13365 lines
322 KiB
JSON
13365 lines
322 KiB
JSON
{
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.9997370496976071,
|
|
"eval_steps": 500,
|
|
"global_step": 1901,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.0005259006047856955,
|
|
"grad_norm": 11.830399513244629,
|
|
"learning_rate": 2.1008403361344538e-08,
|
|
"loss": 1.4499,
|
|
"step": 1
|
|
},
|
|
{
|
|
"epoch": 0.001051801209571391,
|
|
"grad_norm": 12.041601181030273,
|
|
"learning_rate": 4.2016806722689076e-08,
|
|
"loss": 1.5018,
|
|
"step": 2
|
|
},
|
|
{
|
|
"epoch": 0.0015777018143570865,
|
|
"grad_norm": 12.531728744506836,
|
|
"learning_rate": 6.302521008403361e-08,
|
|
"loss": 1.5618,
|
|
"step": 3
|
|
},
|
|
{
|
|
"epoch": 0.002103602419142782,
|
|
"grad_norm": 12.059429168701172,
|
|
"learning_rate": 8.403361344537815e-08,
|
|
"loss": 1.4142,
|
|
"step": 4
|
|
},
|
|
{
|
|
"epoch": 0.0026295030239284773,
|
|
"grad_norm": 13.222566604614258,
|
|
"learning_rate": 1.050420168067227e-07,
|
|
"loss": 1.5547,
|
|
"step": 5
|
|
},
|
|
{
|
|
"epoch": 0.003155403628714173,
|
|
"grad_norm": 13.111701011657715,
|
|
"learning_rate": 1.2605042016806723e-07,
|
|
"loss": 1.565,
|
|
"step": 6
|
|
},
|
|
{
|
|
"epoch": 0.0036813042334998686,
|
|
"grad_norm": 12.945558547973633,
|
|
"learning_rate": 1.4705882352941178e-07,
|
|
"loss": 1.5251,
|
|
"step": 7
|
|
},
|
|
{
|
|
"epoch": 0.004207204838285564,
|
|
"grad_norm": 13.110393524169922,
|
|
"learning_rate": 1.680672268907563e-07,
|
|
"loss": 1.5068,
|
|
"step": 8
|
|
},
|
|
{
|
|
"epoch": 0.00473310544307126,
|
|
"grad_norm": 15.715043067932129,
|
|
"learning_rate": 1.8907563025210085e-07,
|
|
"loss": 1.5738,
|
|
"step": 9
|
|
},
|
|
{
|
|
"epoch": 0.005259006047856955,
|
|
"grad_norm": 13.167400360107422,
|
|
"learning_rate": 2.100840336134454e-07,
|
|
"loss": 1.5641,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.00578490665264265,
|
|
"grad_norm": 10.724669456481934,
|
|
"learning_rate": 2.3109243697478996e-07,
|
|
"loss": 1.3757,
|
|
"step": 11
|
|
},
|
|
{
|
|
"epoch": 0.006310807257428346,
|
|
"grad_norm": 12.651016235351562,
|
|
"learning_rate": 2.5210084033613445e-07,
|
|
"loss": 1.4832,
|
|
"step": 12
|
|
},
|
|
{
|
|
"epoch": 0.0068367078622140415,
|
|
"grad_norm": 13.033792495727539,
|
|
"learning_rate": 2.73109243697479e-07,
|
|
"loss": 1.5997,
|
|
"step": 13
|
|
},
|
|
{
|
|
"epoch": 0.007362608466999737,
|
|
"grad_norm": 12.468499183654785,
|
|
"learning_rate": 2.9411764705882356e-07,
|
|
"loss": 1.5849,
|
|
"step": 14
|
|
},
|
|
{
|
|
"epoch": 0.007888509071785432,
|
|
"grad_norm": 10.891519546508789,
|
|
"learning_rate": 3.151260504201681e-07,
|
|
"loss": 1.4915,
|
|
"step": 15
|
|
},
|
|
{
|
|
"epoch": 0.008414409676571128,
|
|
"grad_norm": 11.143836975097656,
|
|
"learning_rate": 3.361344537815126e-07,
|
|
"loss": 1.4589,
|
|
"step": 16
|
|
},
|
|
{
|
|
"epoch": 0.008940310281356823,
|
|
"grad_norm": 12.372116088867188,
|
|
"learning_rate": 3.5714285714285716e-07,
|
|
"loss": 1.4681,
|
|
"step": 17
|
|
},
|
|
{
|
|
"epoch": 0.00946621088614252,
|
|
"grad_norm": 11.152701377868652,
|
|
"learning_rate": 3.781512605042017e-07,
|
|
"loss": 1.2876,
|
|
"step": 18
|
|
},
|
|
{
|
|
"epoch": 0.009992111490928214,
|
|
"grad_norm": 10.849949836730957,
|
|
"learning_rate": 3.991596638655462e-07,
|
|
"loss": 1.3978,
|
|
"step": 19
|
|
},
|
|
{
|
|
"epoch": 0.01051801209571391,
|
|
"grad_norm": 9.65892219543457,
|
|
"learning_rate": 4.201680672268908e-07,
|
|
"loss": 1.427,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.011043912700499606,
|
|
"grad_norm": 8.445028305053711,
|
|
"learning_rate": 4.4117647058823536e-07,
|
|
"loss": 1.3523,
|
|
"step": 21
|
|
},
|
|
{
|
|
"epoch": 0.0115698133052853,
|
|
"grad_norm": 9.946889877319336,
|
|
"learning_rate": 4.621848739495799e-07,
|
|
"loss": 1.4783,
|
|
"step": 22
|
|
},
|
|
{
|
|
"epoch": 0.012095713910070997,
|
|
"grad_norm": 7.978725910186768,
|
|
"learning_rate": 4.831932773109245e-07,
|
|
"loss": 1.5561,
|
|
"step": 23
|
|
},
|
|
{
|
|
"epoch": 0.012621614514856692,
|
|
"grad_norm": 7.055315017700195,
|
|
"learning_rate": 5.042016806722689e-07,
|
|
"loss": 1.3729,
|
|
"step": 24
|
|
},
|
|
{
|
|
"epoch": 0.013147515119642388,
|
|
"grad_norm": 6.764873504638672,
|
|
"learning_rate": 5.252100840336135e-07,
|
|
"loss": 1.4199,
|
|
"step": 25
|
|
},
|
|
{
|
|
"epoch": 0.013673415724428083,
|
|
"grad_norm": 7.641580104827881,
|
|
"learning_rate": 5.46218487394958e-07,
|
|
"loss": 1.487,
|
|
"step": 26
|
|
},
|
|
{
|
|
"epoch": 0.014199316329213778,
|
|
"grad_norm": 6.178252220153809,
|
|
"learning_rate": 5.672268907563026e-07,
|
|
"loss": 1.3628,
|
|
"step": 27
|
|
},
|
|
{
|
|
"epoch": 0.014725216933999474,
|
|
"grad_norm": 6.144001483917236,
|
|
"learning_rate": 5.882352941176471e-07,
|
|
"loss": 1.4354,
|
|
"step": 28
|
|
},
|
|
{
|
|
"epoch": 0.015251117538785169,
|
|
"grad_norm": 5.47507905960083,
|
|
"learning_rate": 6.092436974789917e-07,
|
|
"loss": 1.4371,
|
|
"step": 29
|
|
},
|
|
{
|
|
"epoch": 0.015777018143570864,
|
|
"grad_norm": 5.208559989929199,
|
|
"learning_rate": 6.302521008403362e-07,
|
|
"loss": 1.4435,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.01630291874835656,
|
|
"grad_norm": 5.5909929275512695,
|
|
"learning_rate": 6.512605042016808e-07,
|
|
"loss": 1.3242,
|
|
"step": 31
|
|
},
|
|
{
|
|
"epoch": 0.016828819353142257,
|
|
"grad_norm": 5.580185413360596,
|
|
"learning_rate": 6.722689075630252e-07,
|
|
"loss": 1.3361,
|
|
"step": 32
|
|
},
|
|
{
|
|
"epoch": 0.017354719957927953,
|
|
"grad_norm": 5.893317222595215,
|
|
"learning_rate": 6.932773109243699e-07,
|
|
"loss": 1.2841,
|
|
"step": 33
|
|
},
|
|
{
|
|
"epoch": 0.017880620562713646,
|
|
"grad_norm": 5.041568756103516,
|
|
"learning_rate": 7.142857142857143e-07,
|
|
"loss": 1.3983,
|
|
"step": 34
|
|
},
|
|
{
|
|
"epoch": 0.018406521167499343,
|
|
"grad_norm": 5.506172180175781,
|
|
"learning_rate": 7.352941176470589e-07,
|
|
"loss": 1.4013,
|
|
"step": 35
|
|
},
|
|
{
|
|
"epoch": 0.01893242177228504,
|
|
"grad_norm": 4.870342254638672,
|
|
"learning_rate": 7.563025210084034e-07,
|
|
"loss": 1.2991,
|
|
"step": 36
|
|
},
|
|
{
|
|
"epoch": 0.019458322377070732,
|
|
"grad_norm": 4.358093738555908,
|
|
"learning_rate": 7.77310924369748e-07,
|
|
"loss": 1.372,
|
|
"step": 37
|
|
},
|
|
{
|
|
"epoch": 0.01998422298185643,
|
|
"grad_norm": 4.625124931335449,
|
|
"learning_rate": 7.983193277310924e-07,
|
|
"loss": 1.3181,
|
|
"step": 38
|
|
},
|
|
{
|
|
"epoch": 0.020510123586642125,
|
|
"grad_norm": 4.190467357635498,
|
|
"learning_rate": 8.193277310924371e-07,
|
|
"loss": 1.3807,
|
|
"step": 39
|
|
},
|
|
{
|
|
"epoch": 0.02103602419142782,
|
|
"grad_norm": 4.260818004608154,
|
|
"learning_rate": 8.403361344537816e-07,
|
|
"loss": 1.3954,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.021561924796213515,
