Grogros's picture
Training in progress, step 4000, checkpoint
d5eb21c verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 4000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0025,
"grad_norm": 276.0,
"learning_rate": 1.1250000000000001e-07,
"loss": 0.4619,
"step": 10
},
{
"epoch": 0.005,
"grad_norm": 236.0,
"learning_rate": 2.3750000000000003e-07,
"loss": 0.457,
"step": 20
},
{
"epoch": 0.0075,
"grad_norm": 276.0,
"learning_rate": 3.625e-07,
"loss": 0.51,
"step": 30
},
{
"epoch": 0.01,
"grad_norm": 194.0,
"learning_rate": 4.875000000000001e-07,
"loss": 0.4631,
"step": 40
},
{
"epoch": 0.0125,
"grad_norm": 264.0,
"learning_rate": 6.125000000000001e-07,
"loss": 0.5266,
"step": 50
},
{
"epoch": 0.015,
"grad_norm": 21.75,
"learning_rate": 7.375e-07,
"loss": 0.5422,
"step": 60
},
{
"epoch": 0.0175,
"grad_norm": 21.875,
"learning_rate": 8.625e-07,
"loss": 0.5314,
"step": 70
},
{
"epoch": 0.02,
"grad_norm": 170.0,
"learning_rate": 9.875e-07,
"loss": 0.4754,
"step": 80
},
{
"epoch": 0.0225,
"grad_norm": 5.625,
"learning_rate": 1.1125000000000001e-06,
"loss": 0.5029,
"step": 90
},
{
"epoch": 0.025,
"grad_norm": 278.0,
"learning_rate": 1.2375e-06,
"loss": 0.4709,
"step": 100
},
{
"epoch": 0.0275,
"grad_norm": 2.546875,
"learning_rate": 1.3625000000000003e-06,
"loss": 0.3525,
"step": 110
},
{
"epoch": 0.03,
"grad_norm": 74.0,
"learning_rate": 1.4875000000000002e-06,
"loss": 0.3973,
"step": 120
},
{
"epoch": 0.0325,
"grad_norm": 2.4375,
"learning_rate": 1.6125000000000002e-06,
"loss": 0.3656,
"step": 130
},
{
"epoch": 0.035,
"grad_norm": 232.0,
"learning_rate": 1.7375e-06,
"loss": 0.3996,
"step": 140
},
{
"epoch": 0.0375,
"grad_norm": 4.09375,
"learning_rate": 1.8625000000000002e-06,
"loss": 0.3652,
"step": 150
},
{
"epoch": 0.04,
"grad_norm": 12.875,
"learning_rate": 1.9875000000000005e-06,
"loss": 0.4156,
"step": 160
},
{
"epoch": 0.0425,
"grad_norm": 22.625,
"learning_rate": 2.1125e-06,
"loss": 0.3318,
"step": 170
},
{
"epoch": 0.045,
"grad_norm": 12.75,
"learning_rate": 2.2375e-06,
"loss": 0.3256,
"step": 180
},
{
"epoch": 0.0475,
"grad_norm": 5.09375,
"learning_rate": 2.3625000000000003e-06,
"loss": 0.3281,
"step": 190
},
{
"epoch": 0.05,
"grad_norm": 25.875,
"learning_rate": 2.4875000000000003e-06,
"loss": 0.3357,
"step": 200
},
{
"epoch": 0.0525,
"grad_norm": 2.703125,
"learning_rate": 2.6125e-06,
"loss": 0.3479,
"step": 210
},
{
"epoch": 0.055,
"grad_norm": 2.765625,
"learning_rate": 2.7375e-06,
"loss": 0.3328,
"step": 220
},
{
"epoch": 0.0575,
"grad_norm": 2.0,
"learning_rate": 2.8625e-06,
"loss": 0.3164,
"step": 230
},
{
"epoch": 0.06,
"grad_norm": 2.34375,
"learning_rate": 2.9875e-06,
"loss": 0.299,
"step": 240
},
{
"epoch": 0.0625,
"grad_norm": 2.84375,
"learning_rate": 3.1125000000000007e-06,
"loss": 0.3129,
"step": 250
},
{
"epoch": 0.065,
"grad_norm": 2.671875,
"learning_rate": 3.2375e-06,
"loss": 0.3131,
"step": 260
},
{
"epoch": 0.0675,
"grad_norm": 4.09375,
"learning_rate": 3.3625000000000004e-06,
"loss": 0.3053,
"step": 270
},
{
"epoch": 0.07,
"grad_norm": 1.9375,
"learning_rate": 3.4875000000000005e-06,
"loss": 0.299,
"step": 280
},
{
"epoch": 0.0725,
"grad_norm": 2.6875,
"learning_rate": 3.6125000000000006e-06,
"loss": 0.3273,
"step": 290
},
{
"epoch": 0.075,
"grad_norm": 1.90625,
"learning_rate": 3.7375000000000006e-06,
"loss": 0.3041,
"step": 300
},
{
"epoch": 0.0775,
"grad_norm": 1.640625,
"learning_rate": 3.8625e-06,
"loss": 0.3057,
"step": 310
},
{
"epoch": 0.08,
"grad_norm": 2.375,
"learning_rate": 3.9875e-06,
"loss": 0.315,
"step": 320
},
{
"epoch": 0.0825,
"grad_norm": 2.015625,
"learning_rate": 4.1125e-06,
"loss": 0.3063,
"step": 330
},
{
"epoch": 0.085,
"grad_norm": 1.96875,
"learning_rate": 4.2375000000000005e-06,
"loss": 0.3023,
"step": 340
},
{
"epoch": 0.0875,
"grad_norm": 2.53125,
"learning_rate": 4.362500000000001e-06,
"loss": 0.2834,
"step": 350
},
{
"epoch": 0.09,
"grad_norm": 2.0,
"learning_rate": 4.4875e-06,
"loss": 0.2767,
"step": 360
},
{
"epoch": 0.0925,
"grad_norm": 1.6015625,
"learning_rate": 4.6125e-06,
"loss": 0.2859,
"step": 370
},
{
"epoch": 0.095,
"grad_norm": 2.375,
"learning_rate": 4.737500000000001e-06,
"loss": 0.2889,
"step": 380
},
{
"epoch": 0.0975,
"grad_norm": 4.75,
"learning_rate": 4.8625000000000005e-06,
"loss": 0.2916,
"step": 390
},
{
"epoch": 0.1,
"grad_norm": 1.671875,
"learning_rate": 4.987500000000001e-06,
"loss": 0.2722,
"step": 400
},
{
"epoch": 0.1025,
"grad_norm": 2.46875,
"learning_rate": 4.999922894111975e-06,
"loss": 0.2822,
"step": 410
},
{
"epoch": 0.105,
"grad_norm": 2.515625,
"learning_rate": 4.999656361346094e-06,
"loss": 0.269,
"step": 420
},
{
"epoch": 0.1075,
"grad_norm": 1.6875,
"learning_rate": 4.999199470070484e-06,
"loss": 0.2836,
"step": 430
},
{
"epoch": 0.11,
"grad_norm": 1.671875,
"learning_rate": 4.998552255079182e-06,
"loss": 0.2577,
"step": 440
},
{
"epoch": 0.1125,
"grad_norm": 1.5234375,
"learning_rate": 4.9977147656601196e-06,
"loss": 0.266,
"step": 450
},
{
"epoch": 0.115,
"grad_norm": 1.890625,
"learning_rate": 4.996687065591355e-06,
"loss": 0.2792,
"step": 460
},
{
"epoch": 0.1175,
"grad_norm": 2.484375,
"learning_rate": 4.9954692331362295e-06,
"loss": 0.2801,
"step": 470
},
{
"epoch": 0.12,
"grad_norm": 1.828125,
"learning_rate": 4.9940613610373974e-06,
"loss": 0.26,
"step": 480
},
{
"epoch": 0.1225,
"grad_norm": 1.9296875,
"learning_rate": 4.992463556509772e-06,
"loss": 0.2613,
"step": 490
},
{
"epoch": 0.125,
"grad_norm": 1.828125,
"learning_rate": 4.990675941232353e-06,
"loss": 0.2385,
"step": 500
},
{
"epoch": 0.1275,
"grad_norm": 1.90625,
"learning_rate": 4.988698651338965e-06,
"loss": 0.2574,
"step": 510
},
{
"epoch": 0.13,
"grad_norm": 1.5,
"learning_rate": 4.986531837407891e-06,
