rate": 8.957690028116495e-05, "loss": -0.9151, "step": 14930 }, { "epoch": 41.5, "grad_norm": 0.5446978211402893, "learning_rate": 8.956005313110928e-05, "loss": -0.9179, "step": 14940 }, { "epoch": 41.52777777777778, "grad_norm": 0.3724380433559418, "learning_rate": 8.9543193963754e-05, "loss": -0.9118, "step": 14950 }, { "epoch": 41.55555555555556, "grad_norm": 0.2818167805671692, "learning_rate": 8.952632278422048e-05, "loss": -0.919, "step": 14960 }, { "epoch": 41.583333333333336, "grad_norm": 0.39863958954811096, "learning_rate": 8.95094395976337e-05, "loss": -0.9137, "step": 14970 }, { "epoch": 41.611111111111114, "grad_norm": 0.3365940749645233, "learning_rate": 8.949254440912239e-05, "loss": -0.9168, "step": 14980 }, { "epoch": 41.638888888888886, "grad_norm": 0.2441442310810089, "learning_rate": 8.94756372238188e-05, "loss": -0.9193, "step": 14990 }, { "epoch": 41.666666666666664, "grad_norm": 0.23682302236557007, "learning_rate": 8.945871804685892e-05, "loss": -0.9215, "step": 15000 }, { "epoch": 41.69444444444444, "grad_norm": 0.2167905569076538, "learning_rate": 8.944178688338236e-05, "loss": -0.9203, "step": 15010 }, { "epoch": 41.72222222222222, "grad_norm": 0.27140095829963684, "learning_rate": 8.942484373853233e-05, "loss": -0.9256, "step": 15020 }, { "epoch": 41.75, "grad_norm": 0.33911487460136414, "learning_rate": 8.940788861745572e-05, "loss": -0.9133, "step": 15030 }, { "epoch": 41.77777777777778, "grad_norm": 0.3487112522125244, "learning_rate": 8.939092152530308e-05, "loss": -0.9167, "step": 15040 }, { "epoch": 41.80555555555556, "grad_norm": 0.5015698671340942, "learning_rate": 8.937394246722853e-05, "loss": -0.9212, "step": 15050 }, { "epoch": 41.833333333333336, "grad_norm": 0.3976120352745056, "learning_rate": 8.935695144838984e-05, "loss": -0.9112, "step": 15060 }, { "epoch": 41.861111111111114, "grad_norm": 0.31186091899871826, "learning_rate": 8.933994847394849e-05, "loss": -0.92, "step": 15070 }, { "epoch": 41.888888888888886, "grad_norm": 0.36908867955207825, "learning_rate": 8.932293354906949e-05, "loss": -0.9229, "step": 15080 }, { "epoch": 41.916666666666664, "grad_norm": 0.28322359919548035, "learning_rate": 8.930590667892153e-05, "loss": -0.9207, "step": 15090 }, { "epoch": 41.94444444444444, "grad_norm": 0.34409669041633606, "learning_rate": 8.928886786867696e-05, "loss": -0.9238, "step": 15100 }, { "epoch": 41.97222222222222, "grad_norm": 0.222192645072937, "learning_rate": 8.927181712351168e-05, "loss": -0.9243, "step": 15110 }, { "epoch": 42.0, "grad_norm": 0.3233698904514313, "learning_rate": 8.925475444860527e-05, "loss": -0.921, "step": 15120 }, { "epoch": 42.02777777777778, "grad_norm": 0.24595172703266144, "learning_rate": 8.923767984914092e-05, "loss": -0.9274, "step": 15130 }, { "epoch": 42.05555555555556, "grad_norm": 0.35354021191596985, "learning_rate": 8.922059333030545e-05, "loss": -0.9139, "step": 15140 }, { "epoch": 42.083333333333336, "grad_norm": 0.2910620868206024, "learning_rate": 8.920349489728928e-05, "loss": -0.9222, "step": 15150 }, { "epoch": 42.111111111111114, "grad_norm": 0.32908013463020325, "learning_rate": 8.918638455528646e-05, "loss": -0.9121, "step": 15160 }, { "epoch": 42.138888888888886, "grad_norm": 0.3573943078517914, "learning_rate": 8.916926230949468e-05, "loss": -0.9179, "step": 15170 }, { "epoch": 42.166666666666664, "grad_norm": 0.22431392967700958, "learning_rate": 8.915212816511522e-05, "loss": -0.9199, "step": 15180 }, { "epoch": 42.19444444444444, "grad_norm": 0.41228538751602173, "learning_rate": 8.913498212735296e-05, "loss": -0.9192, "step": 15190 }, { "epoch": 42.22222222222222, "grad_norm": 0.3062039017677307, "learning_rate": 8.911782420141643e-05, "loss": -0.9178, "step": 15200 }, { "epoch": 42.25, "grad_norm": 0.38851433992385864, "learning_rate": 8.910065439251775e-05, "loss": -0.9228, "step": 15210 }, { "epoch": 42.27777777777778, "grad_norm": 0.36205971240997314, "learning_rate": 8.908347270587268e-05, "loss": -0.9212, "step": 15220 }, { "epoch": 42.30555555555556, "grad_norm": 0.3936983346939087, "learning_rate": 8.906627914670054e-05, "loss": -0.9207, "step": 15230 }, { "epoch": 42.333333333333336, "grad_norm": 0.23957474529743195, "learning_rate": 8.904907372022427e-05, "loss": -0.9226, "step": 15240 }, { "epoch": 42.361111111111114, "grad_norm": 0.3512205481529236, "learning_rate": 8.903185643167042e-05, "loss": -0.919, "step": 15250 }, { "epoch": 42.388888888888886, "grad_norm": 0.2382482886314392, "learning_rate": 8.901462728626919e-05, "loss": -0.9253, "step": 15260 }, { "epoch": 42.416666666666664, "grad_norm": 0.2658940255641937, "learning_rate": 8.899738628925429e-05, "loss": -0.9175, "step": 15270 }, { "epoch": 42.44444444444444, "grad_norm": 0.3209633529186249, "learning_rate": 8.898013344586312e-05, "loss": -0.9193, "step": 15280 }, { "epoch": 42.47222222222222, "grad_norm": 0.28249746561050415, "learning_rate": 8.896286876133661e-05, "loss": -0.9114, "step": 15290 }, { "epoch": 42.5, "grad_norm": 0.328052818775177, "learning_rate": 8.894559224091933e-05, "loss": -0.9195, "step": 15300 }, { "epoch": 42.52777777777778, "grad_norm": 0.28888511657714844, "learning_rate": 8.892830388985942e-05, "loss": -0.9169, "step": 15310 }, { "epoch": 42.55555555555556, "grad_norm": 0.2635376453399658, "learning_rate": 8.891100371340864e-05, "loss": -0.9219, "step": 15320 }, { "epoch": 42.583333333333336, "grad_norm": 0.28054186701774597, "learning_rate": 8.889369171682231e-05, "loss": -0.9178, "step": 15330 }, { "epoch": 42.611111111111114, "grad_norm": 0.31668955087661743, "learning_rate": 8.887636790535936e-05, "loss": -0.9223, "step": 15340 }, { "epoch": 42.638888888888886, "grad_norm": 0.21494342386722565, "learning_rate": 8.885903228428231e-05, "loss": -0.9199, "step": 15350 }, { "epoch": 42.666666666666664, "grad_norm": 0.3185409605503082, "learning_rate": 8.884168485885727e-05, "loss": -0.9251, "step": 15360 }, { "epoch": 42.69444444444444, "grad_norm": 0.29231905937194824, "learning_rate": 8.882432563435393e-05, "loss": -0.9182, "step": 15370 }, { "epoch": 42.72222222222222, "grad_norm": 0.28172439336776733, "learning_rate": 8.880695461604556e-05, "loss": -0.9239, "step": 15380 }, { "epoch": 42.75, "grad_norm": 0.4450286030769348, "learning_rate": 8.878957180920901e-05, "loss": -0.9203, "step": 15390 }, { "epoch": 42.77777777777778, "grad_norm": 0.3442974388599396, "learning_rate": 8.877217721912473e-05, "loss": -0.9191, "step": 15400 }, { "epoch": 42.80555555555556, "grad_norm": 0.28066501021385193, "learning_rate": 8.875477085107673e-05, "loss": -0.918, "step": 15410 }, { "epoch": 42.833333333333336, "grad_norm": 0.3729308247566223, "learning_rate": 8.87373527103526e-05, "loss": -0.9165, "step": 15420 }, { "epoch": 42.861111111111114, "grad_norm": 0.38472118973731995, "learning_rate": 8.871992280224353e-05, "loss": -0.9189, "step": 15430 }, { "epoch": 42.888888888888886, "grad_norm": 0.39201316237449646, "learning_rate": 8.870248113204422e-05, "loss": -0.9231, "step": 15440 }, { "epoch": 42.916666666666664, "grad_norm": 0.3122940957546234, "learning_rate": 8.868502770505306e-05, "loss": -0.9168, "step": 15450 }, { "epoch": 42.94444444444444, "grad_norm": 0.2761979103088379, "learning_rate": 8.86675625265719e-05, "loss": -0.9156, "step": 15460 }, { "epoch": 42.97222222222222, "grad_norm": 0.29190871119499207, "learning_rate": 8.865008560190618e-05, "loss": -0.9243, "step": 15470 }, { "epoch": 43.0, "grad_norm": 0.24968868494033813, "learning_rate": 8.863259693636496e-05, "loss": -0.9207, "step": 15480 }, { "epoch": 43.02777777777778, "grad_norm": 0.2506166398525238, "learning_rate": 8.861509653526083e-05, "loss": -0.9232, "step": 15490 }, { "epoch": 43.05555555555556, "grad_norm": 0.3133154511451721, "learning_rate": 8.859758440390993e-05, "loss": -0.9188, "step": 15500 }, { "epoch": 43.083333333333336, "grad_norm": 0.2780523896217346, "learning_rate": 8.858006054763202e-05, "loss": -0.9255, "step": 15510 }, { "epoch": 43.111111111111114, "grad_norm": 0.30468878149986267, "learning_rate": 8.856252497175035e-05, "loss": -0.9225, "step": 15520 }, { "epoch": 43.138888888888886, "grad_norm": 0.27194806933403015, "learning_rate": 8.854497768159178e-05, "loss": -0.9246, "step": 15530 }, { "epoch": 43.166666666666664, "grad_norm": 0.291984885931015, "learning_rate": 8.852741868248671e-05, "loss": -0.9246, "step": 15540 }, { "epoch": 43.19444444444444, "grad_norm": 0.23698972165584564, "learning_rate": 8.85098479797691e-05, "loss": -0.9181, "step": 15550 }, { "epoch": 43.22222222222222, "grad_norm": 0.47283196449279785, "learning_rate": 8.849226557877646e-05, "loss": -0.9212, "step": 15560 }, { "epoch": 43.25, "grad_norm": 