|
|
"grad_norm": 5.851184844970703,
|
|
"learning_rate": 8.613445378151261e-07,
|
|
"loss": 1.3275,
|
|
"step": 41
|
|
},
|
|
{
|
|
"epoch": 0.02208782540099921,
|
|
"grad_norm": 4.3358683586120605,
|
|
"learning_rate": 8.823529411764707e-07,
|
|
"loss": 1.2826,
|
|
"step": 42
|
|
},
|
|
{
|
|
"epoch": 0.022613726005784908,
|
|
"grad_norm": 4.496443748474121,
|
|
"learning_rate": 9.033613445378152e-07,
|
|
"loss": 1.2972,
|
|
"step": 43
|
|
},
|
|
{
|
|
"epoch": 0.0231396266105706,
|
|
"grad_norm": 4.194251537322998,
|
|
"learning_rate": 9.243697478991598e-07,
|
|
"loss": 1.3098,
|
|
"step": 44
|
|
},
|
|
{
|
|
"epoch": 0.023665527215356297,
|
|
"grad_norm": 4.508425235748291,
|
|
"learning_rate": 9.453781512605043e-07,
|
|
"loss": 1.307,
|
|
"step": 45
|
|
},
|
|
{
|
|
"epoch": 0.024191427820141994,
|
|
"grad_norm": 4.0843682289123535,
|
|
"learning_rate": 9.66386554621849e-07,
|
|
"loss": 1.3442,
|
|
"step": 46
|
|
},
|
|
{
|
|
"epoch": 0.024717328424927687,
|
|
"grad_norm": 4.219754695892334,
|
|
"learning_rate": 9.873949579831934e-07,
|
|
"loss": 1.3489,
|
|
"step": 47
|
|
},
|
|
{
|
|
"epoch": 0.025243229029713384,
|
|
"grad_norm": 3.9498398303985596,
|
|
"learning_rate": 1.0084033613445378e-06,
|
|
"loss": 1.2476,
|
|
"step": 48
|
|
},
|
|
{
|
|
"epoch": 0.02576912963449908,
|
|
"grad_norm": 3.909024953842163,
|
|
"learning_rate": 1.0294117647058825e-06,
|
|
"loss": 1.2408,
|
|
"step": 49
|
|
},
|
|
{
|
|
"epoch": 0.026295030239284777,
|
|
"grad_norm": 4.273558139801025,
|
|
"learning_rate": 1.050420168067227e-06,
|
|
"loss": 1.236,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.02682093084407047,
|
|
"grad_norm": 4.046483516693115,
|
|
"learning_rate": 1.0714285714285714e-06,
|
|
"loss": 1.2726,
|
|
"step": 51
|
|
},
|
|
{
|
|
"epoch": 0.027346831448856166,
|
|
"grad_norm": 3.9503703117370605,
|
|
"learning_rate": 1.092436974789916e-06,
|
|
"loss": 1.3563,
|
|
"step": 52
|
|
},
|
|
{
|
|
"epoch": 0.027872732053641863,
|
|
"grad_norm": 4.400558948516846,
|
|
"learning_rate": 1.1134453781512607e-06,
|
|
"loss": 1.3395,
|
|
"step": 53
|
|
},
|
|
{
|
|
"epoch": 0.028398632658427556,
|
|
"grad_norm": 4.033598899841309,
|
|
"learning_rate": 1.1344537815126051e-06,
|
|
"loss": 1.3143,
|
|
"step": 54
|
|
},
|
|
{
|
|
"epoch": 0.028924533263213252,
|
|
"grad_norm": 3.610640525817871,
|
|
"learning_rate": 1.1554621848739498e-06,
|
|
"loss": 1.3235,
|
|
"step": 55
|
|
},
|
|
{
|
|
"epoch": 0.02945043386799895,
|
|
"grad_norm": 3.798755645751953,
|
|
"learning_rate": 1.1764705882352942e-06,
|
|
"loss": 1.3804,
|
|
"step": 56
|
|
},
|
|
{
|
|
"epoch": 0.029976334472784645,
|
|
"grad_norm": 4.101436138153076,
|
|
"learning_rate": 1.1974789915966389e-06,
|
|
"loss": 1.275,
|
|
"step": 57
|
|
},
|
|
{
|
|
"epoch": 0.030502235077570338,
|
|
"grad_norm": 3.7115893363952637,
|
|
"learning_rate": 1.2184873949579833e-06,
|
|
"loss": 1.1664,
|
|
"step": 58
|
|
},
|
|
{
|
|
"epoch": 0.031028135682356035,
|
|
"grad_norm": 3.9830610752105713,
|
|
"learning_rate": 1.2394957983193278e-06,
|
|
"loss": 1.2959,
|
|
"step": 59
|
|
},
|
|
{
|
|
"epoch": 0.03155403628714173,
|
|
"grad_norm": 3.8779947757720947,
|
|
"learning_rate": 1.2605042016806724e-06,
|
|
"loss": 1.3472,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.032079936891927424,
|
|
"grad_norm": 3.804046154022217,
|
|
"learning_rate": 1.2815126050420169e-06,
|
|
"loss": 1.3735,
|
|
"step": 61
|
|
},
|
|
{
|
|
"epoch": 0.03260583749671312,
|
|
"grad_norm": 3.6065728664398193,
|
|
"learning_rate": 1.3025210084033615e-06,
|
|
"loss": 1.3012,
|
|
"step": 62
|
|
},
|
|
{
|
|
"epoch": 0.03313173810149882,
|
|
"grad_norm": 3.8569400310516357,
|
|
"learning_rate": 1.323529411764706e-06,
|
|
"loss": 1.2677,
|
|
"step": 63
|
|
},
|
|
{
|
|
"epoch": 0.033657638706284514,
|
|
"grad_norm": 3.6886768341064453,
|
|
"learning_rate": 1.3445378151260504e-06,
|
|
"loss": 1.2619,
|
|
"step": 64
|
|
},
|
|
{
|
|
"epoch": 0.03418353931107021,
|
|
"grad_norm": 3.9358179569244385,
|
|
"learning_rate": 1.3655462184873949e-06,
|
|
"loss": 1.3751,
|
|
"step": 65
|
|
},
|
|
{
|
|
"epoch": 0.03470943991585591,
|
|
"grad_norm": 4.193531036376953,
|
|
"learning_rate": 1.3865546218487397e-06,
|
|
"loss": 1.3433,
|
|
"step": 66
|
|
},
|
|
{
|
|
"epoch": 0.035235340520641596,
|
|
"grad_norm": 3.742565631866455,
|
|
"learning_rate": 1.4075630252100842e-06,
|
|
"loss": 1.2714,
|
|
"step": 67
|
|
},
|
|
{
|
|
"epoch": 0.03576124112542729,
|
|
"grad_norm": 3.7572672367095947,
|
|
"learning_rate": 1.4285714285714286e-06,
|
|
"loss": 1.2831,
|
|
"step": 68
|
|
},
|
|
{
|
|
"epoch": 0.03628714173021299,
|
|
"grad_norm": 4.00689172744751,
|
|
"learning_rate": 1.4495798319327733e-06,
|
|
"loss": 1.207,
|
|
"step": 69
|
|
},
|
|
{
|
|
"epoch": 0.036813042334998686,
|
|
"grad_norm": 3.650339365005493,
|
|
"learning_rate": 1.4705882352941177e-06,
|
|
"loss": 1.2977,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.03733894293978438,
|
|
"grad_norm": 3.6257858276367188,
|
|
"learning_rate": 1.4915966386554624e-06,
|
|
"loss": 1.2613,
|
|
"step": 71
|
|
},
|
|
{
|
|
"epoch": 0.03786484354457008,
|
|
"grad_norm": 3.893563747406006,
|
|
"learning_rate": 1.5126050420168068e-06,
|
|
"loss": 1.3507,
|
|
"step": 72
|
|
},
|
|
{
|
|
"epoch": 0.038390744149355775,
|
|
"grad_norm": 3.543694257736206,
|
|
"learning_rate": 1.5336134453781515e-06,
|
|
"loss": 1.1899,
|
|
"step": 73
|
|
},
|
|
{
|
|
"epoch": 0.038916644754141465,
|
|
"grad_norm": 3.6188101768493652,
|
|
"learning_rate": 1.554621848739496e-06,
|
|
"loss": 1.2747,
|
|
"step": 74
|
|
},
|
|
{
|
|
"epoch": 0.03944254535892716,
|
|
"grad_norm": 3.812223196029663,
|
|
"learning_rate": 1.5756302521008404e-06,
|
|
"loss": 1.2941,
|
|
"step": 75
|
|
},
|
|
{
|
|
"epoch": 0.03996844596371286,
|
|
"grad_norm": 3.878098487854004,
|
|
"learning_rate": 1.5966386554621848e-06,
|
|
"loss": 1.306,
|
|
"step": 76
|
|
},
|
|
{
|
|
"epoch": 0.040494346568498554,
|
|
"grad_norm": 3.664268732070923,
|
|
"learning_rate": 1.6176470588235297e-06,
|
|
"loss": 1.2859,
|
|
"step": 77
|
|
},
|
|
{
|
|
"epoch": 0.04102024717328425,
|
|
"grad_norm": 3.751523017883301,
|
|
"learning_rate": 1.6386554621848741e-06,
|
|
"loss": 1.3417,
|
|
"step": 78
|
|
},
|
|
{
|
|
"epoch": 0.04154614777806995,
|
|
"grad_norm": 3.727069616317749,
|
|
"learning_rate": 1.6596638655462186e-06,
|
|
"loss": 1.277,
|
|
"step": 79
|
|
},
|
|
{
|
|
"epoch": 0.04207204838285564,
|
|
"grad_norm": 3.8709964752197266,
|
|
"learning_rate": 1.6806722689075632e-06,
|
|
"loss": 1.271,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.04259794898764133,
|
|
"grad_norm": 3.95377516746521,
|
|
"learning_rate": 1.7016806722689077e-06,
|
|
"loss": 1.2106,
|
|
"step": 81
|
|
},
|
|
{
|
|
"epoch": 0.04312384959242703,
|
|