"loss": 0.2559,
"step": 520
},
{
"epoch": 0.1325,
"grad_norm": 1.484375,
"learning_rate": 4.9841756644503965e-06,
"loss": 0.2622,
"step": 530
},
{
"epoch": 0.135,
"grad_norm": 1.15625,
"learning_rate": 4.981630311898178e-06,
"loss": 0.2537,
"step": 540
},
{
"epoch": 0.1375,
"grad_norm": 1.6484375,
"learning_rate": 4.978895973589686e-06,
"loss": 0.2832,
"step": 550
},
{
"epoch": 0.14,
"grad_norm": 1.6796875,
"learning_rate": 4.975972857755369e-06,
"loss": 0.2471,
"step": 560
},
{
"epoch": 0.1425,
"grad_norm": 1.4296875,
"learning_rate": 4.972861187001815e-06,
"loss": 0.2648,
"step": 570
},
{
"epoch": 0.145,
"grad_norm": 1.6171875,
"learning_rate": 4.9695611982947995e-06,
"loss": 0.2697,
"step": 580
},
{
"epoch": 0.1475,
"grad_norm": 1.8515625,
"learning_rate": 4.966073142941239e-06,
"loss": 0.2592,
"step": 590
},
{
"epoch": 0.15,
"grad_norm": 1.4140625,
"learning_rate": 4.962397286570053e-06,
"loss": 0.267,
"step": 600
},
{
"epoch": 0.1525,
"grad_norm": 1.515625,
"learning_rate": 4.958533909111936e-06,
"loss": 0.253,
"step": 610
},
{
"epoch": 0.155,
"grad_norm": 1.390625,
"learning_rate": 4.95448330477804e-06,
"loss": 0.2488,
"step": 620
},
{
"epoch": 0.1575,
"grad_norm": 1.71875,
"learning_rate": 4.950245782037566e-06,
"loss": 0.2552,
"step": 630
},
{
"epoch": 0.16,
"grad_norm": 1.2890625,
"learning_rate": 4.945821663594277e-06,
"loss": 0.2485,
"step": 640
},
{
"epoch": 0.1625,
"grad_norm": 1.5078125,
"learning_rate": 4.941211286361922e-06,
"loss": 0.2612,
"step": 650
},
{
"epoch": 0.165,
"grad_norm": 1.8984375,
"learning_rate": 4.936415001438577e-06,
"loss": 0.2465,
"step": 660
},
{
"epoch": 0.1675,
"grad_norm": 1.609375,
"learning_rate": 4.9314331740799084e-06,
"loss": 0.2449,
"step": 670
},
{
"epoch": 0.17,
"grad_norm": 1.234375,
"learning_rate": 4.926266183671356e-06,
"loss": 0.2431,
"step": 680
},
{
"epoch": 0.1725,
"grad_norm": 1.359375,
"learning_rate": 4.920914423699247e-06,
"loss": 0.2479,
"step": 690
},
{
"epoch": 0.175,
"grad_norm": 1.28125,
"learning_rate": 4.915378301720822e-06,
"loss": 0.2327,
"step": 700
},
{
"epoch": 0.1775,
"grad_norm": 1.3984375,
"learning_rate": 4.909658239333203e-06,
"loss": 0.2434,
"step": 710
},
{
"epoch": 0.18,
"grad_norm": 1.4296875,
"learning_rate": 4.903754672141288e-06,
"loss": 0.2437,
"step": 720
},
{
"epoch": 0.1825,
"grad_norm": 1.6953125,
"learning_rate": 4.897668049724574e-06,
"loss": 0.248,
"step": 730
},
{
"epoch": 0.185,
"grad_norm": 1.3515625,
"learning_rate": 4.891398835602925e-06,
"loss": 0.2502,
"step": 740
},
{
"epoch": 0.1875,
"grad_norm": 1.5546875,
"learning_rate": 4.884947507201268e-06,
"loss": 0.2368,
"step": 750
},
{
"epoch": 0.19,
"grad_norm": 1.7421875,
"learning_rate": 4.878314555813237e-06,
"loss": 0.2262,
"step": 760
},
{
"epoch": 0.1925,
"grad_norm": 1.6328125,
"learning_rate": 4.8715004865637616e-06,
"loss": 0.2392,
"step": 770
},
{
"epoch": 0.195,
"grad_norm": 2.015625,
"learning_rate": 4.8645058183705976e-06,
"loss": 0.2403,
"step": 780
},
{
"epoch": 0.1975,
"grad_norm": 1.2109375,
"learning_rate": 4.8573310839048085e-06,
"loss": 0.2466,
"step": 790
},
{
"epoch": 0.2,
"grad_norm": 1.3125,
"learning_rate": 4.8499768295502e-06,
"loss": 0.2453,
"step": 800
},
{
"epoch": 0.2025,
"grad_norm": 2.03125,
"learning_rate": 4.842443615361718e-06,
"loss": 0.2566,
"step": 810
},
{
"epoch": 0.205,
"grad_norm": 1.9296875,
"learning_rate": 4.834732015022786e-06,
"loss": 0.2248,
"step": 820
},
{
"epoch": 0.2075,
"grad_norm": 1.59375,
"learning_rate": 4.826842615801628e-06,
"loss": 0.2456,
"step": 830
},
{
"epoch": 0.21,
"grad_norm": 1.484375,
"learning_rate": 4.818776018506538e-06,
"loss": 0.2445,
"step": 840
},
{
"epoch": 0.2125,
"grad_norm": 1.5625,
"learning_rate": 4.810532837440134e-06,
"loss": 0.2451,
"step": 850
},
{
"epoch": 0.215,
"grad_norm": 1.5390625,
"learning_rate": 4.802113700352567e-06,
"loss": 0.2605,
"step": 860
},
{
"epoch": 0.2175,
"grad_norm": 1.2734375,
"learning_rate": 4.793519248393721e-06,
"loss": 0.2394,
"step": 870
},
{
"epoch": 0.22,
"grad_norm": 1.203125,
"learning_rate": 4.78475013606439e-06,
"loss": 0.238,
"step": 880
},
{
"epoch": 0.2225,
"grad_norm": 1.3671875,
"learning_rate": 4.775807031166428e-06,
"loss": 0.236,
"step": 890
},
{
"epoch": 0.225,
"grad_norm": 1.7578125,
"learning_rate": 4.766690614751897e-06,
"loss": 0.2357,
"step": 900
},
{
"epoch": 0.2275,
"grad_norm": 1.3359375,
"learning_rate": 4.757401581071203e-06,
"loss": 0.2436,
"step": 910
},
{
"epoch": 0.23,
"grad_norm": 3.671875,
"learning_rate": 4.747940637520226e-06,
"loss": 0.2309,
"step": 920
},
{
"epoch": 0.2325,
"grad_norm": 1.40625,
"learning_rate": 4.738308504586445e-06,
"loss": 0.2254,
"step": 930
},
{
"epoch": 0.235,
"grad_norm": 1.359375,
"learning_rate": 4.7285059157940765e-06,
"loss": 0.2221,
"step": 940
},
{
"epoch": 0.2375,
"grad_norm": 1.1640625,
"learning_rate": 4.718533617648209e-06,
"loss": 0.2246,
"step": 950
},
{
"epoch": 0.24,
"grad_norm": 1.1171875,
"learning_rate": 4.7083923695779546e-06,
"loss": 0.2228,
"step": 960
},
{
"epoch": 0.2425,
"grad_norm": 1.7109375,
"learning_rate": 4.6980829438786176e-06,
"loss": 0.2198,
"step": 970
},
{
"epoch": 0.245,
"grad_norm": 1.9453125,
"learning_rate": 4.687606125652882e-06,
"loss": 0.223,
"step": 980
},
{
"epoch": 0.2475,
"grad_norm": 1.5,
"learning_rate": 4.676962712751015e-06,
"loss": 0.2337,
"step": 990
},
{
"epoch": 0.25,
"grad_norm": 1.359375,
"learning_rate": 4.666153515710118e-06,
"loss": 0.2273,
"step": 1000
},
{
"epoch": 0.2525,
"grad_norm": 1.2109375,
"learning_rate": 4.655179357692396e-06,
"loss": 0.2134,
"step": 1010
},
{
"epoch": 0.255,
"grad_norm": 1.171875,
"learning_rate": 4.644041074422469e-06,
"loss": 0.2078,
"step": 1020
},
{