0.36166125535964966, "learning_rate": 8.84746714848499e-05, "loss": -0.9176, "step": 15570 }, { "epoch": 43.27777777777778, "grad_norm": 0.18928103148937225, "learning_rate": 8.845706570333397e-05, "loss": -0.9244, "step": 15580 }, { "epoch": 43.30555555555556, "grad_norm": 0.21484871208667755, "learning_rate": 8.84394482395769e-05, "loss": -0.9177, "step": 15590 }, { "epoch": 43.333333333333336, "grad_norm": 0.18512262403964996, "learning_rate": 8.842181909893038e-05, "loss": -0.9272, "step": 15600 }, { "epoch": 43.361111111111114, "grad_norm": 0.2338266223669052, "learning_rate": 8.840417828674969e-05, "loss": -0.9256, "step": 15610 }, { "epoch": 43.388888888888886, "grad_norm": 0.3270546495914459, "learning_rate": 8.838652580839364e-05, "loss": -0.9175, "step": 15620 }, { "epoch": 43.416666666666664, "grad_norm": 0.31915006041526794, "learning_rate": 8.836886166922458e-05, "loss": -0.9153, "step": 15630 }, { "epoch": 43.44444444444444, "grad_norm": 0.24748331308364868, "learning_rate": 8.835118587460844e-05, "loss": -0.9201, "step": 15640 }, { "epoch": 43.47222222222222, "grad_norm": 0.31025031208992004, "learning_rate": 8.83334984299146e-05, "loss": -0.9133, "step": 15650 }, { "epoch": 43.5, "grad_norm": 0.3603517413139343, "learning_rate": 8.83157993405161e-05, "loss": -0.9135, "step": 15660 }, { "epoch": 43.52777777777778, "grad_norm": 0.3515012860298157, "learning_rate": 8.829808861178943e-05, "loss": -0.9226, "step": 15670 }, { "epoch": 43.55555555555556, "grad_norm": 0.5585892796516418, "learning_rate": 8.828036624911464e-05, "loss": -0.9147, "step": 15680 }, { "epoch": 43.583333333333336, "grad_norm": 0.19856752455234528, "learning_rate": 8.826263225787532e-05, "loss": -0.9225, "step": 15690 }, { "epoch": 43.611111111111114, "grad_norm": 0.38493263721466064, "learning_rate": 8.824488664345858e-05, "loss": -0.9247, "step": 15700 }, { "epoch": 43.638888888888886, "grad_norm": 0.3239786922931671, "learning_rate": 8.822712941125508e-05, "loss": -0.9243, "step": 15710 }, { "epoch": 43.666666666666664, "grad_norm": 0.20186147093772888, "learning_rate": 8.820936056665898e-05, "loss": -0.9266, "step": 15720 }, { "epoch": 43.69444444444444, "grad_norm": 0.3504980206489563, "learning_rate": 8.819158011506801e-05, "loss": -0.9211, "step": 15730 }, { "epoch": 43.72222222222222, "grad_norm": 0.37510204315185547, "learning_rate": 8.81737880618834e-05, "loss": -0.9244, "step": 15740 }, { "epoch": 43.75, "grad_norm": 0.24302509427070618, "learning_rate": 8.815598441250987e-05, "loss": -0.9178, "step": 15750 }, { "epoch": 43.77777777777778, "grad_norm": 0.4323848783969879, "learning_rate": 8.813816917235576e-05, "loss": -0.9248, "step": 15760 }, { "epoch": 43.80555555555556, "grad_norm": 0.33055225014686584, "learning_rate": 8.812034234683282e-05, "loss": -0.9196, "step": 15770 }, { "epoch": 43.833333333333336, "grad_norm": 0.2668856680393219, "learning_rate": 8.810250394135637e-05, "loss": -0.9234, "step": 15780 }, { "epoch": 43.861111111111114, "grad_norm": 0.2892667353153229, "learning_rate": 8.808465396134529e-05, "loss": -0.9187, "step": 15790 }, { "epoch": 43.888888888888886, "grad_norm": 0.3907857835292816, "learning_rate": 8.806679241222189e-05, "loss": -0.9229, "step": 15800 }, { "epoch": 43.916666666666664, "grad_norm": 0.2797285318374634, "learning_rate": 8.804891929941203e-05, "loss": -0.914, "step": 15810 }, { "epoch": 43.94444444444444, "grad_norm": 0.39149418473243713, "learning_rate": 8.803103462834514e-05, "loss": -0.9199, "step": 15820 }, { "epoch": 43.97222222222222, "grad_norm": 0.19342191517353058, "learning_rate": 8.801313840445408e-05, "loss": -0.9187, "step": 15830 }, { "epoch": 44.0, "grad_norm": 0.2544451653957367, "learning_rate": 8.799523063317524e-05, "loss": -0.922, "step": 15840 }, { "epoch": 44.02777777777778, "grad_norm": 0.3212417960166931, "learning_rate": 8.797731131994854e-05, "loss": -0.9245, "step": 15850 }, { "epoch": 44.05555555555556, "grad_norm": 0.3155680298805237, "learning_rate": 8.795938047021739e-05, "loss": -0.9173, "step": 15860 }, { "epoch": 44.083333333333336, "grad_norm": 0.43539610505104065, "learning_rate": 8.794143808942872e-05, "loss": -0.9188, "step": 15870 }, { "epoch": 44.111111111111114, "grad_norm": 0.353629469871521, "learning_rate": 8.792348418303296e-05, "loss": -0.9221, "step": 15880 }, { "epoch": 44.138888888888886, "grad_norm": 0.3577032685279846, "learning_rate": 8.790551875648398e-05, "loss": -0.9164, "step": 15890 }, { "epoch": 44.166666666666664, "grad_norm": 0.3417088985443115, "learning_rate": 8.788754181523926e-05, "loss": -0.9162, "step": 15900 }, { "epoch": 44.19444444444444, "grad_norm": 0.3500460386276245, "learning_rate": 8.78695533647597e-05, "loss": -0.9164, "step": 15910 }, { "epoch": 44.22222222222222, "grad_norm": 0.21449729800224304, "learning_rate": 8.785155341050972e-05, "loss": -0.9107, "step": 15920 }, { "epoch": 44.25, "grad_norm": 0.4517425000667572, "learning_rate": 8.783354195795721e-05, "loss": -0.9181, "step": 15930 }, { "epoch": 44.27777777777778, "grad_norm": 0.36741897463798523, "learning_rate": 8.78155190125736e-05, "loss": -0.9174, "step": 15940 }, { "epoch": 44.30555555555556, "grad_norm": 0.21824738383293152, "learning_rate": 8.779748457983378e-05, "loss": -0.9162, "step": 15950 }, { "epoch": 44.333333333333336, "grad_norm": 0.2355160415172577, "learning_rate": 8.777943866521612e-05, "loss": -0.9115, "step": 15960 }, { "epoch": 44.361111111111114, "grad_norm": 0.27987226843833923, "learning_rate": 8.77613812742025e-05, "loss": -0.9244, "step": 15970 }, { "epoch": 44.388888888888886, "grad_norm": 0.30735617876052856, "learning_rate": 8.774331241227829e-05, "loss": -0.9177, "step": 15980 }, { "epoch": 44.416666666666664, "grad_norm": 0.36498209834098816, "learning_rate": 8.772523208493232e-05, "loss": -0.9265, "step": 15990 }, { "epoch": 44.44444444444444, "grad_norm": 0.18248078227043152, "learning_rate": 8.770714029765692e-05, "loss": -0.9203, "step": 16000 }, { "epoch": 44.47222222222222, "grad_norm": 0.23451805114746094, "learning_rate": 8.768903705594789e-05, "loss": -0.9147, "step": 16010 }, { "epoch": 44.5, "grad_norm": 0.26402735710144043, "learning_rate": 8.767092236530453e-05, "loss": -0.9192, "step": 16020 }, { "epoch": 44.52777777777778, "grad_norm": 0.29378747940063477, "learning_rate": 8.76527962312296e-05, "loss": -0.9145, "step": 16030 }, { "epoch": 44.55555555555556, "grad_norm": 0.19429832696914673, "learning_rate": 8.763465865922934e-05, "loss": -0.9261, "step": 16040 }, { "epoch": 44.583333333333336, "grad_norm": 0.33381712436676025, "learning_rate": 8.761650965481347e-05, "loss": -0.9209, "step": 16050 }, { "epoch": 44.611111111111114, "grad_norm": 0.20102912187576294, "learning_rate": 8.759834922349516e-05, "loss": -0.9224, "step": 16060 }, { "epoch": 44.638888888888886, "grad_norm": 0.35689061880111694, "learning_rate": 8.758017737079108e-05, "loss": -0.9203, "step": 16070 }, { "epoch": 44.666666666666664, "grad_norm": 0.31732839345932007, "learning_rate": 8.756199410222137e-05, "loss": -0.9297, "step": 16080 }, { "epoch": 44.69444444444444, "grad_norm": 0.30950474739074707, "learning_rate": 8.754379942330963e-05, "loss": -0.924, "step": 16090 }, { "epoch": 44.72222222222222, "grad_norm": 0.19697345793247223, "learning_rate": 8.75255933395829e-05, "loss": -0.9274, "step": 16100 }, { "epoch": 44.75, "grad_norm": 0.32826483249664307, "learning_rate": 8.750737585657171e-05, "loss": -0.9214, "step": 16110 }, { "epoch": 44.77777777777778, "grad_norm": 0.33892542123794556, "learning_rate": 8.748914697981008e-05, "loss": -0.9221, "step": 16120 }, { "epoch": 44.80555555555556, "grad_norm": 0.34005752205848694, "learning_rate": 8.747090671483542e-05, "loss": -0.92, "step": 16130 }, { "epoch": 44.833333333333336, "grad_norm": 0.23004065454006195, "learning_rate": 8.745265506718869e-05, "loss": -0.9279, "step": 16140 }, { "epoch": 44.861111111111114, "grad_norm": 0.23156709969043732, "learning_rate": 8.74343920424142e-05, "loss": -0.9223, "step": 16150 }, { "epoch": 44.888888888888886, "grad_norm": 0.22454515099525452, "learning_rate": 8.741611764605982e-05, "loss": -0.9233, "step": 16160 }, { "epoch": 44.916666666666664, "grad_norm": 0.23311880230903625, "learning_rate": 8.739783188367682e-05, "loss": -0.9241, "step": 16170 }, { "epoch": 44.94444444444444, "grad_norm": 0.23222874104976654, "learning_rate": 8.737953476081991e-05, "loss": -0.928, "step": 16180 }, { "epoch": 44.97222222222222, "grad_norm": 0.2808738350868225, "learning_rate": 8.73612262830473e-05, "loss": -0.9248, "step": 16190 }, { "epoch": 45.0, "grad_norm": 0.2774062156677246, "learning_rate": 8.734290645592061e-05, "loss": -0.9294, "step": 