"grad_norm": 3.841933012008667,
|
|
"learning_rate": 1.7226890756302521e-06,
|
|
"loss": 1.2871,
|
|
"step": 82
|
|
},
|
|
{
|
|
"epoch": 0.043649750197212726,
|
|
"grad_norm": 3.9010164737701416,
|
|
"learning_rate": 1.7436974789915968e-06,
|
|
"loss": 1.2608,
|
|
"step": 83
|
|
},
|
|
{
|
|
"epoch": 0.04417565080199842,
|
|
"grad_norm": 3.6362926959991455,
|
|
"learning_rate": 1.7647058823529414e-06,
|
|
"loss": 1.1109,
|
|
"step": 84
|
|
},
|
|
{
|
|
"epoch": 0.04470155140678412,
|
|
"grad_norm": 3.612722396850586,
|
|
"learning_rate": 1.7857142857142859e-06,
|
|
"loss": 1.2176,
|
|
"step": 85
|
|
},
|
|
{
|
|
"epoch": 0.045227452011569816,
|
|
"grad_norm": 3.901843547821045,
|
|
"learning_rate": 1.8067226890756303e-06,
|
|
"loss": 1.2323,
|
|
"step": 86
|
|
},
|
|
{
|
|
"epoch": 0.045753352616355505,
|
|
"grad_norm": 3.6690895557403564,
|
|
"learning_rate": 1.8277310924369748e-06,
|
|
"loss": 1.1793,
|
|
"step": 87
|
|
},
|
|
{
|
|
"epoch": 0.0462792532211412,
|
|
"grad_norm": 3.7964813709259033,
|
|
"learning_rate": 1.8487394957983196e-06,
|
|
"loss": 1.3329,
|
|
"step": 88
|
|
},
|
|
{
|
|
"epoch": 0.0468051538259269,
|
|
"grad_norm": 3.670017957687378,
|
|
"learning_rate": 1.869747899159664e-06,
|
|
"loss": 1.2572,
|
|
"step": 89
|
|
},
|
|
{
|
|
"epoch": 0.047331054430712595,
|
|
"grad_norm": 3.7981204986572266,
|
|
"learning_rate": 1.8907563025210085e-06,
|
|
"loss": 1.337,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.04785695503549829,
|
|
"grad_norm": 3.8516929149627686,
|
|
"learning_rate": 1.9117647058823528e-06,
|
|
"loss": 1.3476,
|
|
"step": 91
|
|
},
|
|
{
|
|
"epoch": 0.04838285564028399,
|
|
"grad_norm": 3.6338279247283936,
|
|
"learning_rate": 1.932773109243698e-06,
|
|
"loss": 1.2601,
|
|
"step": 92
|
|
},
|
|
{
|
|
"epoch": 0.048908756245069684,
|
|
"grad_norm": 3.5624072551727295,
|
|
"learning_rate": 1.953781512605042e-06,
|
|
"loss": 1.2877,
|
|
"step": 93
|
|
},
|
|
{
|
|
"epoch": 0.049434656849855374,
|
|
"grad_norm": 3.6358585357666016,
|
|
"learning_rate": 1.9747899159663867e-06,
|
|
"loss": 1.1658,
|
|
"step": 94
|
|
},
|
|
{
|
|
"epoch": 0.04996055745464107,
|
|
"grad_norm": 4.260152816772461,
|
|
"learning_rate": 1.9957983193277314e-06,
|
|
"loss": 1.3132,
|
|
"step": 95
|
|
},
|
|
{
|
|
"epoch": 0.05048645805942677,
|
|
"grad_norm": 3.341580629348755,
|
|
"learning_rate": 2.0168067226890756e-06,
|
|
"loss": 1.172,
|
|
"step": 96
|
|
},
|
|
{
|
|
"epoch": 0.051012358664212464,
|
|
"grad_norm": 3.785923719406128,
|
|
"learning_rate": 2.0378151260504203e-06,
|
|
"loss": 1.3062,
|
|
"step": 97
|
|
},
|
|
{
|
|
"epoch": 0.05153825926899816,
|
|
"grad_norm": 4.143803596496582,
|
|
"learning_rate": 2.058823529411765e-06,
|
|
"loss": 1.2669,
|
|
"step": 98
|
|
},
|
|
{
|
|
"epoch": 0.052064159873783857,
|
|
"grad_norm": 3.6901350021362305,
|
|
"learning_rate": 2.0798319327731096e-06,
|
|
"loss": 1.3231,
|
|
"step": 99
|
|
},
|
|
{
|
|
"epoch": 0.05259006047856955,
|
|
"grad_norm": 3.896963119506836,
|
|
"learning_rate": 2.100840336134454e-06,
|
|
"loss": 1.2137,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.05311596108335524,
|
|
"grad_norm": 4.068558692932129,
|
|
"learning_rate": 2.1218487394957985e-06,
|
|
"loss": 1.358,
|
|
"step": 101
|
|
},
|
|
{
|
|
"epoch": 0.05364186168814094,
|
|
"grad_norm": 3.573665142059326,
|
|
"learning_rate": 2.1428571428571427e-06,
|
|
"loss": 1.2473,
|
|
"step": 102
|
|
},
|
|
{
|
|
"epoch": 0.054167762292926636,
|
|
"grad_norm": 4.117169380187988,
|
|
"learning_rate": 2.163865546218488e-06,
|
|
"loss": 1.2523,
|
|
"step": 103
|
|
},
|
|
{
|
|
"epoch": 0.05469366289771233,
|
|
"grad_norm": 3.93593168258667,
|
|
"learning_rate": 2.184873949579832e-06,
|
|
"loss": 1.2545,
|
|
"step": 104
|
|
},
|
|
{
|
|
"epoch": 0.05521956350249803,
|
|
"grad_norm": 4.127479076385498,
|
|
"learning_rate": 2.2058823529411767e-06,
|
|
"loss": 1.1622,
|
|
"step": 105
|
|
},
|
|
{
|
|
"epoch": 0.055745464107283725,
|
|
"grad_norm": 3.729445219039917,
|
|
"learning_rate": 2.2268907563025214e-06,
|
|
"loss": 1.1717,
|
|
"step": 106
|
|
},
|
|
{
|
|
"epoch": 0.05627136471206942,
|
|
"grad_norm": 3.664088726043701,
|
|
"learning_rate": 2.2478991596638656e-06,
|
|
"loss": 1.1698,
|
|
"step": 107
|
|
},
|
|
{
|
|
"epoch": 0.05679726531685511,
|
|
"grad_norm": 3.9456894397735596,
|
|
"learning_rate": 2.2689075630252102e-06,
|
|
"loss": 1.242,
|
|
"step": 108
|
|
},
|
|
{
|
|
"epoch": 0.05732316592164081,
|
|
"grad_norm": 3.6314444541931152,
|
|
"learning_rate": 2.289915966386555e-06,
|
|
"loss": 1.3059,
|
|
"step": 109
|
|
},
|
|
{
|
|
"epoch": 0.057849066526426504,
|
|
"grad_norm": 3.695945978164673,
|
|
"learning_rate": 2.3109243697478996e-06,
|
|
"loss": 1.2123,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.0583749671312122,
|
|
"grad_norm": 3.625239610671997,
|
|
"learning_rate": 2.331932773109244e-06,
|
|
"loss": 1.1479,
|
|
"step": 111
|
|
},
|
|
{
|
|
"epoch": 0.0589008677359979,
|
|
"grad_norm": 3.5934529304504395,
|
|
"learning_rate": 2.3529411764705885e-06,
|
|
"loss": 1.2274,
|
|
"step": 112
|
|
},
|
|
{
|
|
"epoch": 0.059426768340783594,
|
|
"grad_norm": 4.339212894439697,
|
|
"learning_rate": 2.3739495798319327e-06,
|
|
"loss": 1.2107,
|
|
"step": 113
|
|
},
|
|
{
|
|
"epoch": 0.05995266894556929,
|
|
"grad_norm": 3.587738513946533,
|
|
"learning_rate": 2.3949579831932778e-06,
|
|
"loss": 1.2746,
|
|
"step": 114
|
|
},
|
|
{
|
|
"epoch": 0.06047856955035498,
|
|
"grad_norm": 3.558872938156128,
|
|
"learning_rate": 2.415966386554622e-06,
|
|
"loss": 1.2316,
|
|
"step": 115
|
|
},
|
|
{
|
|
"epoch": 0.061004470155140676,
|
|
"grad_norm": 3.7241933345794678,
|
|
"learning_rate": 2.4369747899159667e-06,
|
|
"loss": 1.2436,
|
|
"step": 116
|
|
},
|
|
{
|
|
"epoch": 0.06153037075992637,
|
|
"grad_norm": 4.053774356842041,
|
|
"learning_rate": 2.457983193277311e-06,
|
|
"loss": 1.3185,
|
|
"step": 117
|
|
},
|
|
{
|
|
"epoch": 0.06205627136471207,
|
|
"grad_norm": 3.832151412963867,
|
|
"learning_rate": 2.4789915966386555e-06,
|
|
"loss": 1.2738,
|
|
"step": 118
|
|
},
|
|
{
|
|
"epoch": 0.06258217196949777,
|
|
"grad_norm": 3.757659912109375,
|
|
"learning_rate": 2.5e-06,
|
|
"loss": 1.1898,
|
|
"step": 119
|
|
},
|
|
{
|
|
"epoch": 0.06310807257428346,
|
|
"grad_norm": 3.5104820728302,
|
|
"learning_rate": 2.521008403361345e-06,
|
|
"loss": 1.2254,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.06363397317906916,
|
|
"grad_norm": 3.90678071975708,
|
|
"learning_rate": 2.542016806722689e-06,
|
|
"loss": 1.251,
|
|
"step": 121
|
|
},
|
|
{
|
|
"epoch": 0.06415987378385485,
|
|
"grad_norm": 3.4690308570861816,
|
|
"learning_rate": 2.5630252100840338e-06,
|
|
"loss": 1.1946,
|
|
"step": 122
|
|
},
|
|
{
|
|
"epoch": 0.06468577438864055,
|
|
"grad_norm": 3.535465955734253,
|
|
"learning_rate": 2.5840336134453784e-06,