"epoch": 0.2575,
"grad_norm": 1.484375,
"learning_rate": 4.632739514123733e-06,
"loss": 0.2229,
"step": 1030
},
{
"epoch": 0.26,
"grad_norm": 1.21875,
"learning_rate": 4.6212755374537596e-06,
"loss": 0.2166,
"step": 1040
},
{
"epoch": 0.2625,
"grad_norm": 1.3359375,
"learning_rate": 4.609650017438757e-06,
"loss": 0.2222,
"step": 1050
},
{
"epoch": 0.265,
"grad_norm": 1.640625,
"learning_rate": 4.5978638394070835e-06,
"loss": 0.2257,
"step": 1060
},
{
"epoch": 0.2675,
"grad_norm": 1.3125,
"learning_rate": 4.58591790092183e-06,
"loss": 0.2269,
"step": 1070
},
{
"epoch": 0.27,
"grad_norm": 1.140625,
"learning_rate": 4.5738131117124605e-06,
"loss": 0.2107,
"step": 1080
},
{
"epoch": 0.2725,
"grad_norm": 1.296875,
"learning_rate": 4.561550393605541e-06,
"loss": 0.2285,
"step": 1090
},
{
"epoch": 0.275,
"grad_norm": 1.2265625,
"learning_rate": 4.549130680454532e-06,
"loss": 0.219,
"step": 1100
},
{
"epoch": 0.2775,
"grad_norm": 1.328125,
"learning_rate": 4.536554918068673e-06,
"loss": 0.2203,
"step": 1110
},
{
"epoch": 0.28,
"grad_norm": 1.359375,
"learning_rate": 4.523824064140961e-06,
"loss": 0.2313,
"step": 1120
},
{
"epoch": 0.2825,
"grad_norm": 1.515625,
"learning_rate": 4.510939088175211e-06,
"loss": 0.2209,
"step": 1130
},
{
"epoch": 0.285,
"grad_norm": 1.46875,
"learning_rate": 4.49790097141223e-06,
"loss": 0.2191,
"step": 1140
},
{
"epoch": 0.2875,
"grad_norm": 1.53125,
"learning_rate": 4.484710706755087e-06,
"loss": 0.2131,
"step": 1150
},
{
"epoch": 0.29,
"grad_norm": 1.5859375,
"learning_rate": 4.471369298693505e-06,
"loss": 0.2121,
"step": 1160
},
{
"epoch": 0.2925,
"grad_norm": 1.21875,
"learning_rate": 4.457877763227361e-06,
"loss": 0.2399,
"step": 1170
},
{
"epoch": 0.295,
"grad_norm": 1.328125,
"learning_rate": 4.444237127789315e-06,
"loss": 0.2145,
"step": 1180
},
{
"epoch": 0.2975,
"grad_norm": 1.171875,
"learning_rate": 4.430448431166567e-06,
"loss": 0.2179,
"step": 1190
},
{
"epoch": 0.3,
"grad_norm": 1.3515625,
"learning_rate": 4.416512723421752e-06,
"loss": 0.2104,
"step": 1200
},
{
"epoch": 0.3025,
"grad_norm": 1.234375,
"learning_rate": 4.402431065812968e-06,
"loss": 0.2194,
"step": 1210
},
{
"epoch": 0.305,
"grad_norm": 1.0546875,
"learning_rate": 4.388204530712959e-06,
"loss": 0.2179,
"step": 1220
},
{
"epoch": 0.3075,
"grad_norm": 1.1796875,
"learning_rate": 4.373834201527457e-06,
"loss": 0.2142,
"step": 1230
},
{
"epoch": 0.31,
"grad_norm": 1.3203125,
"learning_rate": 4.359321172612664e-06,
"loss": 0.2054,
"step": 1240
},
{
"epoch": 0.3125,
"grad_norm": 1.140625,
"learning_rate": 4.344666549191921e-06,
"loss": 0.2196,
"step": 1250
},
{
"epoch": 0.315,
"grad_norm": 1.2265625,
"learning_rate": 4.329871447271541e-06,
"loss": 0.2165,
"step": 1260
},
{
"epoch": 0.3175,
"grad_norm": 1.3515625,
"learning_rate": 4.314936993555816e-06,
"loss": 0.2021,
"step": 1270
},
{
"epoch": 0.32,
"grad_norm": 1.453125,
"learning_rate": 4.299864325361217e-06,
"loss": 0.2113,
"step": 1280
},
{
"epoch": 0.3225,
"grad_norm": 1.375,
"learning_rate": 4.284654590529784e-06,
"loss": 0.1996,
"step": 1290
},
{
"epoch": 0.325,
"grad_norm": 1.4140625,
"learning_rate": 4.269308947341711e-06,
"loss": 0.2101,
"step": 1300
},
{
"epoch": 0.3275,
"grad_norm": 1.3515625,
"learning_rate": 4.25382856442714e-06,
"loss": 0.2086,
"step": 1310
},
{
"epoch": 0.33,
"grad_norm": 1.453125,
"learning_rate": 4.238214620677164e-06,
"loss": 0.2125,
"step": 1320
},
{
"epoch": 0.3325,
"grad_norm": 1.1796875,
"learning_rate": 4.222468305154052e-06,
"loss": 0.2162,
"step": 1330
},
{
"epoch": 0.335,
"grad_norm": 1.25,
"learning_rate": 4.206590817000695e-06,
"loss": 0.2127,
"step": 1340
},
{
"epoch": 0.3375,
"grad_norm": 1.1796875,
"learning_rate": 4.190583365349289e-06,
"loss": 0.2095,
"step": 1350
},
{
"epoch": 0.34,
"grad_norm": 1.2578125,
"learning_rate": 4.174447169229252e-06,
"loss": 0.217,
"step": 1360
},
{
"epoch": 0.3425,
"grad_norm": 1.4140625,
"learning_rate": 4.158183457474392e-06,
"loss": 0.212,
"step": 1370
},
{
"epoch": 0.345,
"grad_norm": 1.4453125,
"learning_rate": 4.141793468629327e-06,
"loss": 0.2221,
"step": 1380
},
{
"epoch": 0.3475,
"grad_norm": 1.125,
"learning_rate": 4.125278450855165e-06,
"loss": 0.2079,
"step": 1390
},
{
"epoch": 0.35,
"grad_norm": 1.28125,
"learning_rate": 4.1086396618344474e-06,
"loss": 0.2085,
"step": 1400
},
{
"epoch": 0.3525,
"grad_norm": 1.3828125,
"learning_rate": 4.09187836867538e-06,
"loss": 0.2123,
"step": 1410
},
{
"epoch": 0.355,
"grad_norm": 1.21875,
"learning_rate": 4.074995847815331e-06,
"loss": 0.2116,
"step": 1420
},
{
"epoch": 0.3575,
"grad_norm": 1.3671875,
"learning_rate": 4.057993384923626e-06,
"loss": 0.2288,
"step": 1430
},
{
"epoch": 0.36,
"grad_norm": 1.1484375,
"learning_rate": 4.0408722748036426e-06,
"loss": 0.2038,
"step": 1440
},
{
"epoch": 0.3625,
"grad_norm": 1.28125,
"learning_rate": 4.023633821294203e-06,
"loss": 0.2123,
"step": 1450
},
{
"epoch": 0.365,
"grad_norm": 1.03125,
"learning_rate": 4.006279337170283e-06,
"loss": 0.1979,
"step": 1460
},
{
"epoch": 0.3675,
"grad_norm": 1.2109375,
"learning_rate": 3.988810144043041e-06,
"loss": 0.2113,
"step": 1470
},
{
"epoch": 0.37,
"grad_norm": 1.4453125,
"learning_rate": 3.971227572259167e-06,
"loss": 0.2077,
"step": 1480
},
{
"epoch": 0.3725,
"grad_norm": 1.1328125,
"learning_rate": 3.953532960799577e-06,
"loss": 0.1986,
"step": 1490
},
{
"epoch": 0.375,
"grad_norm": 1.1484375,
"learning_rate": 3.935727657177439e-06,
"loss": 0.1972,
"step": 1500
},
{
"epoch": 0.3775,
"grad_norm": 1.40625,
"learning_rate": 3.917813017335562e-06,
"loss": 0.2082,
"step": 1510
},
{
"epoch": 0.38,
"grad_norm": 1.375,
"learning_rate": 3.899790405543129e-06,
"loss": 0.2086,
"step": 1520
},
{
"epoch": 0.3825,
"grad_norm": 1.2734375,