16200 }, { "epoch": 45.02777777777778, "grad_norm": 0.19433195888996124, "learning_rate": 8.732457528500493e-05, "loss": -0.923, "step": 16210 }, { "epoch": 45.05555555555556, "grad_norm": 0.24463017284870148, "learning_rate": 8.730623277586875e-05, "loss": -0.9248, "step": 16220 }, { "epoch": 45.083333333333336, "grad_norm": 0.23854681849479675, "learning_rate": 8.72878789340841e-05, "loss": -0.9283, "step": 16230 }, { "epoch": 45.111111111111114, "grad_norm": 0.3087669014930725, "learning_rate": 8.726951376522635e-05, "loss": -0.9191, "step": 16240 }, { "epoch": 45.138888888888886, "grad_norm": 0.28739723563194275, "learning_rate": 8.725113727487435e-05, "loss": -0.9243, "step": 16250 }, { "epoch": 45.166666666666664, "grad_norm": 0.24851255118846893, "learning_rate": 8.723274946861042e-05, "loss": -0.9274, "step": 16260 }, { "epoch": 45.19444444444444, "grad_norm": 0.32095763087272644, "learning_rate": 8.721435035202026e-05, "loss": -0.925, "step": 16270 }, { "epoch": 45.22222222222222, "grad_norm": 0.23268663883209229, "learning_rate": 8.719593993069306e-05, "loss": -0.9161, "step": 16280 }, { "epoch": 45.25, "grad_norm": 0.3381432890892029, "learning_rate": 8.717751821022139e-05, "loss": -0.9236, "step": 16290 }, { "epoch": 45.27777777777778, "grad_norm": 0.22727255523204803, "learning_rate": 8.715908519620134e-05, "loss": -0.9204, "step": 16300 }, { "epoch": 45.30555555555556, "grad_norm": 0.23801955580711365, "learning_rate": 8.71406408942323e-05, "loss": -0.9197, "step": 16310 }, { "epoch": 45.333333333333336, "grad_norm": 0.26239487528800964, "learning_rate": 8.712218530991723e-05, "loss": -0.9254, "step": 16320 }, { "epoch": 45.361111111111114, "grad_norm": 0.28990837931632996, "learning_rate": 8.710371844886241e-05, "loss": -0.9254, "step": 16330 }, { "epoch": 45.388888888888886, "grad_norm": 0.24782408773899078, "learning_rate": 8.708524031667758e-05, "loss": -0.9245, "step": 16340 }, { "epoch": 45.416666666666664, "grad_norm": 0.2584209740161896, "learning_rate": 8.706675091897592e-05, "loss": -0.9211, "step": 16350 }, { "epoch": 45.44444444444444, "grad_norm": 0.2285592257976532, "learning_rate": 8.704825026137404e-05, "loss": -0.9208, "step": 16360 }, { "epoch": 45.47222222222222, "grad_norm": 0.4030211865901947, "learning_rate": 8.702973834949192e-05, "loss": -0.9218, "step": 16370 }, { "epoch": 45.5, "grad_norm": 0.24202477931976318, "learning_rate": 8.701121518895301e-05, "loss": -0.9238, "step": 16380 }, { "epoch": 45.52777777777778, "grad_norm": 0.3094029724597931, "learning_rate": 8.699268078538414e-05, "loss": -0.9173, "step": 16390 }, { "epoch": 45.55555555555556, "grad_norm": 0.20559486746788025, "learning_rate": 8.69741351444156e-05, "loss": -0.9291, "step": 16400 }, { "epoch": 45.583333333333336, "grad_norm": 0.4617546498775482, "learning_rate": 8.695557827168101e-05, "loss": -0.9231, "step": 16410 }, { "epoch": 45.611111111111114, "grad_norm": 0.3108498156070709, "learning_rate": 8.693701017281753e-05, "loss": -0.9241, "step": 16420 }, { "epoch": 45.638888888888886, "grad_norm": 0.3208109438419342, "learning_rate": 8.691843085346563e-05, "loss": -0.915, "step": 16430 }, { "epoch": 45.666666666666664, "grad_norm": 0.20494745671749115, "learning_rate": 8.689984031926919e-05, "loss": -0.9289, "step": 16440 }, { "epoch": 45.69444444444444, "grad_norm": 0.2630619704723358, "learning_rate": 8.688123857587555e-05, "loss": -0.9196, "step": 16450 }, { "epoch": 45.72222222222222, "grad_norm": 0.24305397272109985, "learning_rate": 8.686262562893544e-05, "loss": -0.9205, "step": 16460 }, { "epoch": 45.75, "grad_norm": 0.24307283759117126, "learning_rate": 8.684400148410294e-05, "loss": -0.9268, "step": 16470 }, { "epoch": 45.77777777777778, "grad_norm": 0.20179085433483124, "learning_rate": 8.682536614703562e-05, "loss": -0.9256, "step": 16480 }, { "epoch": 45.80555555555556, "grad_norm": 0.2754277288913727, "learning_rate": 8.680671962339437e-05, "loss": -0.9276, "step": 16490 }, { "epoch": 45.833333333333336, "grad_norm": 0.23345789313316345, "learning_rate": 8.678806191884352e-05, "loss": -0.9271, "step": 16500 }, { "epoch": 45.861111111111114, "grad_norm": 0.27838510274887085, "learning_rate": 8.67693930390508e-05, "loss": -0.9259, "step": 16510 }, { "epoch": 45.888888888888886, "grad_norm": 0.30544888973236084, "learning_rate": 8.67507129896873e-05, "loss": -0.929, "step": 16520 }, { "epoch": 45.916666666666664, "grad_norm": 0.28128841519355774, "learning_rate": 8.673202177642757e-05, "loss": -0.9166, "step": 16530 }, { "epoch": 45.94444444444444, "grad_norm": 0.3116399347782135, "learning_rate": 8.671331940494945e-05, "loss": -0.9155, "step": 16540 }, { "epoch": 45.97222222222222, "grad_norm": 0.28525981307029724, "learning_rate": 8.669460588093427e-05, "loss": -0.9237, "step": 16550 }, { "epoch": 46.0, "grad_norm": 0.33775386214256287, "learning_rate": 8.667588121006667e-05, "loss": -0.9267, "step": 16560 }, { "epoch": 46.02777777777778, "grad_norm": 0.423795610666275, "learning_rate": 8.665714539803475e-05, "loss": -0.9186, "step": 16570 }, { "epoch": 46.05555555555556, "grad_norm": 0.2918187379837036, "learning_rate": 8.663839845052993e-05, "loss": -0.9267, "step": 16580 }, { "epoch": 46.083333333333336, "grad_norm": 0.3479655981063843, "learning_rate": 8.661964037324703e-05, "loss": -0.9233, "step": 16590 }, { "epoch": 46.111111111111114, "grad_norm": 0.29261520504951477, "learning_rate": 8.660087117188427e-05, "loss": -0.9267, "step": 16600 }, { "epoch": 46.138888888888886, "grad_norm": 0.31197410821914673, "learning_rate": 8.658209085214325e-05, "loss": -0.9218, "step": 16610 }, { "epoch": 46.166666666666664, "grad_norm": 0.3799397945404053, "learning_rate": 8.656329941972891e-05, "loss": -0.9179, "step": 16620 }, { "epoch": 46.19444444444444, "grad_norm": 0.3157538175582886, "learning_rate": 8.654449688034963e-05, "loss": -0.9234, "step": 16630 }, { "epoch": 46.22222222222222, "grad_norm": 0.3100598156452179, "learning_rate": 8.652568323971706e-05, "loss": -0.9228, "step": 16640 }, { "epoch": 46.25, "grad_norm": 0.3287033140659332, "learning_rate": 8.650685850354636e-05, "loss": -0.9246, "step": 16650 }, { "epoch": 46.27777777777778, "grad_norm": 0.29683494567871094, "learning_rate": 8.648802267755593e-05, "loss": -0.9228, "step": 16660 }, { "epoch": 46.30555555555556, "grad_norm": 0.3045928478240967, "learning_rate": 8.646917576746764e-05, "loss": -0.9231, "step": 16670 }, { "epoch": 46.333333333333336, "grad_norm": 0.25064218044281006, "learning_rate": 8.645031777900666e-05, "loss": -0.9278, "step": 16680 }, { "epoch": 46.361111111111114, "grad_norm": 0.2605026066303253, "learning_rate": 8.643144871790154e-05, "loss": -0.9206, "step": 16690 }, { "epoch": 46.388888888888886, "grad_norm": 0.27713119983673096, "learning_rate": 8.641256858988424e-05, "loss": -0.9263, "step": 16700 }, { "epoch": 46.416666666666664, "grad_norm": 0.290787011384964, "learning_rate": 8.639367740069e-05, "loss": -0.9252, "step": 16710 }, { "epoch": 46.44444444444444, "grad_norm": 0.1763259619474411, "learning_rate": 8.63747751560575e-05, "loss": -0.9232, "step": 16720 }, { "epoch": 46.47222222222222, "grad_norm": 0.2732510566711426, "learning_rate": 8.635586186172871e-05, "loss": -0.9269, "step": 16730 }, { "epoch": 46.5, "grad_norm": 0.2649574875831604, "learning_rate": 8.633693752344902e-05, "loss": -0.9292, "step": 16740 }, { "epoch": 46.52777777777778, "grad_norm": 0.20868748426437378, "learning_rate": 8.631800214696713e-05, "loss": -0.925, "step": 16750 }, { "epoch": 46.55555555555556, "grad_norm": 0.22685037553310394, "learning_rate": 8.629905573803511e-05, "loss": -0.9231, "step": 16760 }, { "epoch": 46.583333333333336, "grad_norm": 0.29456430673599243, "learning_rate": 8.628009830240839e-05, "loss": -0.9213, "step": 16770 }, { "epoch": 46.611111111111114, "grad_norm": 0.32766175270080566, "learning_rate": 8.626112984584571e-05, "loss": -0.9205, "step": 16780 }, { "epoch": 46.638888888888886, "grad_norm": 0.2432580292224884, "learning_rate": 8.62421503741092e-05, "loss": -0.9248, "step": 16790 }, { "epoch": 46.666666666666664, "grad_norm": 0.17141981422901154, "learning_rate": 8.622315989296432e-05, "loss": -0.9331, "step": 16800 }, { "epoch": 46.69444444444444, "grad_norm": 0.19156743586063385, "learning_rate": 8.62041584081799e-05, "loss": -0.9266, "step": 16810 }, { "epoch": 46.72222222222222, "grad_norm": 0.258279025554657, "learning_rate": 8.618514592552807e-05, "loss": -0.9256, "step": 16820 }, { "epoch": 46.75, "grad_norm": 0.25659018754959106, "learning_rate": 8.616612245078431e-05, "loss": -0.9265, "step": 16830 }, { "epoch": 46.77777777777778, "grad_norm": 