|
|
"loss": 1.2449,
|
|
"step": 123
|
|
},
|
|
{
|
|
"epoch": 0.06521167499342624,
|
|
"grad_norm": 3.671808958053589,
|
|
"learning_rate": 2.605042016806723e-06,
|
|
"loss": 1.2961,
|
|
"step": 124
|
|
},
|
|
{
|
|
"epoch": 0.06573757559821194,
|
|
"grad_norm": 3.796097755432129,
|
|
"learning_rate": 2.6260504201680677e-06,
|
|
"loss": 1.216,
|
|
"step": 125
|
|
},
|
|
{
|
|
"epoch": 0.06626347620299763,
|
|
"grad_norm": 4.2561726570129395,
|
|
"learning_rate": 2.647058823529412e-06,
|
|
"loss": 1.1677,
|
|
"step": 126
|
|
},
|
|
{
|
|
"epoch": 0.06678937680778332,
|
|
"grad_norm": 3.745861768722534,
|
|
"learning_rate": 2.6680672268907566e-06,
|
|
"loss": 1.1591,
|
|
"step": 127
|
|
},
|
|
{
|
|
"epoch": 0.06731527741256903,
|
|
"grad_norm": 4.890429973602295,
|
|
"learning_rate": 2.689075630252101e-06,
|
|
"loss": 1.2631,
|
|
"step": 128
|
|
},
|
|
{
|
|
"epoch": 0.06784117801735472,
|
|
"grad_norm": 3.587217092514038,
|
|
"learning_rate": 2.7100840336134455e-06,
|
|
"loss": 1.2181,
|
|
"step": 129
|
|
},
|
|
{
|
|
"epoch": 0.06836707862214042,
|
|
"grad_norm": 3.594881534576416,
|
|
"learning_rate": 2.7310924369747897e-06,
|
|
"loss": 1.3107,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.06889297922692611,
|
|
"grad_norm": 4.74546480178833,
|
|
"learning_rate": 2.752100840336135e-06,
|
|
"loss": 1.3433,
|
|
"step": 131
|
|
},
|
|
{
|
|
"epoch": 0.06941887983171181,
|
|
"grad_norm": 4.44760274887085,
|
|
"learning_rate": 2.7731092436974795e-06,
|
|
"loss": 1.208,
|
|
"step": 132
|
|
},
|
|
{
|
|
"epoch": 0.0699447804364975,
|
|
"grad_norm": 3.9094603061676025,
|
|
"learning_rate": 2.7941176470588237e-06,
|
|
"loss": 1.1611,
|
|
"step": 133
|
|
},
|
|
{
|
|
"epoch": 0.07047068104128319,
|
|
"grad_norm": 3.731175422668457,
|
|
"learning_rate": 2.8151260504201684e-06,
|
|
"loss": 1.216,
|
|
"step": 134
|
|
},
|
|
{
|
|
"epoch": 0.0709965816460689,
|
|
"grad_norm": 4.075158596038818,
|
|
"learning_rate": 2.8361344537815126e-06,
|
|
"loss": 1.3114,
|
|
"step": 135
|
|
},
|
|
{
|
|
"epoch": 0.07152248225085459,
|
|
"grad_norm": 3.8945798873901367,
|
|
"learning_rate": 2.8571428571428573e-06,
|
|
"loss": 1.1328,
|
|
"step": 136
|
|
},
|
|
{
|
|
"epoch": 0.07204838285564029,
|
|
"grad_norm": 4.227399826049805,
|
|
"learning_rate": 2.8781512605042015e-06,
|
|
"loss": 1.2106,
|
|
"step": 137
|
|
},
|
|
{
|
|
"epoch": 0.07257428346042598,
|
|
"grad_norm": 4.031093120574951,
|
|
"learning_rate": 2.8991596638655466e-06,
|
|
"loss": 1.1766,
|
|
"step": 138
|
|
},
|
|
{
|
|
"epoch": 0.07310018406521168,
|
|
"grad_norm": 3.613487720489502,
|
|
"learning_rate": 2.9201680672268912e-06,
|
|
"loss": 1.2015,
|
|
"step": 139
|
|
},
|
|
{
|
|
"epoch": 0.07362608466999737,
|
|
"grad_norm": 4.50424861907959,
|
|
"learning_rate": 2.9411764705882355e-06,
|
|
"loss": 1.2345,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.07415198527478306,
|
|
"grad_norm": 3.827760934829712,
|
|
"learning_rate": 2.96218487394958e-06,
|
|
"loss": 1.1748,
|
|
"step": 141
|
|
},
|
|
{
|
|
"epoch": 0.07467788587956876,
|
|
"grad_norm": 3.9251503944396973,
|
|
"learning_rate": 2.9831932773109248e-06,
|
|
"loss": 1.4392,
|
|
"step": 142
|
|
},
|
|
{
|
|
"epoch": 0.07520378648435445,
|
|
"grad_norm": 4.420262336730957,
|
|
"learning_rate": 3.004201680672269e-06,
|
|
"loss": 1.1702,
|
|
"step": 143
|
|
},
|
|
{
|
|
"epoch": 0.07572968708914016,
|
|
"grad_norm": 3.731318950653076,
|
|
"learning_rate": 3.0252100840336137e-06,
|
|
"loss": 1.2413,
|
|
"step": 144
|
|
},
|
|
{
|
|
"epoch": 0.07625558769392585,
|
|
"grad_norm": 3.60262393951416,
|
|
"learning_rate": 3.046218487394958e-06,
|
|
"loss": 1.1632,
|
|
"step": 145
|
|
},
|
|
{
|
|
"epoch": 0.07678148829871155,
|
|
"grad_norm": 3.513679265975952,
|
|
"learning_rate": 3.067226890756303e-06,
|
|
"loss": 1.1187,
|
|
"step": 146
|
|
},
|
|
{
|
|
"epoch": 0.07730738890349724,
|
|
"grad_norm": 3.6981728076934814,
|
|
"learning_rate": 3.0882352941176476e-06,
|
|
"loss": 1.1554,
|
|
"step": 147
|
|
},
|
|
{
|
|
"epoch": 0.07783328950828293,
|
|
"grad_norm": 3.7815120220184326,
|
|
"learning_rate": 3.109243697478992e-06,
|
|
"loss": 1.1734,
|
|
"step": 148
|
|
},
|
|
{
|
|
"epoch": 0.07835919011306863,
|
|
"grad_norm": 3.5924789905548096,
|
|
"learning_rate": 3.1302521008403365e-06,
|
|
"loss": 1.1434,
|
|
"step": 149
|
|
},
|
|
{
|
|
"epoch": 0.07888509071785432,
|
|
"grad_norm": 3.626119613647461,
|
|
"learning_rate": 3.1512605042016808e-06,
|
|
"loss": 1.1509,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.07941099132264003,
|
|
"grad_norm": 3.8956491947174072,
|
|
"learning_rate": 3.1722689075630254e-06,
|
|
"loss": 1.3083,
|
|
"step": 151
|
|
},
|
|
{
|
|
"epoch": 0.07993689192742572,
|
|
"grad_norm": 3.8392393589019775,
|
|
"learning_rate": 3.1932773109243696e-06,
|
|
"loss": 1.2639,
|
|
"step": 152
|
|
},
|
|
{
|
|
"epoch": 0.0804627925322114,
|
|
"grad_norm": 3.925689697265625,
|
|
"learning_rate": 3.2142857142857147e-06,
|
|
"loss": 1.3541,
|
|
"step": 153
|
|
},
|
|
{
|
|
"epoch": 0.08098869313699711,
|
|
"grad_norm": 3.6600394248962402,
|
|
"learning_rate": 3.2352941176470594e-06,
|
|
"loss": 1.2265,
|
|
"step": 154
|
|
},
|
|
{
|
|
"epoch": 0.0815145937417828,
|
|
"grad_norm": 3.9275174140930176,
|
|
"learning_rate": 3.2563025210084036e-06,
|
|
"loss": 1.2366,
|
|
"step": 155
|
|
},
|
|
{
|
|
"epoch": 0.0820404943465685,
|
|
"grad_norm": 3.905914783477783,
|
|
"learning_rate": 3.2773109243697483e-06,
|
|
"loss": 1.2398,
|
|
"step": 156
|
|
},
|
|
{
|
|
"epoch": 0.08256639495135419,
|
|
"grad_norm": 3.8108930587768555,
|
|
"learning_rate": 3.2983193277310925e-06,
|
|
"loss": 1.2768,
|
|
"step": 157
|
|
},
|
|
{
|
|
"epoch": 0.0830922955561399,
|
|
"grad_norm": 3.8421032428741455,
|
|
"learning_rate": 3.319327731092437e-06,
|
|
"loss": 1.2458,
|
|
"step": 158
|
|
},
|
|
{
|
|
"epoch": 0.08361819616092558,
|
|
"grad_norm": 3.738586187362671,
|
|
"learning_rate": 3.3403361344537814e-06,
|
|
"loss": 1.3113,
|
|
"step": 159
|
|
},
|
|
{
|
|
"epoch": 0.08414409676571127,
|
|
"grad_norm": 3.6991453170776367,
|
|
"learning_rate": 3.3613445378151265e-06,
|
|
"loss": 1.292,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.08466999737049698,
|
|
"grad_norm": 3.849764347076416,
|
|
"learning_rate": 3.382352941176471e-06,
|
|
"loss": 1.2657,
|
|
"step": 161
|
|
},
|
|
{
|
|
"epoch": 0.08519589797528267,
|
|
"grad_norm": 3.639394521713257,
|
|
"learning_rate": 3.4033613445378154e-06,
|
|
"loss": 1.2088,
|
|
"step": 162
|
|
},
|
|
{
|
|
"epoch": 0.08572179858006837,
|
|
"grad_norm": 3.486828088760376,
|
|
"learning_rate": 3.42436974789916e-06,
|
|
"loss": 1.2339,
|
|
"step": 163
|
|
},
|
|
{
|
|
"epoch": 0.08624769918485406,
|
|
"grad_norm": 3.695509672164917,
|
|
"learning_rate": 3.4453781512605043e-06,
|
|
"loss": 1.1735,
|
|