"learning_rate": 3.881661194291805e-06,
"loss": 0.1968,
"step": 1530
},
{
"epoch": 0.385,
"grad_norm": 1.1640625,
"learning_rate": 3.863426764191216e-06,
"loss": 0.2064,
"step": 1540
},
{
"epoch": 0.3875,
"grad_norm": 1.1015625,
"learning_rate": 3.845088503863813e-06,
"loss": 0.1855,
"step": 1550
},
{
"epoch": 0.39,
"grad_norm": 1.2109375,
"learning_rate": 3.826647809839119e-06,
"loss": 0.2039,
"step": 1560
},
{
"epoch": 0.3925,
"grad_norm": 1.125,
"learning_rate": 3.8081060864473794e-06,
"loss": 0.2079,
"step": 1570
},
{
"epoch": 0.395,
"grad_norm": 1.09375,
"learning_rate": 3.7894647457126188e-06,
"loss": 0.2126,
"step": 1580
},
{
"epoch": 0.3975,
"grad_norm": 1.2734375,
"learning_rate": 3.770725207245106e-06,
"loss": 0.2156,
"step": 1590
},
{
"epoch": 0.4,
"grad_norm": 1.1015625,
"learning_rate": 3.751888898133249e-06,
"loss": 0.2122,
"step": 1600
},
{
"epoch": 0.4025,
"grad_norm": 1.3046875,
"learning_rate": 3.7329572528349145e-06,
"loss": 0.1996,
"step": 1610
},
{
"epoch": 0.405,
"grad_norm": 1.2734375,
"learning_rate": 3.7139317130681886e-06,
"loss": 0.2136,
"step": 1620
},
{
"epoch": 0.4075,
"grad_norm": 0.98046875,
"learning_rate": 3.694813727701584e-06,
"loss": 0.2055,
"step": 1630
},
{
"epoch": 0.41,
"grad_norm": 1.421875,
"learning_rate": 3.675604752643706e-06,
"loss": 0.2184,
"step": 1640
},
{
"epoch": 0.4125,
"grad_norm": 1.2265625,
"learning_rate": 3.6563062507323752e-06,
"loss": 0.2068,
"step": 1650
},
{
"epoch": 0.415,
"grad_norm": 1.09375,
"learning_rate": 3.6369196916232297e-06,
"loss": 0.2034,
"step": 1660
},
{
"epoch": 0.4175,
"grad_norm": 1.1640625,
"learning_rate": 3.6174465516778032e-06,
"loss": 0.1992,
"step": 1670
},
{
"epoch": 0.42,
"grad_norm": 1.1171875,
"learning_rate": 3.5978883138510963e-06,
"loss": 0.2037,
"step": 1680
},
{
"epoch": 0.4225,
"grad_norm": 1.15625,
"learning_rate": 3.578246467578642e-06,
"loss": 0.2185,
"step": 1690
},
{
"epoch": 0.425,
"grad_norm": 1.1875,
"learning_rate": 3.558522508663081e-06,
"loss": 0.2038,
"step": 1700
},
{
"epoch": 0.4275,
"grad_norm": 1.4296875,
"learning_rate": 3.538717939160249e-06,
"loss": 0.2088,
"step": 1710
},
{
"epoch": 0.43,
"grad_norm": 1.453125,
"learning_rate": 3.5188342672647897e-06,
"loss": 0.2041,
"step": 1720
},
{
"epoch": 0.4325,
"grad_norm": 1.125,
"learning_rate": 3.4988730071953005e-06,
"loss": 0.2065,
"step": 1730
},
{
"epoch": 0.435,
"grad_norm": 1.171875,
"learning_rate": 3.478835679079019e-06,
"loss": 0.2041,
"step": 1740
},
{
"epoch": 0.4375,
"grad_norm": 1.5390625,
"learning_rate": 3.4587238088360605e-06,
"loss": 0.1996,
"step": 1750
},
{
"epoch": 0.44,
"grad_norm": 1.390625,
"learning_rate": 3.438538928063208e-06,
"loss": 0.1931,
"step": 1760
},
{
"epoch": 0.4425,
"grad_norm": 1.2265625,
"learning_rate": 3.4182825739172826e-06,
"loss": 0.2095,
"step": 1770
},
{
"epoch": 0.445,
"grad_norm": 1.140625,
"learning_rate": 3.3979562889980777e-06,
"loss": 0.2097,
"step": 1780
},
{
"epoch": 0.4475,
"grad_norm": 1.34375,
"learning_rate": 3.377561621230887e-06,
"loss": 0.1874,
"step": 1790
},
{
"epoch": 0.45,
"grad_norm": 1.171875,
"learning_rate": 3.357100123748621e-06,
"loss": 0.2025,
"step": 1800
},
{
"epoch": 0.4525,
"grad_norm": 0.98828125,
"learning_rate": 3.3365733547735334e-06,
"loss": 0.1846,
"step": 1810
},
{
"epoch": 0.455,
"grad_norm": 1.28125,
"learning_rate": 3.315982877498555e-06,
"loss": 0.2049,
"step": 1820
},
{
"epoch": 0.4575,
"grad_norm": 1.1640625,
"learning_rate": 3.2953302599682487e-06,
"loss": 0.2021,
"step": 1830
},
{
"epoch": 0.46,
"grad_norm": 1.265625,
"learning_rate": 3.2746170749593998e-06,
"loss": 0.1979,
"step": 1840
},
{
"epoch": 0.4625,
"grad_norm": 1.3671875,
"learning_rate": 3.2538448998612394e-06,
"loss": 0.1965,
"step": 1850
},
{
"epoch": 0.465,
"grad_norm": 1.1328125,
"learning_rate": 3.233015316555326e-06,
"loss": 0.2151,
"step": 1860
},
{
"epoch": 0.4675,
"grad_norm": 1.078125,
"learning_rate": 3.212129911295074e-06,
"loss": 0.1824,
"step": 1870
},
{
"epoch": 0.47,
"grad_norm": 1.109375,
"learning_rate": 3.1911902745849526e-06,
"loss": 0.2051,
"step": 1880
},
{
"epoch": 0.4725,
"grad_norm": 1.09375,
"learning_rate": 3.17019800105937e-06,
"loss": 0.2145,
"step": 1890
},
{
"epoch": 0.475,
"grad_norm": 1.125,
"learning_rate": 3.1491546893612296e-06,
"loss": 0.2003,
"step": 1900
},
{
"epoch": 0.4775,
"grad_norm": 1.2109375,
"learning_rate": 3.128061942020189e-06,
"loss": 0.2078,
"step": 1910
},
{
"epoch": 0.48,
"grad_norm": 1.3671875,
"learning_rate": 3.1069213653306242e-06,
"loss": 0.1979,
"step": 1920
},
{
"epoch": 0.4825,
"grad_norm": 1.109375,
"learning_rate": 3.0857345692292968e-06,
"loss": 0.1925,
"step": 1930
},
{
"epoch": 0.485,
"grad_norm": 1.171875,
"learning_rate": 3.0645031671727598e-06,
"loss": 0.2042,
"step": 1940
},
{
"epoch": 0.4875,
"grad_norm": 1.1875,
"learning_rate": 3.0432287760144797e-06,
"loss": 0.2064,
"step": 1950
},
{
"epoch": 0.49,
"grad_norm": 1.4921875,
"learning_rate": 3.0219130158817093e-06,
"loss": 0.2035,
"step": 1960
},
{
"epoch": 0.4925,
"grad_norm": 1.125,
"learning_rate": 3.0005575100521115e-06,
"loss": 0.195,
"step": 1970
},
{
"epoch": 0.495,
"grad_norm": 1.3515625,
"learning_rate": 2.979163884830137e-06,
"loss": 0.2044,
"step": 1980
},
{
"epoch": 0.4975,
"grad_norm": 1.765625,
"learning_rate": 2.957733769423174e-06,
"loss": 0.212,
"step": 1990
},
{
"epoch": 0.5,
"grad_norm": 1.28125,
"learning_rate": 2.93626879581748e-06,
"loss": 0.1985,
"step": 2000
},
{
"epoch": 0.5025,
"grad_norm": 1.359375,
"learning_rate": 2.914770598653902e-06,
"loss": 0.1872,
"step": 2010
},
{
"epoch": 0.505,
"grad_norm": 1.1328125,
"learning_rate": 2.8932408151033868e-06,
"loss": 0.1884,
"step": 2020
},
{
"epoch": 0.5075,