0.28189003467559814, "learning_rate": 8.614708798972746e-05, "loss": -0.9252, "step": 16840 }, { "epoch": 46.80555555555556, "grad_norm": 0.2946809232234955, "learning_rate": 8.61280425481397e-05, "loss": -0.928, "step": 16850 }, { "epoch": 46.833333333333336, "grad_norm": 0.2156822383403778, "learning_rate": 8.61089861318065e-05, "loss": -0.9242, "step": 16860 }, { "epoch": 46.861111111111114, "grad_norm": 0.22242297232151031, "learning_rate": 8.608991874651673e-05, "loss": -0.9291, "step": 16870 }, { "epoch": 46.888888888888886, "grad_norm": 0.30979275703430176, "learning_rate": 8.607084039806255e-05, "loss": -0.9263, "step": 16880 }, { "epoch": 46.916666666666664, "grad_norm": 0.24880072474479675, "learning_rate": 8.605175109223944e-05, "loss": -0.9278, "step": 16890 }, { "epoch": 46.94444444444444, "grad_norm": 0.17043951153755188, "learning_rate": 8.603265083484624e-05, "loss": -0.9288, "step": 16900 }, { "epoch": 46.97222222222222, "grad_norm": 0.21457235515117645, "learning_rate": 8.60135396316851e-05, "loss": -0.9274, "step": 16910 }, { "epoch": 47.0, "grad_norm": 0.2904838025569916, "learning_rate": 8.599441748856152e-05, "loss": -0.927, "step": 16920 }, { "epoch": 47.02777777777778, "grad_norm": 0.17870332300662994, "learning_rate": 8.597528441128427e-05, "loss": -0.9257, "step": 16930 }, { "epoch": 47.05555555555556, "grad_norm": 0.35341671109199524, "learning_rate": 8.595614040566549e-05, "loss": -0.9265, "step": 16940 }, { "epoch": 47.083333333333336, "grad_norm": 0.24762827157974243, "learning_rate": 8.593698547752063e-05, "loss": -0.927, "step": 16950 }, { "epoch": 47.111111111111114, "grad_norm": 0.32072174549102783, "learning_rate": 8.591781963266843e-05, "loss": -0.9129, "step": 16960 }, { "epoch": 47.138888888888886, "grad_norm": 0.27776065468788147, "learning_rate": 8.5898642876931e-05, "loss": -0.9268, "step": 16970 }, { "epoch": 47.166666666666664, "grad_norm": 0.21192510426044464, "learning_rate": 8.587945521613369e-05, "loss": -0.9176, "step": 16980 }, { "epoch": 47.19444444444444, "grad_norm": 0.24543234705924988, "learning_rate": 8.586025665610524e-05, "loss": -0.9299, "step": 16990 }, { "epoch": 47.22222222222222, "grad_norm": 0.2618130147457123, "learning_rate": 8.584104720267765e-05, "loss": -0.9238, "step": 17000 }, { "epoch": 47.25, "grad_norm": 0.23607735335826874, "learning_rate": 8.582182686168625e-05, "loss": -0.927, "step": 17010 }, { "epoch": 47.27777777777778, "grad_norm": 0.2755493223667145, "learning_rate": 8.580259563896967e-05, "loss": -0.9214, "step": 17020 }, { "epoch": 47.30555555555556, "grad_norm": 0.3554861545562744, "learning_rate": 8.578335354036983e-05, "loss": -0.9249, "step": 17030 }, { "epoch": 47.333333333333336, "grad_norm": 0.28501832485198975, "learning_rate": 8.576410057173201e-05, "loss": -0.9258, "step": 17040 }, { "epoch": 47.361111111111114, "grad_norm": 0.2633955180644989, "learning_rate": 8.574483673890474e-05, "loss": -0.9265, "step": 17050 }, { "epoch": 47.388888888888886, "grad_norm": 0.19416774809360504, "learning_rate": 8.572556204773983e-05, "loss": -0.9285, "step": 17060 }, { "epoch": 47.416666666666664, "grad_norm": 0.361386239528656, "learning_rate": 8.570627650409246e-05, "loss": -0.9307, "step": 17070 }, { "epoch": 47.44444444444444, "grad_norm": 0.18803465366363525, "learning_rate": 8.568698011382107e-05, "loss": -0.9235, "step": 17080 }, { "epoch": 47.47222222222222, "grad_norm": 0.2268105298280716, "learning_rate": 8.566767288278738e-05, "loss": -0.9183, "step": 17090 }, { "epoch": 47.5, "grad_norm": 0.28922030329704285, "learning_rate": 8.56483548168564e-05, "loss": -0.9245, "step": 17100 }, { "epoch": 47.52777777777778, "grad_norm": 0.22344227135181427, "learning_rate": 8.562902592189648e-05, "loss": -0.9249, "step": 17110 }, { "epoch": 47.55555555555556, "grad_norm": 0.26333871483802795, "learning_rate": 8.560968620377921e-05, "loss": -0.9294, "step": 17120 }, { "epoch": 47.583333333333336, "grad_norm": 0.2838645279407501, "learning_rate": 8.559033566837951e-05, "loss": -0.9258, "step": 17130 }, { "epoch": 47.611111111111114, "grad_norm": 0.20298711955547333, "learning_rate": 8.557097432157551e-05, "loss": -0.9225, "step": 17140 }, { "epoch": 47.638888888888886, "grad_norm": 0.2690686583518982, "learning_rate": 8.555160216924872e-05, "loss": -0.9289, "step": 17150 }, { "epoch": 47.666666666666664, "grad_norm": 0.2669510841369629, "learning_rate": 8.55322192172839e-05, "loss": -0.9303, "step": 17160 }, { "epoch": 47.69444444444444, "grad_norm": 0.2418321818113327, "learning_rate": 8.551282547156902e-05, "loss": -0.9299, "step": 17170 }, { "epoch": 47.72222222222222, "grad_norm": 0.2647036910057068, "learning_rate": 8.549342093799544e-05, "loss": -0.9258, "step": 17180 }, { "epoch": 47.75, "grad_norm": 0.2762335538864136, "learning_rate": 8.547400562245773e-05, "loss": -0.9206, "step": 17190 }, { "epoch": 47.77777777777778, "grad_norm": 0.2656387388706207, "learning_rate": 8.545457953085374e-05, "loss": -0.9221, "step": 17200 }, { "epoch": 47.80555555555556, "grad_norm": 0.3330097794532776, "learning_rate": 8.543514266908463e-05, "loss": -0.9239, "step": 17210 }, { "epoch": 47.833333333333336, "grad_norm": 0.30362993478775024, "learning_rate": 8.541569504305478e-05, "loss": -0.9179, "step": 17220 }, { "epoch": 47.861111111111114, "grad_norm": 0.24856744706630707, "learning_rate": 8.539623665867187e-05, "loss": -0.9228, "step": 17230 }, { "epoch": 47.888888888888886, "grad_norm": 0.26128438115119934, "learning_rate": 8.537676752184685e-05, "loss": -0.9255, "step": 17240 }, { "epoch": 47.916666666666664, "grad_norm": 0.3736089766025543, "learning_rate": 8.53572876384939e-05, "loss": -0.9247, "step": 17250 }, { "epoch": 47.94444444444444, "grad_norm": 0.3580988049507141, "learning_rate": 8.533779701453056e-05, "loss": -0.9255, "step": 17260 }, { "epoch": 47.97222222222222, "grad_norm": 0.3170616924762726, "learning_rate": 8.53182956558775e-05, "loss": -0.9236, "step": 17270 }, { "epoch": 48.0, "grad_norm": 0.2570171356201172, "learning_rate": 8.529878356845877e-05, "loss": -0.927, "step": 17280 }, { "epoch": 48.02777777777778, "grad_norm": 0.3387782573699951, "learning_rate": 8.527926075820158e-05, "loss": -0.9237, "step": 17290 }, { "epoch": 48.05555555555556, "grad_norm": 0.2722717225551605, "learning_rate": 8.525972723103648e-05, "loss": -0.9237, "step": 17300 }, { "epoch": 48.083333333333336, "grad_norm": 0.26538416743278503, "learning_rate": 8.524018299289722e-05, "loss": -0.9264, "step": 17310 }, { "epoch": 48.111111111111114, "grad_norm": 0.3735320270061493, "learning_rate": 8.522062804972083e-05, "loss": -0.9262, "step": 17320 }, { "epoch": 48.138888888888886, "grad_norm": 0.15319697558879852, "learning_rate": 8.520106240744759e-05, "loss": -0.9264, "step": 17330 }, { "epoch": 48.166666666666664, "grad_norm": 0.3817085921764374, "learning_rate": 8.518148607202102e-05, "loss": -0.924, "step": 17340 }, { "epoch": 48.19444444444444, "grad_norm": 0.29327887296676636, "learning_rate": 8.51618990493879e-05, "loss": -0.9245, "step": 17350 }, { "epoch": 48.22222222222222, "grad_norm": 0.1740938276052475, "learning_rate": 8.514230134549823e-05, "loss": -0.9207, "step": 17360 }, { "epoch": 48.25, "grad_norm": 0.23025989532470703, "learning_rate": 8.51226929663053e-05, "loss": -0.9257, "step": 17370 }, { "epoch": 48.27777777777778, "grad_norm": 0.2554945945739746, "learning_rate": 8.51030739177656e-05, "loss": -0.9251, "step": 17380 }, { "epoch": 48.30555555555556, "grad_norm": 0.19967332482337952, "learning_rate": 8.508344420583889e-05, "loss": -0.9221, "step": 17390 }, { "epoch": 48.333333333333336, "grad_norm": 0.20565378665924072, "learning_rate": 8.506380383648816e-05, "loss": -0.929, "step": 17400 }, { "epoch": 48.361111111111114, "grad_norm": 0.2775403559207916, "learning_rate": 8.504415281567963e-05, "loss": -0.9304, "step": 17410 }, { "epoch": 48.388888888888886, "grad_norm": 0.1864451766014099, "learning_rate": 8.502449114938275e-05, "loss": -0.9321, "step": 17420 }, { "epoch": 48.416666666666664, "grad_norm": 0.23881064355373383, "learning_rate": 8.500481884357025e-05, "loss": -0.9259, "step": 17430 }, { "epoch": 48.44444444444444, "grad_norm": 0.3122458755970001, "learning_rate": 8.498513590421801e-05, "loss": -0.9257, "step": 17440 }, { "epoch": 48.47222222222222, "grad_norm": 0.23839035630226135, "learning_rate": 8.496544233730522e-05, "loss": -0.9228, "step": 17450 }, { "epoch": 48.5, "grad_norm": 0.2733101546764374, "learning_rate": 8.494573814881426e-05, "loss": -0.9275, "step": 17460 }, { "epoch": 48.52777777777778, "grad_norm": 0.22155752778053284, "learning_rate": 8.492602334473074e-05, "loss": -0.9297, "step": 17470 }, { "epoch": 