"step": 164
|
|
},
|
|
{
|
|
"epoch": 0.08677359978963976,
|
|
"grad_norm": 3.5470075607299805,
|
|
"learning_rate": 3.466386554621849e-06,
|
|
"loss": 1.2415,
|
|
"step": 165
|
|
},
|
|
{
|
|
"epoch": 0.08729950039442545,
|
|
"grad_norm": 3.863962411880493,
|
|
"learning_rate": 3.4873949579831936e-06,
|
|
"loss": 1.2642,
|
|
"step": 166
|
|
},
|
|
{
|
|
"epoch": 0.08782540099921114,
|
|
"grad_norm": 3.804619550704956,
|
|
"learning_rate": 3.508403361344538e-06,
|
|
"loss": 1.1933,
|
|
"step": 167
|
|
},
|
|
{
|
|
"epoch": 0.08835130160399685,
|
|
"grad_norm": 3.7575409412384033,
|
|
"learning_rate": 3.529411764705883e-06,
|
|
"loss": 1.2255,
|
|
"step": 168
|
|
},
|
|
{
|
|
"epoch": 0.08887720220878254,
|
|
"grad_norm": 3.7897167205810547,
|
|
"learning_rate": 3.550420168067227e-06,
|
|
"loss": 1.2385,
|
|
"step": 169
|
|
},
|
|
{
|
|
"epoch": 0.08940310281356824,
|
|
"grad_norm": 3.8045732975006104,
|
|
"learning_rate": 3.5714285714285718e-06,
|
|
"loss": 1.2222,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.08992900341835393,
|
|
"grad_norm": 3.9293887615203857,
|
|
"learning_rate": 3.5924369747899164e-06,
|
|
"loss": 1.145,
|
|
"step": 171
|
|
},
|
|
{
|
|
"epoch": 0.09045490402313963,
|
|
"grad_norm": 3.499516487121582,
|
|
"learning_rate": 3.6134453781512607e-06,
|
|
"loss": 1.261,
|
|
"step": 172
|
|
},
|
|
{
|
|
"epoch": 0.09098080462792532,
|
|
"grad_norm": 3.7555673122406006,
|
|
"learning_rate": 3.6344537815126053e-06,
|
|
"loss": 1.3161,
|
|
"step": 173
|
|
},
|
|
{
|
|
"epoch": 0.09150670523271101,
|
|
"grad_norm": 3.5064549446105957,
|
|
"learning_rate": 3.6554621848739496e-06,
|
|
"loss": 1.2329,
|
|
"step": 174
|
|
},
|
|
{
|
|
"epoch": 0.09203260583749671,
|
|
"grad_norm": 3.658947467803955,
|
|
"learning_rate": 3.6764705882352946e-06,
|
|
"loss": 1.2809,
|
|
"step": 175
|
|
},
|
|
{
|
|
"epoch": 0.0925585064422824,
|
|
"grad_norm": 3.804522752761841,
|
|
"learning_rate": 3.6974789915966393e-06,
|
|
"loss": 1.2044,
|
|
"step": 176
|
|
},
|
|
{
|
|
"epoch": 0.09308440704706811,
|
|
"grad_norm": 3.4374892711639404,
|
|
"learning_rate": 3.7184873949579835e-06,
|
|
"loss": 1.1522,
|
|
"step": 177
|
|
},
|
|
{
|
|
"epoch": 0.0936103076518538,
|
|
"grad_norm": 3.721381425857544,
|
|
"learning_rate": 3.739495798319328e-06,
|
|
"loss": 1.2353,
|
|
"step": 178
|
|
},
|
|
{
|
|
"epoch": 0.0941362082566395,
|
|
"grad_norm": 3.4102025032043457,
|
|
"learning_rate": 3.7605042016806724e-06,
|
|
"loss": 1.2997,
|
|
"step": 179
|
|
},
|
|
{
|
|
"epoch": 0.09466210886142519,
|
|
"grad_norm": 3.507432699203491,
|
|
"learning_rate": 3.781512605042017e-06,
|
|
"loss": 1.2067,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.09518800946621088,
|
|
"grad_norm": 3.4599626064300537,
|
|
"learning_rate": 3.8025210084033613e-06,
|
|
"loss": 1.2015,
|
|
"step": 181
|
|
},
|
|
{
|
|
"epoch": 0.09571391007099658,
|
|
"grad_norm": 3.514089584350586,
|
|
"learning_rate": 3.8235294117647055e-06,
|
|
"loss": 1.2022,
|
|
"step": 182
|
|
},
|
|
{
|
|
"epoch": 0.09623981067578227,
|
|
"grad_norm": 3.543154239654541,
|
|
"learning_rate": 3.844537815126051e-06,
|
|
"loss": 1.2066,
|
|
"step": 183
|
|
},
|
|
{
|
|
"epoch": 0.09676571128056798,
|
|
"grad_norm": 3.556584596633911,
|
|
"learning_rate": 3.865546218487396e-06,
|
|
"loss": 1.1981,
|
|
"step": 184
|
|
},
|
|
{
|
|
"epoch": 0.09729161188535367,
|
|
"grad_norm": 3.6025426387786865,
|
|
"learning_rate": 3.8865546218487395e-06,
|
|
"loss": 1.2051,
|
|
"step": 185
|
|
},
|
|
{
|
|
"epoch": 0.09781751249013937,
|
|
"grad_norm": 3.7175467014312744,
|
|
"learning_rate": 3.907563025210084e-06,
|
|
"loss": 1.2132,
|
|
"step": 186
|
|
},
|
|
{
|
|
"epoch": 0.09834341309492506,
|
|
"grad_norm": 3.4796833992004395,
|
|
"learning_rate": 3.928571428571429e-06,
|
|
"loss": 1.2783,
|
|
"step": 187
|
|
},
|
|
{
|
|
"epoch": 0.09886931369971075,
|
|
"grad_norm": 3.7726001739501953,
|
|
"learning_rate": 3.9495798319327735e-06,
|
|
"loss": 1.2287,
|
|
"step": 188
|
|
},
|
|
{
|
|
"epoch": 0.09939521430449645,
|
|
"grad_norm": 4.104244232177734,
|
|
"learning_rate": 3.970588235294118e-06,
|
|
"loss": 1.3332,
|
|
"step": 189
|
|
},
|
|
{
|
|
"epoch": 0.09992111490928214,
|
|
"grad_norm": 3.625800848007202,
|
|
"learning_rate": 3.991596638655463e-06,
|
|
"loss": 1.3385,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.10044701551406784,
|
|
"grad_norm": 3.666543960571289,
|
|
"learning_rate": 4.0126050420168075e-06,
|
|
"loss": 1.2193,
|
|
"step": 191
|
|
},
|
|
{
|
|
"epoch": 0.10097291611885353,
|
|
"grad_norm": 3.689282178878784,
|
|
"learning_rate": 4.033613445378151e-06,
|
|
"loss": 1.0812,
|
|
"step": 192
|
|
},
|
|
{
|
|
"epoch": 0.10149881672363924,
|
|
"grad_norm": 3.7951157093048096,
|
|
"learning_rate": 4.054621848739496e-06,
|
|
"loss": 1.154,
|
|
"step": 193
|
|
},
|
|
{
|
|
"epoch": 0.10202471732842493,
|
|
"grad_norm": 4.021623134613037,
|
|
"learning_rate": 4.075630252100841e-06,
|
|
"loss": 1.3738,
|
|
"step": 194
|
|
},
|
|
{
|
|
"epoch": 0.10255061793321062,
|
|
"grad_norm": 3.6783387660980225,
|
|
"learning_rate": 4.096638655462185e-06,
|
|
"loss": 1.2919,
|
|
"step": 195
|
|
},
|
|
{
|
|
"epoch": 0.10307651853799632,
|
|
"grad_norm": 3.5767123699188232,
|
|
"learning_rate": 4.11764705882353e-06,
|
|
"loss": 1.2121,
|
|
"step": 196
|
|
},
|
|
{
|
|
"epoch": 0.10360241914278201,
|
|
"grad_norm": 3.512733221054077,
|
|
"learning_rate": 4.138655462184874e-06,
|
|
"loss": 1.2519,
|
|
"step": 197
|
|
},
|
|
{
|
|
"epoch": 0.10412831974756771,
|
|
"grad_norm": 3.793973922729492,
|
|
"learning_rate": 4.159663865546219e-06,
|
|
"loss": 1.271,
|
|
"step": 198
|
|
},
|
|
{
|
|
"epoch": 0.1046542203523534,
|
|
"grad_norm": 3.6994495391845703,
|
|
"learning_rate": 4.180672268907564e-06,
|
|
"loss": 1.3392,
|
|
"step": 199
|
|
},
|
|
{
|
|
"epoch": 0.1051801209571391,
|
|
"grad_norm": 3.399963140487671,
|
|
"learning_rate": 4.201680672268908e-06,
|
|
"loss": 1.1958,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.1057060215619248,
|
|
"grad_norm": 3.6358935832977295,
|
|
"learning_rate": 4.222689075630252e-06,
|
|
"loss": 1.1853,
|
|
"step": 201
|
|
},
|
|
{
|
|
"epoch": 0.10623192216671049,
|
|
"grad_norm": 3.4237492084503174,
|
|
"learning_rate": 4.243697478991597e-06,
|
|
"loss": 1.2506,
|
|
"step": 202
|
|
},
|
|
{
|
|
"epoch": 0.10675782277149619,
|
|
"grad_norm": 3.539459466934204,
|
|
"learning_rate": 4.264705882352942e-06,
|
|
"loss": 1.2299,
|
|
"step": 203
|
|
},
|
|
{
|
|
"epoch": 0.10728372337628188,
|
|
"grad_norm": 3.8455276489257812,
|
|
"learning_rate": 4.2857142857142855e-06,
|
|
"loss": 1.2794,
|
|
"step": 204
|
|
},
|
|
{
|
|
"epoch": 0.10780962398106758,
|
|
"grad_norm": 3.849560260772705,
|
|
"learning_rate": 4.306722689075631e-06,
|
|
"loss": 1.1586,
|
|
"step": 205
|
|
},
|
|
{
|
|