"grad_norm": 1.203125,
"learning_rate": 2.8716810847423083e-06,
"loss": 0.2003,
"step": 2030
},
{
"epoch": 0.51,
"grad_norm": 1.2265625,
"learning_rate": 2.8500930494276035e-06,
"loss": 0.2057,
"step": 2040
},
{
"epoch": 0.5125,
"grad_norm": 1.2890625,
"learning_rate": 2.828478353171745e-06,
"loss": 0.197,
"step": 2050
},
{
"epoch": 0.515,
"grad_norm": 1.2890625,
"learning_rate": 2.8068386420175376e-06,
"loss": 0.1982,
"step": 2060
},
{
"epoch": 0.5175,
"grad_norm": 1.0390625,
"learning_rate": 2.785175563912766e-06,
"loss": 0.1865,
"step": 2070
},
{
"epoch": 0.52,
"grad_norm": 1.1171875,
"learning_rate": 2.7634907685846995e-06,
"loss": 0.1963,
"step": 2080
},
{
"epoch": 0.5225,
"grad_norm": 1.3515625,
"learning_rate": 2.7417859074144604e-06,
"loss": 0.1976,
"step": 2090
},
{
"epoch": 0.525,
"grad_norm": 1.234375,
"learning_rate": 2.7200626333112595e-06,
"loss": 0.1997,
"step": 2100
},
{
"epoch": 0.5275,
"grad_norm": 1.0703125,
"learning_rate": 2.6983226005865236e-06,
"loss": 0.1997,
"step": 2110
},
{
"epoch": 0.53,
"grad_norm": 1.046875,
"learning_rate": 2.676567464827917e-06,
"loss": 0.2114,
"step": 2120
},
{
"epoch": 0.5325,
"grad_norm": 1.1953125,
"learning_rate": 2.6547988827732546e-06,
"loss": 0.194,
"step": 2130
},
{
"epoch": 0.535,
"grad_norm": 1.1015625,
"learning_rate": 2.633018512184341e-06,
"loss": 0.198,
"step": 2140
},
{
"epoch": 0.5375,
"grad_norm": 1.2265625,
"learning_rate": 2.6112280117207223e-06,
"loss": 0.2126,
"step": 2150
},
{
"epoch": 0.54,
"grad_norm": 1.40625,
"learning_rate": 2.5894290408133744e-06,
"loss": 0.2021,
"step": 2160
},
{
"epoch": 0.5425,
"grad_norm": 1.2890625,
"learning_rate": 2.56762325953833e-06,
"loss": 0.1997,
"step": 2170
},
{
"epoch": 0.545,
"grad_norm": 1.140625,
"learning_rate": 2.5458123284902577e-06,
"loss": 0.1984,
"step": 2180
},
{
"epoch": 0.5475,
"grad_norm": 1.0078125,
"learning_rate": 2.5239979086560003e-06,
"loss": 0.1981,
"step": 2190
},
{
"epoch": 0.55,
"grad_norm": 1.2890625,
"learning_rate": 2.5021816612880884e-06,
"loss": 0.2044,
"step": 2200
},
{
"epoch": 0.5525,
"grad_norm": 1.1953125,
"learning_rate": 2.4803652477782228e-06,
"loss": 0.2039,
"step": 2210
},
{
"epoch": 0.555,
"grad_norm": 1.078125,
"learning_rate": 2.4585503295307565e-06,
"loss": 0.203,
"step": 2220
},
{
"epoch": 0.5575,
"grad_norm": 1.109375,
"learning_rate": 2.436738567836176e-06,
"loss": 0.1996,
"step": 2230
},
{
"epoch": 0.56,
"grad_norm": 1.09375,
"learning_rate": 2.4149316237445813e-06,
"loss": 0.1979,
"step": 2240
},
{
"epoch": 0.5625,
"grad_norm": 1.078125,
"learning_rate": 2.3931311579391946e-06,
"loss": 0.1939,
"step": 2250
},
{
"epoch": 0.565,
"grad_norm": 1.3359375,
"learning_rate": 2.37133883060989e-06,
"loss": 0.1991,
"step": 2260
},
{
"epoch": 0.5675,
"grad_norm": 1.140625,
"learning_rate": 2.3495563013267668e-06,
"loss": 0.19,
"step": 2270
},
{
"epoch": 0.57,
"grad_norm": 1.125,
"learning_rate": 2.3277852289137636e-06,
"loss": 0.1957,
"step": 2280
},
{
"epoch": 0.5725,
"grad_norm": 1.2578125,
"learning_rate": 2.306027271322336e-06,
"loss": 0.1966,
"step": 2290
},
{
"epoch": 0.575,
"grad_norm": 1.203125,
"learning_rate": 2.284284085505192e-06,
"loss": 0.2099,
"step": 2300
},
{
"epoch": 0.5775,
"grad_norm": 1.2265625,
"learning_rate": 2.2625573272901156e-06,
"loss": 0.1997,
"step": 2310
},
{
"epoch": 0.58,
"grad_norm": 1.1796875,
"learning_rate": 2.240848651253863e-06,
"loss": 0.1897,
"step": 2320
},
{
"epoch": 0.5825,
"grad_norm": 0.9921875,
"learning_rate": 2.2191597105961613e-06,
"loss": 0.1826,
"step": 2330
},
{
"epoch": 0.585,
"grad_norm": 1.109375,
"learning_rate": 2.1974921570138155e-06,
"loss": 0.1917,
"step": 2340
},
{
"epoch": 0.5875,
"grad_norm": 1.265625,
"learning_rate": 2.1758476405749207e-06,
"loss": 0.1963,
"step": 2350
},
{
"epoch": 0.59,
"grad_norm": 1.140625,
"learning_rate": 2.154227809593203e-06,
"loss": 0.1931,
"step": 2360
},
{
"epoch": 0.5925,
"grad_norm": 1.1953125,
"learning_rate": 2.1326343105024962e-06,
"loss": 0.202,
"step": 2370
},
{
"epoch": 0.595,
"grad_norm": 1.2265625,
"learning_rate": 2.111068787731358e-06,
"loss": 0.2061,
"step": 2380
},
{
"epoch": 0.5975,
"grad_norm": 1.578125,
"learning_rate": 2.089532883577843e-06,
"loss": 0.1867,
"step": 2390
},
{
"epoch": 0.6,
"grad_norm": 1.234375,
"learning_rate": 2.068028238084432e-06,
"loss": 0.2039,
"step": 2400
},
{
"epoch": 0.6025,
"grad_norm": 1.171875,
"learning_rate": 2.046556488913137e-06,
"loss": 0.209,
"step": 2410
},
{
"epoch": 0.605,
"grad_norm": 1.234375,
"learning_rate": 2.025119271220789e-06,
"loss": 0.1894,
"step": 2420
},
{
"epoch": 0.6075,
"grad_norm": 1.125,
"learning_rate": 2.0037182175345137e-06,
"loss": 0.2098,
"step": 2430
},
{
"epoch": 0.61,
"grad_norm": 1.1171875,
"learning_rate": 1.9823549576274048e-06,
"loss": 0.1955,
"step": 2440
},
{
"epoch": 0.6125,
"grad_norm": 1.1640625,
"learning_rate": 1.961031118394418e-06,
"loss": 0.1881,
"step": 2450
},
{
"epoch": 0.615,
"grad_norm": 1.453125,
"learning_rate": 1.939748323728468e-06,
"loss": 0.1976,
"step": 2460
},
{
"epoch": 0.6175,
"grad_norm": 1.4296875,
"learning_rate": 1.918508194396769e-06,
"loss": 0.2028,
"step": 2470
},
{
"epoch": 0.62,
"grad_norm": 1.828125,
"learning_rate": 1.8973123479174038e-06,
"loss": 0.2065,
"step": 2480
},
{
"epoch": 0.6225,
"grad_norm": 1.3359375,
"learning_rate": 1.8761623984361444e-06,
"loss": 0.198,
"step": 2490
},
{
"epoch": 0.625,
"grad_norm": 1.1171875,
"learning_rate": 1.8550599566035299e-06,
"loss": 0.192,
"step": 2500
},
{
"epoch": 0.6275,
"grad_norm": 1.2265625,
"learning_rate": 1.834006629452207e-06,
"loss": 0.1894,
"step": 2510
},
{
"epoch": 0.63,
"grad_norm": 1.15625,
"learning_rate": 1.8130040202745488e-06,
"loss": 0.203,
"step": 2520
},
{