48.55555555555556, "grad_norm": 0.4200872480869293, "learning_rate": 8.49062979310435e-05, "loss": -0.9201, "step": 17480 }, { "epoch": 48.583333333333336, "grad_norm": 0.3241598904132843, "learning_rate": 8.488656191374458e-05, "loss": -0.9212, "step": 17490 }, { "epoch": 48.611111111111114, "grad_norm": 0.36301928758621216, "learning_rate": 8.48668152988293e-05, "loss": -0.9267, "step": 17500 }, { "epoch": 48.638888888888886, "grad_norm": 0.2559203803539276, "learning_rate": 8.484705809229612e-05, "loss": -0.9266, "step": 17510 }, { "epoch": 48.666666666666664, "grad_norm": 0.34503230452537537, "learning_rate": 8.482729030014677e-05, "loss": -0.9237, "step": 17520 }, { "epoch": 48.69444444444444, "grad_norm": 0.28684550523757935, "learning_rate": 8.48075119283862e-05, "loss": -0.929, "step": 17530 }, { "epoch": 48.72222222222222, "grad_norm": 0.2855939269065857, "learning_rate": 8.478772298302254e-05, "loss": -0.9268, "step": 17540 }, { "epoch": 48.75, "grad_norm": 0.31941157579421997, "learning_rate": 8.476792347006716e-05, "loss": -0.9222, "step": 17550 }, { "epoch": 48.77777777777778, "grad_norm": 0.5061282515525818, "learning_rate": 8.474811339553462e-05, "loss": -0.9267, "step": 17560 }, { "epoch": 48.80555555555556, "grad_norm": 0.4198087751865387, "learning_rate": 8.47282927654427e-05, "loss": -0.9141, "step": 17570 }, { "epoch": 48.833333333333336, "grad_norm": 0.29300937056541443, "learning_rate": 8.470846158581238e-05, "loss": -0.931, "step": 17580 }, { "epoch": 48.861111111111114, "grad_norm": 0.316471666097641, "learning_rate": 8.468861986266787e-05, "loss": -0.9236, "step": 17590 }, { "epoch": 48.888888888888886, "grad_norm": 0.20738740265369415, "learning_rate": 8.466876760203654e-05, "loss": -0.9267, "step": 17600 }, { "epoch": 48.916666666666664, "grad_norm": 0.30087316036224365, "learning_rate": 8.464890480994898e-05, "loss": -0.9203, "step": 17610 }, { "epoch": 48.94444444444444, "grad_norm": 0.2256436049938202, "learning_rate": 8.462903149243899e-05, "loss": -0.9256, "step": 17620 }, { "epoch": 48.97222222222222, "grad_norm": 0.25340551137924194, "learning_rate": 8.460914765554357e-05, "loss": -0.9299, "step": 17630 }, { "epoch": 49.0, "grad_norm": 0.5104928612709045, "learning_rate": 8.458925330530288e-05, "loss": -0.9192, "step": 17640 }, { "epoch": 49.02777777777778, "grad_norm": 0.24924127757549286, "learning_rate": 8.456934844776032e-05, "loss": -0.9288, "step": 17650 }, { "epoch": 49.05555555555556, "grad_norm": 0.3491559326648712, "learning_rate": 8.454943308896246e-05, "loss": -0.9224, "step": 17660 }, { "epoch": 49.083333333333336, "grad_norm": 0.29652220010757446, "learning_rate": 8.452950723495905e-05, "loss": -0.9303, "step": 17670 }, { "epoch": 49.111111111111114, "grad_norm": 0.35996878147125244, "learning_rate": 8.450957089180303e-05, "loss": -0.9278, "step": 17680 }, { "epoch": 49.138888888888886, "grad_norm": 0.36424025893211365, "learning_rate": 8.448962406555055e-05, "loss": -0.9267, "step": 17690 }, { "epoch": 49.166666666666664, "grad_norm": 0.3754829168319702, "learning_rate": 8.446966676226093e-05, "loss": -0.922, "step": 17700 }, { "epoch": 49.19444444444444, "grad_norm": 0.4001710116863251, "learning_rate": 8.444969898799667e-05, "loss": -0.9238, "step": 17710 }, { "epoch": 49.22222222222222, "grad_norm": 0.39788952469825745, "learning_rate": 8.442972074882343e-05, "loss": -0.9242, "step": 17720 }, { "epoch": 49.25, "grad_norm": 0.26204365491867065, "learning_rate": 8.44097320508101e-05, "loss": -0.9264, "step": 17730 }, { "epoch": 49.27777777777778, "grad_norm": 0.3325068950653076, "learning_rate": 8.43897329000287e-05, "loss": -0.9297, "step": 17740 }, { "epoch": 49.30555555555556, "grad_norm": 0.2992180287837982, "learning_rate": 8.436972330255448e-05, "loss": -0.9261, "step": 17750 }, { "epoch": 49.333333333333336, "grad_norm": 0.32437780499458313, "learning_rate": 8.434970326446579e-05, "loss": -0.9235, "step": 17760 }, { "epoch": 49.361111111111114, "grad_norm": 0.35037168860435486, "learning_rate": 8.432967279184418e-05, "loss": -0.9246, "step": 17770 }, { "epoch": 49.388888888888886, "grad_norm": 0.30388757586479187, "learning_rate": 8.430963189077441e-05, "loss": -0.9306, "step": 17780 }, { "epoch": 49.416666666666664, "grad_norm": 0.28164705634117126, "learning_rate": 8.428958056734437e-05, "loss": -0.9299, "step": 17790 }, { "epoch": 49.44444444444444, "grad_norm": 0.30533546209335327, "learning_rate": 8.426951882764513e-05, "loss": -0.916, "step": 17800 }, { "epoch": 49.47222222222222, "grad_norm": 0.1883441060781479, "learning_rate": 8.424944667777089e-05, "loss": -0.9285, "step": 17810 }, { "epoch": 49.5, "grad_norm": 0.2723890244960785, "learning_rate": 8.422936412381905e-05, "loss": -0.9306, "step": 17820 }, { "epoch": 49.52777777777778, "grad_norm": 0.26253849267959595, "learning_rate": 8.420927117189017e-05, "loss": -0.9242, "step": 17830 }, { "epoch": 49.55555555555556, "grad_norm": 0.26928290724754333, "learning_rate": 8.418916782808795e-05, "loss": -0.928, "step": 17840 }, { "epoch": 49.583333333333336, "grad_norm": 0.3666425943374634, "learning_rate": 8.416905409851926e-05, "loss": -0.929, "step": 17850 }, { "epoch": 49.611111111111114, "grad_norm": 0.2703123390674591, "learning_rate": 8.41489299892941e-05, "loss": -0.9309, "step": 17860 }, { "epoch": 49.638888888888886, "grad_norm": 0.28211531043052673, "learning_rate": 8.412879550652566e-05, "loss": -0.925, "step": 17870 }, { "epoch": 49.666666666666664, "grad_norm": 0.30671608448028564, "learning_rate": 8.410865065633029e-05, "loss": -0.9289, "step": 17880 }, { "epoch": 49.69444444444444, "grad_norm": 0.38424545526504517, "learning_rate": 8.408849544482742e-05, "loss": -0.9247, "step": 17890 }, { "epoch": 49.72222222222222, "grad_norm": 0.4131236672401428, "learning_rate": 8.406832987813968e-05, "loss": -0.9264, "step": 17900 }, { "epoch": 49.75, "grad_norm": 0.37098219990730286, "learning_rate": 8.404815396239286e-05, "loss": -0.9274, "step": 17910 }, { "epoch": 49.77777777777778, "grad_norm": 0.3801688253879547, "learning_rate": 8.402796770371587e-05, "loss": -0.9235, "step": 17920 }, { "epoch": 49.80555555555556, "grad_norm": 0.3665392994880676, "learning_rate": 8.400777110824071e-05, "loss": -0.9312, "step": 17930 }, { "epoch": 49.833333333333336, "grad_norm": 0.26704350113868713, "learning_rate": 8.398756418210263e-05, "loss": -0.9304, "step": 17940 }, { "epoch": 49.861111111111114, "grad_norm": 0.25499069690704346, "learning_rate": 8.396734693143993e-05, "loss": -0.9296, "step": 17950 }, { "epoch": 49.888888888888886, "grad_norm": 0.23365779221057892, "learning_rate": 8.39471193623941e-05, "loss": -0.9222, "step": 17960 }, { "epoch": 49.916666666666664, "grad_norm": 0.25554215908050537, "learning_rate": 8.392688148110974e-05, "loss": -0.9272, "step": 17970 }, { "epoch": 49.94444444444444, "grad_norm": 0.22708114981651306, "learning_rate": 8.390663329373456e-05, "loss": -0.9287, "step": 17980 }, { "epoch": 49.97222222222222, "grad_norm": 0.27692919969558716, "learning_rate": 8.388637480641944e-05, "loss": -0.9216, "step": 17990 }, { "epoch": 50.0, "grad_norm": 0.3385431468486786, "learning_rate": 8.386610602531837e-05, "loss": -0.9256, "step": 18000 }, { "epoch": 50.02777777777778, "grad_norm": 0.26961615681648254, "learning_rate": 8.384582695658847e-05, "loss": -0.9316, "step": 18010 }, { "epoch": 50.05555555555556, "grad_norm": 0.2092764526605606, "learning_rate": 8.382553760638999e-05, "loss": -0.9295, "step": 18020 }, { "epoch": 50.083333333333336, "grad_norm": 0.32751521468162537, "learning_rate": 8.380523798088631e-05, "loss": -0.9314, "step": 18030 }, { "epoch": 50.111111111111114, "grad_norm": 0.2657199501991272, "learning_rate": 8.378492808624389e-05, "loss": -0.9305, "step": 18040 }, { "epoch": 50.138888888888886, "grad_norm": 0.30841463804244995, "learning_rate": 8.376460792863237e-05, "loss": -0.9268, "step": 18050 }, { "epoch": 50.166666666666664, "grad_norm": 0.28344428539276123, "learning_rate": 8.374427751422444e-05, "loss": -0.9339, "step": 18060 }, { "epoch": 50.19444444444444, "grad_norm": 0.27209872007369995, "learning_rate": 8.3723936849196e-05, "loss": -0.9311, "step": 18070 }, { "epoch": 50.22222222222222, "grad_norm": 0.20227903127670288, "learning_rate": 8.370358593972595e-05, "loss": -0.9297, "step": 18080 }, { "epoch": 50.25, "grad_norm": 0.2513710856437683, "learning_rate": 8.36832247919964e-05, "loss": -0.9318, "step": 18090 }, { "epoch": 50.27777777777778, "grad_norm": 0.1808832287788391, "learning_rate": 8.36628534121925e-05, "loss": -0.9268, "step": 18100 }, { "epoch": 50.30555555555556, "grad_norm": 0.2596690058708191, "learning_rate": 