"epoch": 0.10833552458585327,
|
|
"grad_norm": 3.576690673828125,
|
|
"learning_rate": 4.327731092436976e-06,
|
|
"loss": 1.1681,
|
|
"step": 206
|
|
},
|
|
{
|
|
"epoch": 0.10886142519063897,
|
|
"grad_norm": 3.4814934730529785,
|
|
"learning_rate": 4.3487394957983194e-06,
|
|
"loss": 1.1492,
|
|
"step": 207
|
|
},
|
|
{
|
|
"epoch": 0.10938732579542466,
|
|
"grad_norm": 3.7272002696990967,
|
|
"learning_rate": 4.369747899159664e-06,
|
|
"loss": 1.1062,
|
|
"step": 208
|
|
},
|
|
{
|
|
"epoch": 0.10991322640021035,
|
|
"grad_norm": 3.6134095191955566,
|
|
"learning_rate": 4.390756302521009e-06,
|
|
"loss": 1.1727,
|
|
"step": 209
|
|
},
|
|
{
|
|
"epoch": 0.11043912700499606,
|
|
"grad_norm": 3.567073345184326,
|
|
"learning_rate": 4.411764705882353e-06,
|
|
"loss": 1.2827,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.11096502760978175,
|
|
"grad_norm": 3.8407320976257324,
|
|
"learning_rate": 4.432773109243698e-06,
|
|
"loss": 1.2433,
|
|
"step": 211
|
|
},
|
|
{
|
|
"epoch": 0.11149092821456745,
|
|
"grad_norm": 3.4349818229675293,
|
|
"learning_rate": 4.453781512605043e-06,
|
|
"loss": 1.219,
|
|
"step": 212
|
|
},
|
|
{
|
|
"epoch": 0.11201682881935314,
|
|
"grad_norm": 3.6192538738250732,
|
|
"learning_rate": 4.474789915966387e-06,
|
|
"loss": 1.2087,
|
|
"step": 213
|
|
},
|
|
{
|
|
"epoch": 0.11254272942413884,
|
|
"grad_norm": 3.8862030506134033,
|
|
"learning_rate": 4.495798319327731e-06,
|
|
"loss": 1.2022,
|
|
"step": 214
|
|
},
|
|
{
|
|
"epoch": 0.11306863002892453,
|
|
"grad_norm": 3.509019374847412,
|
|
"learning_rate": 4.516806722689076e-06,
|
|
"loss": 1.1656,
|
|
"step": 215
|
|
},
|
|
{
|
|
"epoch": 0.11359453063371022,
|
|
"grad_norm": 3.881326198577881,
|
|
"learning_rate": 4.5378151260504205e-06,
|
|
"loss": 1.243,
|
|
"step": 216
|
|
},
|
|
{
|
|
"epoch": 0.11412043123849593,
|
|
"grad_norm": 3.7966248989105225,
|
|
"learning_rate": 4.558823529411765e-06,
|
|
"loss": 1.1968,
|
|
"step": 217
|
|
},
|
|
{
|
|
"epoch": 0.11464633184328162,
|
|
"grad_norm": 3.7038373947143555,
|
|
"learning_rate": 4.57983193277311e-06,
|
|
"loss": 1.1756,
|
|
"step": 218
|
|
},
|
|
{
|
|
"epoch": 0.11517223244806732,
|
|
"grad_norm": 3.7482972145080566,
|
|
"learning_rate": 4.600840336134454e-06,
|
|
"loss": 1.3234,
|
|
"step": 219
|
|
},
|
|
{
|
|
"epoch": 0.11569813305285301,
|
|
"grad_norm": 3.7005937099456787,
|
|
"learning_rate": 4.621848739495799e-06,
|
|
"loss": 1.3358,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.11622403365763871,
|
|
"grad_norm": 3.3589909076690674,
|
|
"learning_rate": 4.642857142857144e-06,
|
|
"loss": 1.1711,
|
|
"step": 221
|
|
},
|
|
{
|
|
"epoch": 0.1167499342624244,
|
|
"grad_norm": 3.6279635429382324,
|
|
"learning_rate": 4.663865546218488e-06,
|
|
"loss": 1.2837,
|
|
"step": 222
|
|
},
|
|
{
|
|
"epoch": 0.11727583486721009,
|
|
"grad_norm": 3.5633528232574463,
|
|
"learning_rate": 4.684873949579832e-06,
|
|
"loss": 1.2683,
|
|
"step": 223
|
|
},
|
|
{
|
|
"epoch": 0.1178017354719958,
|
|
"grad_norm": 3.5712976455688477,
|
|
"learning_rate": 4.705882352941177e-06,
|
|
"loss": 1.179,
|
|
"step": 224
|
|
},
|
|
{
|
|
"epoch": 0.11832763607678148,
|
|
"grad_norm": 3.625666618347168,
|
|
"learning_rate": 4.7268907563025216e-06,
|
|
"loss": 1.1401,
|
|
"step": 225
|
|
},
|
|
{
|
|
"epoch": 0.11885353668156719,
|
|
"grad_norm": 3.8824515342712402,
|
|
"learning_rate": 4.747899159663865e-06,
|
|
"loss": 1.259,
|
|
"step": 226
|
|
},
|
|
{
|
|
"epoch": 0.11937943728635288,
|
|
"grad_norm": 3.759141206741333,
|
|
"learning_rate": 4.768907563025211e-06,
|
|
"loss": 1.2644,
|
|
"step": 227
|
|
},
|
|
{
|
|
"epoch": 0.11990533789113858,
|
|
"grad_norm": 3.6927576065063477,
|
|
"learning_rate": 4.7899159663865555e-06,
|
|
"loss": 1.2812,
|
|
"step": 228
|
|
},
|
|
{
|
|
"epoch": 0.12043123849592427,
|
|
"grad_norm": 3.827273368835449,
|
|
"learning_rate": 4.810924369747899e-06,
|
|
"loss": 1.2071,
|
|
"step": 229
|
|
},
|
|
{
|
|
"epoch": 0.12095713910070996,
|
|
"grad_norm": 3.4929311275482178,
|
|
"learning_rate": 4.831932773109244e-06,
|
|
"loss": 1.2014,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.12148303970549566,
|
|
"grad_norm": 3.8108766078948975,
|
|
"learning_rate": 4.852941176470589e-06,
|
|
"loss": 1.3057,
|
|
"step": 231
|
|
},
|
|
{
|
|
"epoch": 0.12200894031028135,
|
|
"grad_norm": 4.087435245513916,
|
|
"learning_rate": 4.873949579831933e-06,
|
|
"loss": 1.2873,
|
|
"step": 232
|
|
},
|
|
{
|
|
"epoch": 0.12253484091506706,
|
|
"grad_norm": 3.931680202484131,
|
|
"learning_rate": 4.894957983193277e-06,
|
|
"loss": 1.2746,
|
|
"step": 233
|
|
},
|
|
{
|
|
"epoch": 0.12306074151985275,
|
|
"grad_norm": 3.501211166381836,
|
|
"learning_rate": 4.915966386554622e-06,
|
|
"loss": 1.2223,
|
|
"step": 234
|
|
},
|
|
{
|
|
"epoch": 0.12358664212463845,
|
|
"grad_norm": 3.4216043949127197,
|
|
"learning_rate": 4.936974789915967e-06,
|
|
"loss": 1.2072,
|
|
"step": 235
|
|
},
|
|
{
|
|
"epoch": 0.12411254272942414,
|
|
"grad_norm": 3.4759089946746826,
|
|
"learning_rate": 4.957983193277311e-06,
|
|
"loss": 1.2565,
|
|
"step": 236
|
|
},
|
|
{
|
|
"epoch": 0.12463844333420983,
|
|
"grad_norm": 3.4260342121124268,
|
|
"learning_rate": 4.978991596638656e-06,
|
|
"loss": 1.2142,
|
|
"step": 237
|
|
},
|
|
{
|
|
"epoch": 0.12516434393899553,
|
|
"grad_norm": 3.6531078815460205,
|
|
"learning_rate": 5e-06,
|
|
"loss": 1.1979,
|
|
"step": 238
|
|
},
|
|
{
|
|
"epoch": 0.12569024454378122,
|
|
"grad_norm": 3.5414624214172363,
|
|
"learning_rate": 5.021008403361345e-06,
|
|
"loss": 1.2312,
|
|
"step": 239
|
|
},
|
|
{
|
|
"epoch": 0.1262161451485669,
|
|
"grad_norm": 3.6826086044311523,
|
|
"learning_rate": 5.04201680672269e-06,
|
|
"loss": 1.2944,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.12674204575335263,
|
|
"grad_norm": 3.5233047008514404,
|
|
"learning_rate": 5.0630252100840335e-06,
|
|
"loss": 1.2619,
|
|
"step": 241
|
|
},
|
|
{
|
|
"epoch": 0.12726794635813832,
|
|
"grad_norm": 3.420653820037842,
|
|
"learning_rate": 5.084033613445378e-06,
|
|
"loss": 1.1992,
|
|
"step": 242
|
|
},
|
|
{
|
|
"epoch": 0.127793846962924,
|
|
"grad_norm": 3.5980186462402344,
|
|
"learning_rate": 5.105042016806723e-06,
|
|
"loss": 1.2209,
|
|
"step": 243
|
|
},
|
|
{
|
|
"epoch": 0.1283197475677097,
|
|
"grad_norm": 4.239027500152588,
|
|
"learning_rate": 5.1260504201680675e-06,
|
|
"loss": 1.2232,
|
|
"step": 244
|
|
},
|
|
{
|
|
"epoch": 0.1288456481724954,
|
|
"grad_norm": 3.590466260910034,
|
|
"learning_rate": 5.147058823529411e-06,
|
|
"loss": 1.1775,
|
|
"step": 245
|
|
},
|
|
{
|
|
"epoch": 0.1293715487772811,
|
|
"grad_norm": 3.6280338764190674,
|
|
"learning_rate": 5.168067226890757e-06,
|
|
"loss": 1.2451,
|
|
"step": 246
|
|
},
|
|
{
|
|
"epoch": 0.1298974493820668,
|
|
"grad_norm": 3.6897988319396973,