"epoch": 0.6325,
"grad_norm": 1.171875,
"learning_rate": 1.7920537285005607e-06,
"loss": 0.1879,
"step": 2530
},
{
"epoch": 0.635,
"grad_norm": 1.1953125,
"learning_rate": 1.7711573495760725e-06,
"loss": 0.1887,
"step": 2540
},
{
"epoch": 0.6375,
"grad_norm": 1.2578125,
"learning_rate": 1.750316474841242e-06,
"loss": 0.1894,
"step": 2550
},
{
"epoch": 0.64,
"grad_norm": 1.1015625,
"learning_rate": 1.7295326914093713e-06,
"loss": 0.1991,
"step": 2560
},
{
"epoch": 0.6425,
"grad_norm": 1.3984375,
"learning_rate": 1.7088075820460348e-06,
"loss": 0.1982,
"step": 2570
},
{
"epoch": 0.645,
"grad_norm": 1.1171875,
"learning_rate": 1.6881427250485516e-06,
"loss": 0.1876,
"step": 2580
},
{
"epoch": 0.6475,
"grad_norm": 1.1015625,
"learning_rate": 1.6675396941257896e-06,
"loss": 0.1853,
"step": 2590
},
{
"epoch": 0.65,
"grad_norm": 1.1171875,
"learning_rate": 1.6470000582783205e-06,
"loss": 0.1917,
"step": 2600
},
{
"epoch": 0.6525,
"grad_norm": 1.296875,
"learning_rate": 1.6265253816789372e-06,
"loss": 0.2017,
"step": 2610
},
{
"epoch": 0.655,
"grad_norm": 1.203125,
"learning_rate": 1.6061172235535342e-06,
"loss": 0.193,
"step": 2620
},
{
"epoch": 0.6575,
"grad_norm": 1.25,
"learning_rate": 1.5857771380623643e-06,
"loss": 0.2018,
"step": 2630
},
{
"epoch": 0.66,
"grad_norm": 1.4765625,
"learning_rate": 1.5655066741816898e-06,
"loss": 0.2062,
"step": 2640
},
{
"epoch": 0.6625,
"grad_norm": 1.1484375,
"learning_rate": 1.545307375585814e-06,
"loss": 0.1961,
"step": 2650
},
{
"epoch": 0.665,
"grad_norm": 1.421875,
"learning_rate": 1.5251807805295302e-06,
"loss": 0.208,
"step": 2660
},
{
"epoch": 0.6675,
"grad_norm": 1.125,
"learning_rate": 1.5051284217309743e-06,
"loss": 0.1951,
"step": 2670
},
{
"epoch": 0.67,
"grad_norm": 1.0859375,
"learning_rate": 1.4851518262549058e-06,
"loss": 0.1936,
"step": 2680
},
{
"epoch": 0.6725,
"grad_norm": 1.25,
"learning_rate": 1.465252515396413e-06,
"loss": 0.2048,
"step": 2690
},
{
"epoch": 0.675,
"grad_norm": 1.1328125,
"learning_rate": 1.4454320045650606e-06,
"loss": 0.1979,
"step": 2700
},
{
"epoch": 0.6775,
"grad_norm": 1.265625,
"learning_rate": 1.4256918031694866e-06,
"loss": 0.198,
"step": 2710
},
{
"epoch": 0.68,
"grad_norm": 1.15625,
"learning_rate": 1.4060334145024543e-06,
"loss": 0.1902,
"step": 2720
},
{
"epoch": 0.6825,
"grad_norm": 1.046875,
"learning_rate": 1.3864583356263706e-06,
"loss": 0.1911,
"step": 2730
},
{
"epoch": 0.685,
"grad_norm": 1.15625,
"learning_rate": 1.366968057259282e-06,
"loss": 0.2109,
"step": 2740
},
{
"epoch": 0.6875,
"grad_norm": 1.140625,
"learning_rate": 1.3475640636613447e-06,
"loss": 0.2015,
"step": 2750
},
{
"epoch": 0.69,
"grad_norm": 1.265625,
"learning_rate": 1.3282478325217961e-06,
"loss": 0.2006,
"step": 2760
},
{
"epoch": 0.6925,
"grad_norm": 1.046875,
"learning_rate": 1.3090208348464244e-06,
"loss": 0.1842,
"step": 2770
},
{
"epoch": 0.695,
"grad_norm": 1.2890625,
"learning_rate": 1.289884534845542e-06,
"loss": 0.2091,
"step": 2780
},
{
"epoch": 0.6975,
"grad_norm": 0.953125,
"learning_rate": 1.2708403898224839e-06,
"loss": 0.185,
"step": 2790
},
{
"epoch": 0.7,
"grad_norm": 1.0625,
"learning_rate": 1.2518898500626259e-06,
"loss": 0.1875,
"step": 2800
},
{
"epoch": 0.7025,
"grad_norm": 1.140625,
"learning_rate": 1.2330343587229397e-06,
"loss": 0.1926,
"step": 2810
},
{
"epoch": 0.705,
"grad_norm": 1.234375,
"learning_rate": 1.2142753517220945e-06,
"loss": 0.1988,
"step": 2820
},
{
"epoch": 0.7075,
"grad_norm": 1.046875,
"learning_rate": 1.1956142576311011e-06,
"loss": 0.1988,
"step": 2830
},
{
"epoch": 0.71,
"grad_norm": 1.1484375,
"learning_rate": 1.177052497564524e-06,
"loss": 0.1817,
"step": 2840
},
{
"epoch": 0.7125,
"grad_norm": 1.2109375,
"learning_rate": 1.1585914850722565e-06,
"loss": 0.1887,
"step": 2850
},
{
"epoch": 0.715,
"grad_norm": 1.140625,
"learning_rate": 1.1402326260318752e-06,
"loss": 0.1821,
"step": 2860
},
{
"epoch": 0.7175,
"grad_norm": 1.234375,
"learning_rate": 1.121977318541575e-06,
"loss": 0.1948,
"step": 2870
},
{
"epoch": 0.72,
"grad_norm": 1.2265625,
"learning_rate": 1.1038269528136989e-06,
"loss": 0.2054,
"step": 2880
},
{
"epoch": 0.7225,
"grad_norm": 1.0859375,
"learning_rate": 1.0857829110688695e-06,
"loss": 0.1956,
"step": 2890
},
{
"epoch": 0.725,
"grad_norm": 1.09375,
"learning_rate": 1.0678465674307273e-06,
"loss": 0.2033,
"step": 2900
},
{
"epoch": 0.7275,
"grad_norm": 1.046875,
"learning_rate": 1.0500192878212826e-06,
"loss": 0.1986,
"step": 2910
},
{
"epoch": 0.73,
"grad_norm": 1.1328125,
"learning_rate": 1.032302429856899e-06,
"loss": 0.1979,
"step": 2920
},
{
"epoch": 0.7325,
"grad_norm": 1.28125,
"learning_rate": 1.014697342744904e-06,
"loss": 0.1994,
"step": 2930
},
{
"epoch": 0.735,
"grad_norm": 1.0859375,
"learning_rate": 9.97205367180842e-07,
"loss": 0.1921,
"step": 2940
},
{
"epoch": 0.7375,
"grad_norm": 1.1328125,
"learning_rate": 9.798278352463752e-07,
"loss": 0.1999,
"step": 2950
},
{
"epoch": 0.74,
"grad_norm": 1.1015625,
"learning_rate": 9.625660703078394e-07,
"loss": 0.2021,
"step": 2960
},
{
"epoch": 0.7425,
"grad_norm": 1.296875,
"learning_rate": 9.45421386915468e-07,
"loss": 0.2027,
"step": 2970
},
{
"epoch": 0.745,
"grad_norm": 1.34375,
"learning_rate": 9.283950907032788e-07,
"loss": 0.1898,
"step": 2980
},
{
"epoch": 0.7475,
"grad_norm": 1.1484375,
"learning_rate": 9.114884782896482e-07,
"loss": 0.1943,
"step": 2990
},
{
"epoch": 0.75,
"grad_norm": 1.3515625,
"learning_rate": 8.947028371785677e-07,
"loss": 0.2189,
"step": 3000
},
{
"epoch": 0.7525,
"grad_norm": 1.0703125,
"learning_rate": 8.780394456615974e-07,
"loss": 0.1889,
"step": 3010
},
{
"epoch": 0.755,
"grad_norm": 1.3359375,
"learning_rate": 8.614995727205155e-07,
"loss": 0.2013,
"step": 3020