8.364247180650254e-05, "loss": -0.9203, "step": 18110 }, { "epoch": 50.333333333333336, "grad_norm": 0.2589612901210785, "learning_rate": 8.362207998111794e-05, "loss": -0.931, "step": 18120 }, { "epoch": 50.361111111111114, "grad_norm": 0.18995700776576996, "learning_rate": 8.360167794223318e-05, "loss": -0.9306, "step": 18130 }, { "epoch": 50.388888888888886, "grad_norm": 0.3235227167606354, "learning_rate": 8.358126569604586e-05, "loss": -0.9293, "step": 18140 }, { "epoch": 50.416666666666664, "grad_norm": 0.19670957326889038, "learning_rate": 8.356084324875668e-05, "loss": -0.9268, "step": 18150 }, { "epoch": 50.44444444444444, "grad_norm": 0.3152536451816559, "learning_rate": 8.354041060656945e-05, "loss": -0.9305, "step": 18160 }, { "epoch": 50.47222222222222, "grad_norm": 0.36967387795448303, "learning_rate": 8.351996777569106e-05, "loss": -0.9311, "step": 18170 }, { "epoch": 50.5, "grad_norm": 0.21266795694828033, "learning_rate": 8.349951476233148e-05, "loss": -0.9282, "step": 18180 }, { "epoch": 50.52777777777778, "grad_norm": 0.21623827517032623, "learning_rate": 8.347905157270386e-05, "loss": -0.926, "step": 18190 }, { "epoch": 50.55555555555556, "grad_norm": 0.25522878766059875, "learning_rate": 8.345857821302432e-05, "loss": -0.9242, "step": 18200 }, { "epoch": 50.583333333333336, "grad_norm": 0.3262898623943329, "learning_rate": 8.343809468951213e-05, "loss": -0.9245, "step": 18210 }, { "epoch": 50.611111111111114, "grad_norm": 0.2299589216709137, "learning_rate": 8.341760100838965e-05, "loss": -0.9207, "step": 18220 }, { "epoch": 50.638888888888886, "grad_norm": 0.2756154537200928, "learning_rate": 8.339709717588233e-05, "loss": -0.9245, "step": 18230 }, { "epoch": 50.666666666666664, "grad_norm": 0.282968670129776, "learning_rate": 8.33765831982187e-05, "loss": -0.931, "step": 18240 }, { "epoch": 50.69444444444444, "grad_norm": 0.28091952204704285, "learning_rate": 8.335605908163035e-05, "loss": -0.9316, "step": 18250 }, { "epoch": 50.72222222222222, "grad_norm": 0.21066391468048096, "learning_rate": 8.333552483235196e-05, "loss": -0.9327, "step": 18260 }, { "epoch": 50.75, "grad_norm": 0.26812630891799927, "learning_rate": 8.33149804566213e-05, "loss": -0.9277, "step": 18270 }, { "epoch": 50.77777777777778, "grad_norm": 0.1870378702878952, "learning_rate": 8.329442596067921e-05, "loss": -0.9279, "step": 18280 }, { "epoch": 50.80555555555556, "grad_norm": 0.2930411696434021, "learning_rate": 8.32738613507696e-05, "loss": -0.9253, "step": 18290 }, { "epoch": 50.833333333333336, "grad_norm": 0.3506859540939331, "learning_rate": 8.325328663313946e-05, "loss": -0.9221, "step": 18300 }, { "epoch": 50.861111111111114, "grad_norm": 0.22614112496376038, "learning_rate": 8.323270181403884e-05, "loss": -0.9277, "step": 18310 }, { "epoch": 50.888888888888886, "grad_norm": 0.1901329755783081, "learning_rate": 8.321210689972086e-05, "loss": -0.9268, "step": 18320 }, { "epoch": 50.916666666666664, "grad_norm": 0.2508808970451355, "learning_rate": 8.319150189644174e-05, "loss": -0.9278, "step": 18330 }, { "epoch": 50.94444444444444, "grad_norm": 0.27597522735595703, "learning_rate": 8.31708868104607e-05, "loss": -0.92, "step": 18340 }, { "epoch": 50.97222222222222, "grad_norm": 0.331464022397995, "learning_rate": 8.315026164804007e-05, "loss": -0.9225, "step": 18350 }, { "epoch": 51.0, "grad_norm": 0.3818557560443878, "learning_rate": 8.312962641544524e-05, "loss": -0.9274, "step": 18360 }, { "epoch": 51.02777777777778, "grad_norm": 0.317154198884964, "learning_rate": 8.310898111894465e-05, "loss": -0.9288, "step": 18370 }, { "epoch": 51.05555555555556, "grad_norm": 0.20453256368637085, "learning_rate": 8.308832576480977e-05, "loss": -0.9294, "step": 18380 }, { "epoch": 51.083333333333336, "grad_norm": 0.24128085374832153, "learning_rate": 8.306766035931519e-05, "loss": -0.9207, "step": 18390 }, { "epoch": 51.111111111111114, "grad_norm": 0.42857152223587036, "learning_rate": 8.304698490873847e-05, "loss": -0.9274, "step": 18400 }, { "epoch": 51.138888888888886, "grad_norm": 0.3255842924118042, "learning_rate": 8.30262994193603e-05, "loss": -0.9263, "step": 18410 }, { "epoch": 51.166666666666664, "grad_norm": 0.3867410123348236, "learning_rate": 8.300560389746438e-05, "loss": -0.9292, "step": 18420 }, { "epoch": 51.19444444444444, "grad_norm": 0.28810396790504456, "learning_rate": 8.298489834933745e-05, "loss": -0.9241, "step": 18430 }, { "epoch": 51.22222222222222, "grad_norm": 0.2979995012283325, "learning_rate": 8.296418278126934e-05, "loss": -0.9234, "step": 18440 }, { "epoch": 51.25, "grad_norm": 0.2602136433124542, "learning_rate": 8.294345719955284e-05, "loss": -0.927, "step": 18450 }, { "epoch": 51.27777777777778, "grad_norm": 0.26829975843429565, "learning_rate": 8.29227216104839e-05, "loss": -0.9267, "step": 18460 }, { "epoch": 51.30555555555556, "grad_norm": 0.38337117433547974, "learning_rate": 8.290197602036137e-05, "loss": -0.9241, "step": 18470 }, { "epoch": 51.333333333333336, "grad_norm": 0.3181128203868866, "learning_rate": 8.288122043548725e-05, "loss": -0.93, "step": 18480 }, { "epoch": 51.361111111111114, "grad_norm": 0.2652621567249298, "learning_rate": 8.286045486216657e-05, "loss": -0.9225, "step": 18490 }, { "epoch": 51.388888888888886, "grad_norm": 0.4685180187225342, "learning_rate": 8.283967930670733e-05, "loss": -0.9278, "step": 18500 }, { "epoch": 51.416666666666664, "grad_norm": 0.2250700443983078, "learning_rate": 8.281889377542058e-05, "loss": -0.9238, "step": 18510 }, { "epoch": 51.44444444444444, "grad_norm": 0.24106374382972717, "learning_rate": 8.279809827462045e-05, "loss": -0.9276, "step": 18520 }, { "epoch": 51.47222222222222, "grad_norm": 0.20782864093780518, "learning_rate": 8.277729281062402e-05, "loss": -0.9238, "step": 18530 }, { "epoch": 51.5, "grad_norm": 0.22525563836097717, "learning_rate": 8.27564773897515e-05, "loss": -0.9285, "step": 18540 }, { "epoch": 51.52777777777778, "grad_norm": 0.1809854805469513, "learning_rate": 8.273565201832602e-05, "loss": -0.931, "step": 18550 }, { "epoch": 51.55555555555556, "grad_norm": 0.19473648071289062, "learning_rate": 8.27148167026738e-05, "loss": -0.9314, "step": 18560 }, { "epoch": 51.583333333333336, "grad_norm": 0.2519212067127228, "learning_rate": 8.269397144912405e-05, "loss": -0.9288, "step": 18570 }, { "epoch": 51.611111111111114, "grad_norm": 0.17066359519958496, "learning_rate": 8.267311626400899e-05, "loss": -0.9318, "step": 18580 }, { "epoch": 51.638888888888886, "grad_norm": 0.2729027569293976, "learning_rate": 8.26522511536639e-05, "loss": -0.9231, "step": 18590 }, { "epoch": 51.666666666666664, "grad_norm": 0.2856464087963104, "learning_rate": 8.263137612442706e-05, "loss": -0.9313, "step": 18600 }, { "epoch": 51.69444444444444, "grad_norm": 0.27650532126426697, "learning_rate": 8.261049118263971e-05, "loss": -0.9313, "step": 18610 }, { "epoch": 51.72222222222222, "grad_norm": 0.20680877566337585, "learning_rate": 8.258959633464619e-05, "loss": -0.9287, "step": 18620 }, { "epoch": 51.75, "grad_norm": 0.22460432350635529, "learning_rate": 8.256869158679377e-05, "loss": -0.9283, "step": 18630 }, { "epoch": 51.77777777777778, "grad_norm": 0.23662880063056946, "learning_rate": 8.254777694543278e-05, "loss": -0.9272, "step": 18640 }, { "epoch": 51.80555555555556, "grad_norm": 0.26218706369400024, "learning_rate": 8.252685241691651e-05, "loss": -0.9299, "step": 18650 }, { "epoch": 51.833333333333336, "grad_norm": 0.3053811490535736, "learning_rate": 8.250591800760133e-05, "loss": -0.9215, "step": 18660 }, { "epoch": 51.861111111111114, "grad_norm": 0.24201275408267975, "learning_rate": 8.248497372384649e-05, "loss": -0.9262, "step": 18670 }, { "epoch": 51.888888888888886, "grad_norm": 0.3588934540748596, "learning_rate": 8.246401957201437e-05, "loss": -0.9282, "step": 18680 }, { "epoch": 51.916666666666664, "grad_norm": 0.32443007826805115, "learning_rate": 8.244305555847027e-05, "loss": -0.9302, "step": 18690 }, { "epoch": 51.94444444444444, "grad_norm": 0.1740061342716217, "learning_rate": 8.24220816895825e-05, "loss": -0.9322, "step": 18700 }, { "epoch": 51.97222222222222, "grad_norm": 0.24226170778274536, "learning_rate": 8.240109797172237e-05, "loss": -0.9276, "step": 18710 }, { "epoch": 52.0, "grad_norm": 0.2524919807910919, "learning_rate": 8.238010441126416e-05, "loss": -0.9265, "step": 18720 }, { "epoch": 52.02777777777778, "grad_norm": 0.3073617219924927, "learning_rate": 8.23591010145852e-05, "loss": -0.9224, "step": 18730 }, { "epoch": 52.05555555555556, "grad_norm": 0.24248215556144714, "learning_rate": 8.233808778806571e-05, "loss": -0.9251, "step": 18740 }, { "epoch": 52.083333333333336, "grad_norm": 