|
|
"learning_rate": 5.1890756302521015e-06,
|
|
"loss": 1.3339,
|
|
"step": 247
|
|
},
|
|
{
|
|
"epoch": 0.13042334998685248,
|
|
"grad_norm": 3.5509109497070312,
|
|
"learning_rate": 5.210084033613446e-06,
|
|
"loss": 1.1666,
|
|
"step": 248
|
|
},
|
|
{
|
|
"epoch": 0.13094925059163817,
|
|
"grad_norm": 3.652493953704834,
|
|
"learning_rate": 5.231092436974791e-06,
|
|
"loss": 1.2753,
|
|
"step": 249
|
|
},
|
|
{
|
|
"epoch": 0.1314751511964239,
|
|
"grad_norm": 3.528027057647705,
|
|
"learning_rate": 5.2521008403361354e-06,
|
|
"loss": 1.2509,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.13200105180120958,
|
|
"grad_norm": 3.572392225265503,
|
|
"learning_rate": 5.273109243697479e-06,
|
|
"loss": 1.2788,
|
|
"step": 251
|
|
},
|
|
{
|
|
"epoch": 0.13252695240599527,
|
|
"grad_norm": 3.8545901775360107,
|
|
"learning_rate": 5.294117647058824e-06,
|
|
"loss": 1.1701,
|
|
"step": 252
|
|
},
|
|
{
|
|
"epoch": 0.13305285301078096,
|
|
"grad_norm": 3.737743616104126,
|
|
"learning_rate": 5.3151260504201686e-06,
|
|
"loss": 1.212,
|
|
"step": 253
|
|
},
|
|
{
|
|
"epoch": 0.13357875361556665,
|
|
"grad_norm": 3.450867176055908,
|
|
"learning_rate": 5.336134453781513e-06,
|
|
"loss": 1.2599,
|
|
"step": 254
|
|
},
|
|
{
|
|
"epoch": 0.13410465422035237,
|
|
"grad_norm": 3.5492677688598633,
|
|
"learning_rate": 5.357142857142857e-06,
|
|
"loss": 1.142,
|
|
"step": 255
|
|
},
|
|
{
|
|
"epoch": 0.13463055482513805,
|
|
"grad_norm": 3.76678466796875,
|
|
"learning_rate": 5.378151260504202e-06,
|
|
"loss": 1.2123,
|
|
"step": 256
|
|
},
|
|
{
|
|
"epoch": 0.13515645542992374,
|
|
"grad_norm": 3.4145233631134033,
|
|
"learning_rate": 5.399159663865546e-06,
|
|
"loss": 1.2292,
|
|
"step": 257
|
|
},
|
|
{
|
|
"epoch": 0.13568235603470943,
|
|
"grad_norm": 3.737891912460327,
|
|
"learning_rate": 5.420168067226891e-06,
|
|
"loss": 1.2513,
|
|
"step": 258
|
|
},
|
|
{
|
|
"epoch": 0.13620825663949512,
|
|
"grad_norm": 3.754944086074829,
|
|
"learning_rate": 5.441176470588236e-06,
|
|
"loss": 1.2545,
|
|
"step": 259
|
|
},
|
|
{
|
|
"epoch": 0.13673415724428084,
|
|
"grad_norm": 3.334968328475952,
|
|
"learning_rate": 5.4621848739495795e-06,
|
|
"loss": 1.2334,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.13726005784906653,
|
|
"grad_norm": 3.6741573810577393,
|
|
"learning_rate": 5.483193277310925e-06,
|
|
"loss": 1.3083,
|
|
"step": 261
|
|
},
|
|
{
|
|
"epoch": 0.13778595845385222,
|
|
"grad_norm": 3.7089602947235107,
|
|
"learning_rate": 5.50420168067227e-06,
|
|
"loss": 1.2868,
|
|
"step": 262
|
|
},
|
|
{
|
|
"epoch": 0.1383118590586379,
|
|
"grad_norm": 3.6002771854400635,
|
|
"learning_rate": 5.525210084033614e-06,
|
|
"loss": 1.2947,
|
|
"step": 263
|
|
},
|
|
{
|
|
"epoch": 0.13883775966342363,
|
|
"grad_norm": 3.2446248531341553,
|
|
"learning_rate": 5.546218487394959e-06,
|
|
"loss": 1.1274,
|
|
"step": 264
|
|
},
|
|
{
|
|
"epoch": 0.13936366026820932,
|
|
"grad_norm": 3.627769708633423,
|
|
"learning_rate": 5.567226890756303e-06,
|
|
"loss": 1.2136,
|
|
"step": 265
|
|
},
|
|
{
|
|
"epoch": 0.139889560872995,
|
|
"grad_norm": 3.5511953830718994,
|
|
"learning_rate": 5.588235294117647e-06,
|
|
"loss": 1.2305,
|
|
"step": 266
|
|
},
|
|
{
|
|
"epoch": 0.1404154614777807,
|
|
"grad_norm": 3.588629722595215,
|
|
"learning_rate": 5.609243697478992e-06,
|
|
"loss": 1.3093,
|
|
"step": 267
|
|
},
|
|
{
|
|
"epoch": 0.14094136208256638,
|
|
"grad_norm": 3.8218297958374023,
|
|
"learning_rate": 5.630252100840337e-06,
|
|
"loss": 1.3218,
|
|
"step": 268
|
|
},
|
|
{
|
|
"epoch": 0.1414672626873521,
|
|
"grad_norm": 3.510584831237793,
|
|
"learning_rate": 5.651260504201681e-06,
|
|
"loss": 1.2365,
|
|
"step": 269
|
|
},
|
|
{
|
|
"epoch": 0.1419931632921378,
|
|
"grad_norm": 3.6596295833587646,
|
|
"learning_rate": 5.672268907563025e-06,
|
|
"loss": 1.1781,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.14251906389692348,
|
|
"grad_norm": 3.506701946258545,
|
|
"learning_rate": 5.69327731092437e-06,
|
|
"loss": 1.1984,
|
|
"step": 271
|
|
},
|
|
{
|
|
"epoch": 0.14304496450170917,
|
|
"grad_norm": 3.638305902481079,
|
|
"learning_rate": 5.7142857142857145e-06,
|
|
"loss": 1.1348,
|
|
"step": 272
|
|
},
|
|
{
|
|
"epoch": 0.14357086510649486,
|
|
"grad_norm": 3.486974000930786,
|
|
"learning_rate": 5.735294117647059e-06,
|
|
"loss": 1.1972,
|
|
"step": 273
|
|
},
|
|
{
|
|
"epoch": 0.14409676571128058,
|
|
"grad_norm": 4.081687927246094,
|
|
"learning_rate": 5.756302521008403e-06,
|
|
"loss": 1.273,
|
|
"step": 274
|
|
},
|
|
{
|
|
"epoch": 0.14462266631606627,
|
|
"grad_norm": 3.659921646118164,
|
|
"learning_rate": 5.777310924369748e-06,
|
|
"loss": 1.2205,
|
|
"step": 275
|
|
},
|
|
{
|
|
"epoch": 0.14514856692085196,
|
|
"grad_norm": 3.4537105560302734,
|
|
"learning_rate": 5.798319327731093e-06,
|
|
"loss": 1.2776,
|
|
"step": 276
|
|
},
|
|
{
|
|
"epoch": 0.14567446752563765,
|
|
"grad_norm": 3.3502252101898193,
|
|
"learning_rate": 5.819327731092438e-06,
|
|
"loss": 1.1629,
|
|
"step": 277
|
|
},
|
|
{
|
|
"epoch": 0.14620036813042336,
|
|
"grad_norm": 4.160674095153809,
|
|
"learning_rate": 5.8403361344537825e-06,
|
|
"loss": 1.2712,
|
|
"step": 278
|
|
},
|
|
{
|
|
"epoch": 0.14672626873520905,
|
|
"grad_norm": 3.790989637374878,
|
|
"learning_rate": 5.861344537815127e-06,
|
|
"loss": 1.2997,
|
|
"step": 279
|
|
},
|
|
{
|
|
"epoch": 0.14725216933999474,
|
|
"grad_norm": 3.4639837741851807,
|
|
"learning_rate": 5.882352941176471e-06,
|
|
"loss": 1.2569,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.14777806994478043,
|
|
"grad_norm": 3.414987564086914,
|
|
"learning_rate": 5.903361344537816e-06,
|
|
"loss": 1.2711,
|
|
"step": 281
|
|
},
|
|
{
|
|
"epoch": 0.14830397054956612,
|
|
"grad_norm": 3.599836587905884,
|
|
"learning_rate": 5.92436974789916e-06,
|
|
"loss": 1.2504,
|
|
"step": 282
|
|
},
|
|
{
|
|
"epoch": 0.14882987115435184,
|
|
"grad_norm": 4.437165260314941,
|
|
"learning_rate": 5.945378151260505e-06,
|
|
"loss": 1.1674,
|
|
"step": 283
|
|
},
|
|
{
|
|
"epoch": 0.14935577175913753,
|
|
"grad_norm": 3.45229434967041,
|
|
"learning_rate": 5.9663865546218495e-06,
|
|
"loss": 1.1958,
|
|
"step": 284
|
|
},
|
|
{
|
|
"epoch": 0.14988167236392322,
|
|
"grad_norm": 3.5962073802948,
|
|
"learning_rate": 5.987394957983193e-06,
|
|
"loss": 1.172,
|
|
"step": 285
|
|
},
|
|
{
|
|
"epoch": 0.1504075729687089,
|
|
"grad_norm": 3.5451295375823975,
|
|
"learning_rate": 6.008403361344538e-06,
|
|
"loss": 1.2841,
|
|
"step": 286
|
|
},
|
|
{
|
|
"epoch": 0.1509334735734946,
|
|
"grad_norm": 3.4825387001037598,
|
|
"learning_rate": 6.029411764705883e-06,
|
|
"loss": 1.2698,
|
|
"step": 287
|
|
},
|
|
{
|
|
"epoch": 0.15145937417828031,
|
|
"grad_norm": 3.5313754081726074,
|
|
"learning_rate": 6.050420168067227e-06,
|
|
"loss": 1.3121,