},
{
"epoch": 0.7575,
"grad_norm": 1.0703125,
"learning_rate": 8.450844779306827e-07,
"loss": 0.1867,
"step": 3030
},
{
"epoch": 0.76,
"grad_norm": 1.0234375,
"learning_rate": 8.28795411365122e-07,
"loss": 0.2004,
"step": 3040
},
{
"epoch": 0.7625,
"grad_norm": 1.484375,
"learning_rate": 8.126336134993176e-07,
"loss": 0.1959,
"step": 3050
},
{
"epoch": 0.765,
"grad_norm": 1.1484375,
"learning_rate": 7.966003151167498e-07,
"loss": 0.1855,
"step": 3060
},
{
"epoch": 0.7675,
"grad_norm": 1.125,
"learning_rate": 7.806967372151661e-07,
"loss": 0.1901,
"step": 3070
},
{
"epoch": 0.77,
"grad_norm": 1.609375,
"learning_rate": 7.649240909135966e-07,
"loss": 0.2125,
"step": 3080
},
{
"epoch": 0.7725,
"grad_norm": 1.171875,
"learning_rate": 7.492835773601234e-07,
"loss": 0.1916,
"step": 3090
},
{
"epoch": 0.775,
"grad_norm": 1.140625,
"learning_rate": 7.337763876404078e-07,
"loss": 0.1844,
"step": 3100
},
{
"epoch": 0.7775,
"grad_norm": 1.0390625,
"learning_rate": 7.184037026869867e-07,
"loss": 0.1908,
"step": 3110
},
{
"epoch": 0.78,
"grad_norm": 1.1015625,
"learning_rate": 7.031666931893361e-07,
"loss": 0.1858,
"step": 3120
},
{
"epoch": 0.7825,
"grad_norm": 1.171875,
"learning_rate": 6.880665195047226e-07,
"loss": 0.1976,
"step": 3130
},
{
"epoch": 0.785,
"grad_norm": 1.1015625,
"learning_rate": 6.731043315698346e-07,
"loss": 0.1901,
"step": 3140
},
{
"epoch": 0.7875,
"grad_norm": 1.15625,
"learning_rate": 6.58281268813212e-07,
"loss": 0.2021,
"step": 3150
},
{
"epoch": 0.79,
"grad_norm": 1.4296875,
"learning_rate": 6.435984600684731e-07,
"loss": 0.2088,
"step": 3160
},
{
"epoch": 0.7925,
"grad_norm": 1.0390625,
"learning_rate": 6.290570234883506e-07,
"loss": 0.2004,
"step": 3170
},
{
"epoch": 0.795,
"grad_norm": 4.46875,
"learning_rate": 6.146580664595391e-07,
"loss": 0.1964,
"step": 3180
},
{
"epoch": 0.7975,
"grad_norm": 1.15625,
"learning_rate": 6.004026855183656e-07,
"loss": 0.1996,
"step": 3190
},
{
"epoch": 0.8,
"grad_norm": 1.0390625,
"learning_rate": 5.862919662672801e-07,
"loss": 0.1919,
"step": 3200
},
{
"epoch": 0.8025,
"grad_norm": 1.1796875,
"learning_rate": 5.723269832921849e-07,
"loss": 0.202,
"step": 3210
},
{
"epoch": 0.805,
"grad_norm": 1.1015625,
"learning_rate": 5.585088000806016e-07,
"loss": 0.2024,
"step": 3220
},
{
"epoch": 0.8075,
"grad_norm": 1.2734375,
"learning_rate": 5.448384689406804e-07,
"loss": 0.1962,
"step": 3230
},
{
"epoch": 0.81,
"grad_norm": 1.2890625,
"learning_rate": 5.313170309210655e-07,
"loss": 0.1879,
"step": 3240
},
{
"epoch": 0.8125,
"grad_norm": 0.99609375,
"learning_rate": 5.179455157316124e-07,
"loss": 0.1987,
"step": 3250
},
{
"epoch": 0.815,
"grad_norm": 1.15625,
"learning_rate": 5.047249416649713e-07,
"loss": 0.2078,
"step": 3260
},
{
"epoch": 0.8175,
"grad_norm": 14.25,
"learning_rate": 4.916563155190446e-07,
"loss": 0.2002,
"step": 3270
},
{
"epoch": 0.82,
"grad_norm": 1.09375,
"learning_rate": 4.787406325203101e-07,
"loss": 0.1948,
"step": 3280
},
{
"epoch": 0.8225,
"grad_norm": 1.0859375,
"learning_rate": 4.6597887624803273e-07,
"loss": 0.1981,
"step": 3290
},
{
"epoch": 0.825,
"grad_norm": 1.1796875,
"learning_rate": 4.533720185593621e-07,
"loss": 0.1965,
"step": 3300
},
{
"epoch": 0.8275,
"grad_norm": 1.1875,
"learning_rate": 4.4092101951532076e-07,
"loss": 0.1926,
"step": 3310
},
{
"epoch": 0.83,
"grad_norm": 1.2578125,
"learning_rate": 4.2862682730769157e-07,
"loss": 0.1952,
"step": 3320
},
{
"epoch": 0.8325,
"grad_norm": 1.1484375,
"learning_rate": 4.164903781868096e-07,
"loss": 0.1853,
"step": 3330
},
{
"epoch": 0.835,
"grad_norm": 1.6015625,
"learning_rate": 4.045125963902641e-07,
"loss": 0.2008,
"step": 3340
},
{
"epoch": 0.8375,
"grad_norm": 1.1015625,
"learning_rate": 3.9269439407251365e-07,
"loss": 0.2064,
"step": 3350
},
{
"epoch": 0.84,
"grad_norm": 1.1640625,
"learning_rate": 3.810366712354199e-07,
"loss": 0.1833,
"step": 3360
},
{
"epoch": 0.8425,
"grad_norm": 1.0546875,
"learning_rate": 3.6954031565971187e-07,
"loss": 0.1889,
"step": 3370
},
{
"epoch": 0.845,
"grad_norm": 0.984375,
"learning_rate": 3.5820620283737615e-07,
"loss": 0.1785,
"step": 3380
},
{
"epoch": 0.8475,
"grad_norm": 1.46875,
"learning_rate": 3.4703519590498615e-07,
"loss": 0.2022,
"step": 3390
},
{
"epoch": 0.85,
"grad_norm": 1.140625,
"learning_rate": 3.360281455779704e-07,
"loss": 0.1942,
"step": 3400
},
{
"epoch": 0.8525,
"grad_norm": 1.1640625,
"learning_rate": 3.2518589008582597e-07,
"loss": 0.1933,
"step": 3410
},
{
"epoch": 0.855,
"grad_norm": 1.1875,
"learning_rate": 3.1450925510828705e-07,
"loss": 0.1902,
"step": 3420
},
{
"epoch": 0.8575,
"grad_norm": 1.1796875,
"learning_rate": 3.039990537124432e-07,
"loss": 0.2161,
"step": 3430
},
{
"epoch": 0.86,
"grad_norm": 1.078125,
"learning_rate": 2.936560862908225e-07,
"loss": 0.178,
"step": 3440
},
{
"epoch": 0.8625,
"grad_norm": 1.234375,
"learning_rate": 2.8348114050043813e-07,
"loss": 0.2021,
"step": 3450
},
{
"epoch": 0.865,
"grad_norm": 1.578125,
"learning_rate": 2.7347499120280677e-07,
"loss": 0.2043,
"step": 3460
},
{
"epoch": 0.8675,
"grad_norm": 1.328125,
"learning_rate": 2.636384004049375e-07,
"loss": 0.1921,
"step": 3470
},
{
"epoch": 0.87,
"grad_norm": 1.125,
"learning_rate": 2.5397211720130267e-07,
"loss": 0.19,
"step": 3480
},
{
"epoch": 0.8725,
"grad_norm": 1.171875,
"learning_rate": 2.4447687771679414e-07,
"loss": 0.1914,
"step": 3490
},
{
"epoch": 0.875,
"grad_norm": 1.09375,
"learning_rate": 2.3515340505066043e-07,
"loss": 0.1919,
"step": 3500
},
{
"epoch": 0.8775,
"grad_norm": 1.0859375,
"learning_rate": 2.260024092214419e-07,
"loss": 0.2006,
"step": 3510
},
{
"epoch": 0.88,
"grad_norm": 1.1640625,
"learning_rate": 2.170245871129012e-07,
"loss": 0.1982,
"step": 3520