0.30122026801109314, "learning_rate": 8.231706473808903e-05, "loss": -0.9302, "step": 18750 }, { "epoch": 52.111111111111114, "grad_norm": 0.2811446785926819, "learning_rate": 8.229603187104133e-05, "loss": -0.9225, "step": 18760 }, { "epoch": 52.138888888888886, "grad_norm": 0.31804558634757996, "learning_rate": 8.22749891933119e-05, "loss": -0.9264, "step": 18770 }, { "epoch": 52.166666666666664, "grad_norm": 0.29202553629875183, "learning_rate": 8.225393671129291e-05, "loss": -0.9287, "step": 18780 }, { "epoch": 52.19444444444444, "grad_norm": 0.18562178313732147, "learning_rate": 8.223287443137957e-05, "loss": -0.9286, "step": 18790 }, { "epoch": 52.22222222222222, "grad_norm": 0.175180584192276, "learning_rate": 8.221180235997004e-05, "loss": -0.9311, "step": 18800 }, { "epoch": 52.25, "grad_norm": 0.20933377742767334, "learning_rate": 8.219072050346544e-05, "loss": -0.9305, "step": 18810 }, { "epoch": 52.27777777777778, "grad_norm": 0.26551756262779236, "learning_rate": 8.216962886826992e-05, "loss": -0.9298, "step": 18820 }, { "epoch": 52.30555555555556, "grad_norm": 0.36722907423973083, "learning_rate": 8.214852746079054e-05, "loss": -0.9269, "step": 18830 }, { "epoch": 52.333333333333336, "grad_norm": 0.19565162062644958, "learning_rate": 8.212741628743732e-05, "loss": -0.9313, "step": 18840 }, { "epoch": 52.361111111111114, "grad_norm": 0.2617698609828949, "learning_rate": 8.210629535462333e-05, "loss": -0.9288, "step": 18850 }, { "epoch": 52.388888888888886, "grad_norm": 0.2536715567111969, "learning_rate": 8.208516466876453e-05, "loss": -0.9328, "step": 18860 }, { "epoch": 52.416666666666664, "grad_norm": 0.3239142596721649, "learning_rate": 8.206402423627986e-05, "loss": -0.9349, "step": 18870 }, { "epoch": 52.44444444444444, "grad_norm": 0.3087070882320404, "learning_rate": 8.204287406359124e-05, "loss": -0.9156, "step": 18880 }, { "epoch": 52.47222222222222, "grad_norm": 0.3915445804595947, "learning_rate": 8.20217141571235e-05, "loss": -0.9287, "step": 18890 }, { "epoch": 52.5, "grad_norm": 0.22285793721675873, "learning_rate": 8.200054452330449e-05, "loss": -0.9284, "step": 18900 }, { "epoch": 52.52777777777778, "grad_norm": 0.2796308398246765, "learning_rate": 8.197936516856499e-05, "loss": -0.9228, "step": 18910 }, { "epoch": 52.55555555555556, "grad_norm": 0.2136061191558838, "learning_rate": 8.195817609933871e-05, "loss": -0.9288, "step": 18920 }, { "epoch": 52.583333333333336, "grad_norm": 0.2734050452709198, "learning_rate": 8.193697732206233e-05, "loss": -0.9274, "step": 18930 }, { "epoch": 52.611111111111114, "grad_norm": 0.2237454652786255, "learning_rate": 8.19157688431755e-05, "loss": -0.9259, "step": 18940 }, { "epoch": 52.638888888888886, "grad_norm": 0.30390995740890503, "learning_rate": 8.189455066912077e-05, "loss": -0.931, "step": 18950 }, { "epoch": 52.666666666666664, "grad_norm": 0.20408426225185394, "learning_rate": 8.187332280634369e-05, "loss": -0.9309, "step": 18960 }, { "epoch": 52.69444444444444, "grad_norm": 0.245172917842865, "learning_rate": 8.18520852612927e-05, "loss": -0.9282, "step": 18970 }, { "epoch": 52.72222222222222, "grad_norm": 0.19634851813316345, "learning_rate": 8.183083804041921e-05, "loss": -0.9242, "step": 18980 }, { "epoch": 52.75, "grad_norm": 0.1981768012046814, "learning_rate": 8.180958115017757e-05, "loss": -0.9293, "step": 18990 }, { "epoch": 52.77777777777778, "grad_norm": 0.28867536783218384, "learning_rate": 8.178831459702505e-05, "loss": -0.9235, "step": 19000 }, { "epoch": 52.80555555555556, "grad_norm": 0.2713887691497803, "learning_rate": 8.17670383874219e-05, "loss": -0.9281, "step": 19010 }, { "epoch": 52.833333333333336, "grad_norm": 0.20129822194576263, "learning_rate": 8.174575252783124e-05, "loss": -0.9287, "step": 19020 }, { "epoch": 52.861111111111114, "grad_norm": 0.2666698396205902, "learning_rate": 8.172445702471914e-05, "loss": -0.9227, "step": 19030 }, { "epoch": 52.888888888888886, "grad_norm": 0.30339315533638, "learning_rate": 8.170315188455466e-05, "loss": -0.9295, "step": 19040 }, { "epoch": 52.916666666666664, "grad_norm": 0.27939966320991516, "learning_rate": 8.168183711380969e-05, "loss": -0.9291, "step": 19050 }, { "epoch": 52.94444444444444, "grad_norm": 0.35389429330825806, "learning_rate": 8.166051271895913e-05, "loss": -0.93, "step": 19060 }, { "epoch": 52.97222222222222, "grad_norm": 0.19156979024410248, "learning_rate": 8.163917870648075e-05, "loss": -0.9231, "step": 19070 }, { "epoch": 53.0, "grad_norm": 0.21537216007709503, "learning_rate": 8.161783508285526e-05, "loss": -0.9282, "step": 19080 }, { "epoch": 53.02777777777778, "grad_norm": 0.33997049927711487, "learning_rate": 8.159648185456628e-05, "loss": -0.9271, "step": 19090 }, { "epoch": 53.05555555555556, "grad_norm": 0.31960076093673706, "learning_rate": 8.157511902810038e-05, "loss": -0.9237, "step": 19100 }, { "epoch": 53.083333333333336, "grad_norm": 0.3041817545890808, "learning_rate": 8.155374660994701e-05, "loss": -0.9299, "step": 19110 }, { "epoch": 53.111111111111114, "grad_norm": 0.28550687432289124, "learning_rate": 8.153236460659857e-05, "loss": -0.921, "step": 19120 }, { "epoch": 53.138888888888886, "grad_norm": 0.31145110726356506, "learning_rate": 8.151097302455031e-05, "loss": -0.9311, "step": 19130 }, { "epoch": 53.166666666666664, "grad_norm": 0.26477330923080444, "learning_rate": 8.148957187030044e-05, "loss": -0.9251, "step": 19140 }, { "epoch": 53.19444444444444, "grad_norm": 0.25161227583885193, "learning_rate": 8.146816115035006e-05, "loss": -0.9278, "step": 19150 }, { "epoch": 53.22222222222222, "grad_norm": 0.2685237228870392, "learning_rate": 8.14467408712032e-05, "loss": -0.9239, "step": 19160 }, { "epoch": 53.25, "grad_norm": 0.2291652411222458, "learning_rate": 8.142531103936678e-05, "loss": -0.9286, "step": 19170 }, { "epoch": 53.27777777777778, "grad_norm": 0.20274388790130615, "learning_rate": 8.14038716613506e-05, "loss": -0.9336, "step": 19180 }, { "epoch": 53.30555555555556, "grad_norm": 0.1743096262216568, "learning_rate": 8.138242274366736e-05, "loss": -0.9288, "step": 19190 }, { "epoch": 53.333333333333336, "grad_norm": 0.3476232886314392, "learning_rate": 8.136096429283271e-05, "loss": -0.9267, "step": 19200 }, { "epoch": 53.361111111111114, "grad_norm": 0.3955804407596588, "learning_rate": 8.133949631536515e-05, "loss": -0.9247, "step": 19210 }, { "epoch": 53.388888888888886, "grad_norm": 0.31840622425079346, "learning_rate": 8.131801881778607e-05, "loss": -0.9255, "step": 19220 }, { "epoch": 53.416666666666664, "grad_norm": 0.2718217074871063, "learning_rate": 8.129653180661978e-05, "loss": -0.9277, "step": 19230 }, { "epoch": 53.44444444444444, "grad_norm": 0.1934608668088913, "learning_rate": 8.127503528839346e-05, "loss": -0.9215, "step": 19240 }, { "epoch": 53.47222222222222, "grad_norm": 0.25919127464294434, "learning_rate": 8.125352926963721e-05, "loss": -0.9226, "step": 19250 }, { "epoch": 53.5, "grad_norm": 0.262900710105896, "learning_rate": 8.123201375688395e-05, "loss": -0.9279, "step": 19260 }, { "epoch": 53.52777777777778, "grad_norm": 0.234915092587471, "learning_rate": 8.121048875666954e-05, "loss": -0.928, "step": 19270 }, { "epoch": 53.55555555555556, "grad_norm": 0.21214887499809265, "learning_rate": 8.118895427553274e-05, "loss": -0.9329, "step": 19280 }, { "epoch": 53.583333333333336, "grad_norm": 0.24737632274627686, "learning_rate": 8.116741032001511e-05, "loss": -0.9289, "step": 19290 }, { "epoch": 53.611111111111114, "grad_norm": 0.28007665276527405, "learning_rate": 8.114585689666114e-05, "loss": -0.9303, "step": 19300 }, { "epoch": 53.638888888888886, "grad_norm": 0.37387800216674805, "learning_rate": 8.112429401201821e-05, "loss": -0.9318, "step": 19310 }, { "epoch": 53.666666666666664, "grad_norm": 0.2719254195690155, "learning_rate": 8.110272167263656e-05, "loss": -0.9294, "step": 19320 }, { "epoch": 53.69444444444444, "grad_norm": 0.23895223438739777, "learning_rate": 8.108113988506929e-05, "loss": -0.9295, "step": 19330 }, { "epoch": 53.72222222222222, "grad_norm": 0.22390739619731903, "learning_rate": 8.105954865587235e-05, "loss": -0.9277, "step": 19340 }, { "epoch": 53.75, "grad_norm": 0.2892605662345886, "learning_rate": 8.103794799160463e-05, "loss": -0.9306, "step": 19350 }, { "epoch": 53.77777777777778, "grad_norm": 0.2823115885257721, "learning_rate": 8.101633789882781e-05, "loss": -0.9261, "step": 19360 }, { "epoch": 53.80555555555556, "grad_norm": 0.25385063886642456, "learning_rate": 8.099471838410648e-05, "loss": -0.9321, "step": 19370 }, { "epoch": 53.833333333333336, "grad_norm": 0.2041885107755661, "learning_rate": 8.097308945400806e-05, "loss": -0.9294, "step": 19380 }, { "epoch": 53.861111111111114, "grad_norm": 0.28316229581832886, "learning_rate": 8.095145111510288e-05, "loss": -0.9284, "step": 19390 }, { "epoch": 53.888888888888886, "grad_norm": 0.14962337911128998, "learning_rate": 8.092980337396406e-05, "loss": -0.9338, "step": 19400 }, { "epoch": 53.916666666666664, "grad_norm": 0.31711050868034363, "learning_rate": 8.090814623716763e-05, "loss": -0.9262, "step": 19410 }, { "epoch": 53.94444444444444, "grad_norm": 0.15670174360275269, "learning_rate": 8.088647971129246e-05, "loss": -0.9286, "step": 19420 }, { "epoch": 53.97222222222222, "grad_norm": 0.2579851448535919, "learning_rate": 8.086480380292026e-05, "loss": -0.9262, "step": 19430 }, { "epoch": 54.0, "grad_norm": 0.32167643308639526, "learning_rate": 8.084311851863562e-05, "loss": -0.9295, "step": 19440 }, { "epoch": 54.02777777777778, "grad_norm": 0.28364452719688416, "learning_rate": 8.082142386502591e-05, "loss": -0.9283, "step": 19450 }, { "epoch": 54.05555555555556, "grad_norm": 0.31260964274406433, "learning_rate": 8.079971984868145e-05, "loss": -0.9232, "step": 19460 }, { "epoch": 54.083333333333336, "grad_norm": 0.22724099457263947, "learning_rate": 8.077800647619532e-05, "loss": -0.9299, "step": 19470 }, { "epoch": 54.111111111111114, "grad_norm": 0.3783111870288849, "learning_rate": 8.075628375416345e-05, "loss": -0.9301, "step": 19480 }, { "epoch": 54.138888888888886, "grad_norm": 0.2695923447608948, "learning_rate": 8.073455168918464e-05, "loss": -0.9274, "step": 19490 }, { "epoch": 54.166666666666664, "grad_norm": 0.28777170181274414, "learning_rate": 8.071281028786055e-05, "loss": -0.9309, "step": 19500 }, { "epoch": 54.19444444444444, "grad_norm": 0.3367520570755005, "learning_rate": 8.069105955679562e-05, "loss": -0.9321, "step": 19510 }, { "epoch": 54.22222222222222, "grad_norm": 0.20988047122955322, "learning_rate": 8.066929950259713e-05, "loss": -0.9293, "step": 19520 }, { "epoch": 54.25, "grad_norm": 0.23608635365962982, "learning_rate": 8.064753013187522e-05, "loss": -0.9289, "step": 19530 }, { "epoch": 54.27777777777778, "grad_norm": 0.20542855560779572, "learning_rate": 8.062575145124289e-05, "loss": -0.9329, "step": 19540 }, { "epoch": 54.30555555555556, "grad_norm": 0.2507700026035309, "learning_rate": 8.060396346731587e-05, "loss": -0.926, "step": 19550 }, { "epoch": 54.333333333333336, "grad_norm": 0.29900529980659485, "learning_rate": 8.058216618671281e-05, "loss": -0.9307, "step": 19560 }, { "epoch": 54.361111111111114, "grad_norm": 0.27493762969970703, "learning_rate": 8.056035961605514e-05, "loss": -0.9338, "step": 19570 }, { "epoch": 54.388888888888886, "grad_norm": 0.3389262855052948, "learning_rate": 8.05385437619671e-05, "loss": -0.9222, "step": 19580 }, { "epoch": 54.416666666666664, "grad_norm": 0.1920335739850998, "learning_rate": 8.05167186310758e-05, "loss": -0.9332, "step": 19590 }, { "epoch": 54.44444444444444, "grad_norm": 0.2852037847042084, "learning_rate": 8.049488423001113e-05, "loss": -0.93, "step": 19600 }, { "epoch": 54.47222222222222, "grad_norm": 0.21544529497623444, "learning_rate": 8.047304056540581e-05, "loss": -0.9288, "step": 19610 }, { "epoch": 54.5, "grad_norm": 0.36582162976264954, "learning_rate": 8.045118764389534e-05, "loss": -0.9292, "step": 19620 }, { "epoch": 54.52777777777778, "grad_norm": 0.20115746557712555, "learning_rate": 8.042932547211809e-05, "loss": -0.9336, "step": 19630 }, { "epoch": 54.55555555555556, "grad_norm": 0.2550508975982666, "learning_rate": 8.04074540567152e-05, "loss": -0.9352, "step": 19640 }, { "epoch": 54.583333333333336, "grad_norm": 0.23136156797409058, "learning_rate": 8.038557340433063e-05, "loss": -0.9227, "step": 19650 }, { "epoch": 54.611111111111114, "grad_norm": 0.2663789391517639, "learning_rate": 8.036368352161115e-05, "loss": -0.931, "step": 19660 }, { "epoch": 54.638888888888886, "grad_norm": 0.18590182065963745, "learning_rate": 8.034178441520633e-05, "loss": -0.9315, "step": 19670 }, { "epoch": 54.666666666666664, "grad_norm": 0.22093647718429565, "learning_rate": 8.031987609176852e-05, "loss": -0.9311, "step": 19680 }, { "epoch": 54.69444444444444, "grad_norm": 0.26555976271629333, "learning_rate": 8.02979585579529e-05, "loss": -0.9294, "step": 19690 }, { "epoch": 54.72222222222222, "grad_norm": 0.18595227599143982, "learning_rate": 8.027603182041745e-05, "loss": -0.9243, "step": 19700 }, { "epoch": 54.75, "grad_norm": 0.22542454302310944, "learning_rate": 8.025409588582292e-05, "loss": -0.9268, "step": 19710 }, { "epoch": 54.77777777777778, "grad_norm": 0.3552703857421875, "learning_rate": 8.023215076083288e-05, "loss": -0.9274, "step": 19720 }, { "epoch": 54.80555555555556, "grad_norm": 0.4019477665424347, "learning_rate": 8.021019645211367e-05, "loss": -0.9343, "step": 19730 }, { "epoch": 54.833333333333336, "grad_norm": 0.26406601071357727, "learning_rate": 8.018823296633441e-05, "loss": -0.9304, "step": 19740 }, { "epoch": 54.861111111111114, "grad_norm": 0.22312355041503906, "learning_rate": 8.016626031016708e-05, "loss": -0.9231, "step": 19750 }, { "epoch": 54.888888888888886, "grad_norm": 0.19096523523330688, "learning_rate": 8.014427849028636e-05, "loss": -0.9343, "step": 19760 }, { "epoch": 54.916666666666664, "grad_norm": 0.32752397656440735, "learning_rate": 8.012228751336974e-05, "loss": -0.9287, "step": 19770 }, { "epoch": 54.94444444444444, "grad_norm": 0.30542922019958496, "learning_rate": 8.01002873860975e-05, "loss": -0.9199, "step": 19780 }, { "epoch": 54.97222222222222, "grad_norm": 0.2759529948234558, "learning_rate": 8.00782781151527e-05, "loss": -0.9337, "step": 19790 }, { "epoch": 55.0, "grad_norm": 0.33866235613822937, "learning_rate": 8.005625970722119e-05, "loss": -0.9217, "step": 19800 }, { "epoch": 55.02777777777778, "grad_norm": 0.2012053281068802, "learning_rate": 8.003423216899158e-05, "loss": -0.934, "step": 19810 }, { "epoch": 55.05555555555556, "grad_norm": 0.24997374415397644, "learning_rate": 8.001219550715522e-05, "loss": -0.9264, "step": 19820 }, { "epoch": 55.083333333333336, "grad_norm": 0.20483052730560303, "learning_rate": 7.999014972840632e-05, "loss": -0.9247, "step": 19830 }, { "epoch": 55.111111111111114, "grad_norm": 0.27314871549606323, "learning_rate": 7.996809483944174e-05, "loss": -0.9332, "step": 19840 }, { "epoch": 55.138888888888886, "grad_norm": 0.23275864124298096, "learning_rate": 7.994603084696124e-05, "loss": -0.9332, "step": 19850 }, { "epoch": 55.166666666666664, "grad_norm": 0.1740139126777649, "learning_rate": 7.992395775766724e-05, "loss": -0.9352, "step": 19860 }, { "epoch": 55.19444444444444, "grad_norm": 0.18091215193271637, "learning_rate": 7.990187557826497e-05, "loss": -0.9217, "step": 19870 }, { "epoch": 55.22222222222222, "grad_norm": 0.292536199092865, "learning_rate": 7.987978431546242e-05, "loss": -0.9298, "step": 19880 }, { "epoch": 55.25, "grad_norm": 0.23134984076023102, "learning_rate": 7.985768397597031e-05, "loss": -0.9263, "step": 19890 }, { "epoch": 55.27777777777778, "grad_norm": 0.3546929657459259, "learning_rate": 7.983557456650216e-05, "loss": -0.9234, "step": 19900 }, { "epoch": 55.30555555555556, "grad_norm": 0.3900480270385742, "learning_rate": 7.981345609377422e-05, "loss": -0.9252, "step": 19910 }, { "epoch": 55.333333333333336, "grad_norm": 0.217746302485466, "learning_rate": 7.97913285645055e-05, "loss": -0.9313, "step": 19920 }, { "epoch": 55.361111111111114, "grad_norm": 0.33384978771209717, "learning_rate": 7.976919198541776e-05, "loss": -0.9307, "step": 19930 }, { "epoch": 55.388888888888886, "grad_norm": 0.33057084679603577, "learning_rate": 7.974704636323548e-05, "loss": -0.9277, "step": 19940 }, { "epoch": 55.416666666666664, "grad_norm": 0.1995295137166977, "learning_rate": 7.972489170468597e-05, "loss": -0.9339, "step": 19950 }, { "epoch": 55.44444444444444, "grad_norm": 0.24730521440505981, "learning_rate": 7.970272801649918e-05, "loss": -0.9256, "step": 19960 }, { "epoch": 55.47222222222222, "grad_norm": 0.21971429884433746, "learning_rate": 7.96805553054079e-05, "loss": -0.9273, "step": 19970 }, { "epoch": 55.5, "grad_norm": 0.2690114974975586, "learning_rate": 7.965837357814756e-05, "loss": -0.9302, "step": 19980 }, { "epoch": 55.52777777777778, "grad_norm": 0.25520768761634827, "learning_rate": 7.963618284145643e-05, "loss": -0.9242, "step": 19990 }, { "epoch": 55.55555555555556, "grad_norm": 0.2960836589336395, "learning_rate": 7.961398310207544e-05, "loss": -0.9297, "step": 20000 } ], "logging_steps": 10, "max_steps": 60000, "num_input_tokens_seen": 0, "num_train_epochs": 167, "save_steps": 10000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 16, "trial_name": null, "trial_params": null }