},
{
"epoch": 0.8825,
"grad_norm": 1.9453125,
"learning_rate": 2.0822062242095015e-07,
"loss": 0.2113,
"step": 3530
},
{
"epoch": 0.885,
"grad_norm": 1.1171875,
"learning_rate": 1.995911856015867e-07,
"loss": 0.1969,
"step": 3540
},
{
"epoch": 0.8875,
"grad_norm": 1.1796875,
"learning_rate": 1.9113693381983405e-07,
"loss": 0.1974,
"step": 3550
},
{
"epoch": 0.89,
"grad_norm": 1.203125,
"learning_rate": 1.8285851089969802e-07,
"loss": 0.1987,
"step": 3560
},
{
"epoch": 0.8925,
"grad_norm": 1.3046875,
"learning_rate": 1.7475654727513502e-07,
"loss": 0.201,
"step": 3570
},
{
"epoch": 0.895,
"grad_norm": 1.21875,
"learning_rate": 1.668316599420433e-07,
"loss": 0.185,
"step": 3580
},
{
"epoch": 0.8975,
"grad_norm": 1.0703125,
"learning_rate": 1.5908445241127528e-07,
"loss": 0.1872,
"step": 3590
},
{
"epoch": 0.9,
"grad_norm": 1.1953125,
"learning_rate": 1.5151551466267956e-07,
"loss": 0.1985,
"step": 3600
},
{
"epoch": 0.9025,
"grad_norm": 1.1796875,
"learning_rate": 1.441254231001696e-07,
"loss": 0.2036,
"step": 3610
},
{
"epoch": 0.905,
"grad_norm": 1.125,
"learning_rate": 1.3691474050782972e-07,
"loss": 0.1902,
"step": 3620
},
{
"epoch": 0.9075,
"grad_norm": 1.078125,
"learning_rate": 1.2988401600705635e-07,
"loss": 0.1921,
"step": 3630
},
{
"epoch": 0.91,
"grad_norm": 1.1640625,
"learning_rate": 1.2303378501474174e-07,
"loss": 0.1752,
"step": 3640
},
{
"epoch": 0.9125,
"grad_norm": 1.1953125,
"learning_rate": 1.16364569202497e-07,
"loss": 0.1893,
"step": 3650
},
{
"epoch": 0.915,
"grad_norm": 1.1953125,
"learning_rate": 1.0987687645692746e-07,
"loss": 0.2005,
"step": 3660
},
{
"epoch": 0.9175,
"grad_norm": 1.4375,
"learning_rate": 1.035712008409534e-07,
"loss": 0.1948,
"step": 3670
},
{
"epoch": 0.92,
"grad_norm": 1.125,
"learning_rate": 9.744802255618662e-08,
"loss": 0.1903,
"step": 3680
},
{
"epoch": 0.9225,
"grad_norm": 1.109375,
"learning_rate": 9.150780790636054e-08,
"loss": 0.2147,
"step": 3690
},
{
"epoch": 0.925,
"grad_norm": 1.2734375,
"learning_rate": 8.575100926181884e-08,
"loss": 0.19,
"step": 3700
},
{
"epoch": 0.9275,
"grad_norm": 1.09375,
"learning_rate": 8.017806502506692e-08,
"loss": 0.2004,
"step": 3710
},
{
"epoch": 0.93,
"grad_norm": 1.2421875,
"learning_rate": 7.478939959738502e-08,
"loss": 0.2059,
"step": 3720
},
{
"epoch": 0.9325,
"grad_norm": 1.234375,
"learning_rate": 6.958542334650847e-08,
"loss": 0.1908,
"step": 3730
},
{
"epoch": 0.935,
"grad_norm": 1.6328125,
"learning_rate": 6.456653257537665e-08,
"loss": 0.1979,
"step": 3740
},
{
"epoch": 0.9375,
"grad_norm": 1.0859375,
"learning_rate": 5.973310949195343e-08,
"loss": 0.1934,
"step": 3750
},
{
"epoch": 0.94,
"grad_norm": 1.078125,
"learning_rate": 5.50855221801197e-08,
"loss": 0.1893,
"step": 3760
},
{
"epoch": 0.9425,
"grad_norm": 1.15625,
"learning_rate": 5.062412457164323e-08,
"loss": 0.1985,
"step": 3770
},
{
"epoch": 0.945,
"grad_norm": 1.1953125,
"learning_rate": 4.634925641922472e-08,
"loss": 0.2008,
"step": 3780
},
{
"epoch": 0.9475,
"grad_norm": 1.234375,
"learning_rate": 4.226124327062514e-08,
"loss": 0.2056,
"step": 3790
},
{
"epoch": 0.95,
"grad_norm": 1.0625,
"learning_rate": 3.836039644387307e-08,
"loss": 0.2036,
"step": 3800
},
{
"epoch": 0.9525,
"grad_norm": 16.25,
"learning_rate": 3.4647013003556996e-08,
"loss": 0.199,
"step": 3810
},
{
"epoch": 0.955,
"grad_norm": 1.1015625,
"learning_rate": 3.112137573820284e-08,
"loss": 0.1917,
"step": 3820
},
{
"epoch": 0.9575,
"grad_norm": 1.0625,
"learning_rate": 2.7783753138738713e-08,
"loss": 0.1946,
"step": 3830
},
{
"epoch": 0.96,
"grad_norm": 1.1171875,
"learning_rate": 2.463439937804707e-08,
"loss": 0.1816,
"step": 3840
},
{
"epoch": 0.9625,
"grad_norm": 1.2421875,
"learning_rate": 2.1673554291610775e-08,
"loss": 0.1859,
"step": 3850
},
{
"epoch": 0.965,
"grad_norm": 1.375,
"learning_rate": 1.8901443359245765e-08,
"loss": 0.1993,
"step": 3860
},
{
"epoch": 0.9675,
"grad_norm": 1.328125,
"learning_rate": 1.6318277687932816e-08,
"loss": 0.2007,
"step": 3870
},
{
"epoch": 0.97,
"grad_norm": 1.1484375,
"learning_rate": 1.3924253995738769e-08,
"loss": 0.1925,
"step": 3880
},
{
"epoch": 0.9725,
"grad_norm": 1.15625,
"learning_rate": 1.1719554596836546e-08,
"loss": 0.1996,
"step": 3890
},
{
"epoch": 0.975,
"grad_norm": 1.125,
"learning_rate": 9.704347387620994e-09,
"loss": 0.1836,
"step": 3900
},
{
"epoch": 0.9775,
"grad_norm": 1.1640625,
"learning_rate": 7.878785833923819e-09,
"loss": 0.1928,
"step": 3910
},
{
"epoch": 0.98,
"grad_norm": 1.46875,
"learning_rate": 6.243008959324892e-09,
"loss": 0.1967,
"step": 3920
},
{
"epoch": 0.9825,
"grad_norm": 1.171875,
"learning_rate": 4.797141334566268e-09,
"loss": 0.1992,
"step": 3930
},
{
"epoch": 0.985,
"grad_norm": 1.2734375,
"learning_rate": 3.5412930680658876e-09,
"loss": 0.1993,
"step": 3940
},
{
"epoch": 0.9875,
"grad_norm": 1.125,
"learning_rate": 2.475559797531224e-09,
"loss": 0.1989,
"step": 3950
},
{
"epoch": 0.99,
"grad_norm": 1.09375,
"learning_rate": 1.6000226826770604e-09,
"loss": 0.1826,
"step": 3960
},
{
"epoch": 0.9925,
"grad_norm": 1.171875,
"learning_rate": 9.147483990443184e-10,
"loss": 0.1969,
"step": 3970
},
{
"epoch": 0.995,
"grad_norm": 1.21875,
"learning_rate": 4.197891329230097e-10,
"loss": 0.2085,
"step": 3980
},
{
"epoch": 0.9975,
"grad_norm": 1.2890625,
"learning_rate": 1.1518257737763716e-10,
"loss": 0.1928,
"step": 3990
},
{
"epoch": 1.0,
"grad_norm": 1.125,
"learning_rate": 9.51929376435956e-13,
"loss": 0.1986,
"step": 4000
}
],
"logging_steps": 10,
"max_steps": 4000,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 2000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 3.82656621576192e+17,
"train_batch_size": 32,
"trial_name": null,
"trial_params": null
}