|
|
"step": 288
|
|
},
|
|
{
|
|
"epoch": 0.151985274783066,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 6.071428571428571e-06,
|
|
"loss": 1.2479,
|
|
"step": 289
|
|
},
|
|
{
|
|
"epoch": 0.1525111753878517,
|
|
"grad_norm": 3.520141363143921,
|
|
"learning_rate": 6.092436974789916e-06,
|
|
"loss": 1.254,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.15303707599263738,
|
|
"grad_norm": 3.683520555496216,
|
|
"learning_rate": 6.113445378151261e-06,
|
|
"loss": 1.2748,
|
|
"step": 291
|
|
},
|
|
{
|
|
"epoch": 0.1535629765974231,
|
|
"grad_norm": 3.302438735961914,
|
|
"learning_rate": 6.134453781512606e-06,
|
|
"loss": 1.2966,
|
|
"step": 292
|
|
},
|
|
{
|
|
"epoch": 0.1540888772022088,
|
|
"grad_norm": 3.4478492736816406,
|
|
"learning_rate": 6.155462184873951e-06,
|
|
"loss": 1.2007,
|
|
"step": 293
|
|
},
|
|
{
|
|
"epoch": 0.15461477780699448,
|
|
"grad_norm": 4.157530307769775,
|
|
"learning_rate": 6.176470588235295e-06,
|
|
"loss": 1.2243,
|
|
"step": 294
|
|
},
|
|
{
|
|
"epoch": 0.15514067841178017,
|
|
"grad_norm": 3.8873863220214844,
|
|
"learning_rate": 6.197478991596639e-06,
|
|
"loss": 1.2526,
|
|
"step": 295
|
|
},
|
|
{
|
|
"epoch": 0.15566657901656586,
|
|
"grad_norm": 4.097506999969482,
|
|
"learning_rate": 6.218487394957984e-06,
|
|
"loss": 1.3835,
|
|
"step": 296
|
|
},
|
|
{
|
|
"epoch": 0.15619247962135158,
|
|
"grad_norm": 3.7234766483306885,
|
|
"learning_rate": 6.239495798319328e-06,
|
|
"loss": 1.2491,
|
|
"step": 297
|
|
},
|
|
{
|
|
"epoch": 0.15671838022613727,
|
|
"grad_norm": 3.8953440189361572,
|
|
"learning_rate": 6.260504201680673e-06,
|
|
"loss": 1.2188,
|
|
"step": 298
|
|
},
|
|
{
|
|
"epoch": 0.15724428083092296,
|
|
"grad_norm": 3.6170642375946045,
|
|
"learning_rate": 6.281512605042017e-06,
|
|
"loss": 1.2182,
|
|
"step": 299
|
|
},
|
|
{
|
|
"epoch": 0.15777018143570865,
|
|
"grad_norm": 3.718691110610962,
|
|
"learning_rate": 6.3025210084033615e-06,
|
|
"loss": 1.1974,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.15829608204049433,
|
|
"grad_norm": 3.4003992080688477,
|
|
"learning_rate": 6.323529411764706e-06,
|
|
"loss": 1.2844,
|
|
"step": 301
|
|
},
|
|
{
|
|
"epoch": 0.15882198264528005,
|
|
"grad_norm": 3.6379525661468506,
|
|
"learning_rate": 6.344537815126051e-06,
|
|
"loss": 1.1409,
|
|
"step": 302
|
|
},
|
|
{
|
|
"epoch": 0.15934788325006574,
|
|
"grad_norm": 3.502467393875122,
|
|
"learning_rate": 6.3655462184873955e-06,
|
|
"loss": 1.206,
|
|
"step": 303
|
|
},
|
|
{
|
|
"epoch": 0.15987378385485143,
|
|
"grad_norm": 3.7654476165771484,
|
|
"learning_rate": 6.386554621848739e-06,
|
|
"loss": 1.201,
|
|
"step": 304
|
|
},
|
|
{
|
|
"epoch": 0.16039968445963712,
|
|
"grad_norm": 4.523637771606445,
|
|
"learning_rate": 6.407563025210085e-06,
|
|
"loss": 1.3212,
|
|
"step": 305
|
|
},
|
|
{
|
|
"epoch": 0.1609255850644228,
|
|
"grad_norm": 3.585219383239746,
|
|
"learning_rate": 6.4285714285714295e-06,
|
|
"loss": 1.3029,
|
|
"step": 306
|
|
},
|
|
{
|
|
"epoch": 0.16145148566920853,
|
|
"grad_norm": 3.505274534225464,
|
|
"learning_rate": 6.449579831932774e-06,
|
|
"loss": 1.2445,
|
|
"step": 307
|
|
},
|
|
{
|
|
"epoch": 0.16197738627399422,
|
|
"grad_norm": 3.751278877258301,
|
|
"learning_rate": 6.470588235294119e-06,
|
|
"loss": 1.2978,
|
|
"step": 308
|
|
},
|
|
{
|
|
"epoch": 0.1625032868787799,
|
|
"grad_norm": 3.4292526245117188,
|
|
"learning_rate": 6.491596638655463e-06,
|
|
"loss": 1.2591,
|
|
"step": 309
|
|
},
|
|
{
|
|
"epoch": 0.1630291874835656,
|
|
"grad_norm": 3.5352745056152344,
|
|
"learning_rate": 6.512605042016807e-06,
|
|
"loss": 1.2843,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 0.1635550880883513,
|
|
"grad_norm": 3.5294947624206543,
|
|
"learning_rate": 6.533613445378152e-06,
|
|
"loss": 1.1661,
|
|
"step": 311
|
|
},
|
|
{
|
|
"epoch": 0.164080988693137,
|
|
"grad_norm": 3.5288474559783936,
|
|
"learning_rate": 6.5546218487394966e-06,
|
|
"loss": 1.3153,
|
|
"step": 312
|
|
},
|
|
{
|
|
"epoch": 0.1646068892979227,
|
|
"grad_norm": 3.735635757446289,
|
|
"learning_rate": 6.575630252100841e-06,
|
|
"loss": 1.2236,
|
|
"step": 313
|
|
},
|
|
{
|
|
"epoch": 0.16513278990270838,
|
|
"grad_norm": 3.3471696376800537,
|
|
"learning_rate": 6.596638655462185e-06,
|
|
"loss": 1.176,
|
|
"step": 314
|
|
},
|
|
{
|
|
"epoch": 0.16565869050749407,
|
|
"grad_norm": 3.378708839416504,
|
|
"learning_rate": 6.61764705882353e-06,
|
|
"loss": 1.2418,
|
|
"step": 315
|
|
},
|
|
{
|
|
"epoch": 0.1661845911122798,
|
|
"grad_norm": 3.578310012817383,
|
|
"learning_rate": 6.638655462184874e-06,
|
|
"loss": 1.2343,
|
|
"step": 316
|
|
},
|
|
{
|
|
"epoch": 0.16671049171706548,
|
|
"grad_norm": 3.636845827102661,
|
|
"learning_rate": 6.659663865546219e-06,
|
|
"loss": 1.2516,
|
|
"step": 317
|
|
},
|
|
{
|
|
"epoch": 0.16723639232185117,
|
|
"grad_norm": 3.3973867893218994,
|
|
"learning_rate": 6.680672268907563e-06,
|
|
"loss": 1.1963,
|
|
"step": 318
|
|
},
|
|
{
|
|
"epoch": 0.16776229292663686,
|
|
"grad_norm": 3.4658966064453125,
|
|
"learning_rate": 6.7016806722689075e-06,
|
|
"loss": 1.2647,
|
|
"step": 319
|
|
},
|
|
{
|
|
"epoch": 0.16828819353142255,
|
|
"grad_norm": 3.5659782886505127,
|
|
"learning_rate": 6.722689075630253e-06,
|
|
"loss": 1.2902,
|
|
"step": 320
|
|
},
|
|
{
|
|
"epoch": 0.16881409413620826,
|
|
"grad_norm": 3.7267565727233887,
|
|
"learning_rate": 6.743697478991598e-06,
|
|
"loss": 1.3023,
|
|
"step": 321
|
|
},
|
|
{
|
|
"epoch": 0.16933999474099395,
|
|
"grad_norm": 3.6304492950439453,
|
|
"learning_rate": 6.764705882352942e-06,
|
|
"loss": 1.241,
|
|
"step": 322
|
|
},
|
|
{
|
|
"epoch": 0.16986589534577964,
|
|
"grad_norm": 3.532696485519409,
|
|
"learning_rate": 6.785714285714287e-06,
|
|
"loss": 1.2993,
|
|
"step": 323
|
|
},
|
|
{
|
|
"epoch": 0.17039179595056533,
|
|
"grad_norm": 3.6817219257354736,
|
|
"learning_rate": 6.806722689075631e-06,
|
|
"loss": 1.2719,
|
|
"step": 324
|
|
},
|
|
{
|
|
"epoch": 0.17091769655535105,
|
|
"grad_norm": 3.572549819946289,
|
|
"learning_rate": 6.827731092436975e-06,
|
|
"loss": 1.0911,
|
|
"step": 325
|
|
},
|
|
{
|
|
"epoch": 0.17144359716013674,
|
|
"grad_norm": 3.580209255218506,
|
|
"learning_rate": 6.84873949579832e-06,
|
|
"loss": 1.3011,
|
|
"step": 326
|
|
},
|
|
{
|
|
"epoch": 0.17196949776492243,
|
|
"grad_norm": 3.642347812652588,
|
|
"learning_rate": 6.869747899159665e-06,
|
|
"loss": 1.2494,
|
|
"step": 327
|
|
},
|
|
{
|
|
"epoch": 0.17249539836970812,
|
|
"grad_norm": 3.510411500930786,
|
|
"learning_rate": 6.8907563025210085e-06,
|
|
"loss": 1.2127,
|
|
"step": 328
|
|
},
|
|
{
|
|
"epoch": 0.1730212989744938,
|
|
"grad_norm": 3.5221219062805176,
|
|
"learning_rate": 6.911764705882353e-06,
|
|
"loss": 1.2347,
|
|
"step": 329
|
|
},
|
|
{
|
|
"epoch": 0.17354719957927953,
|
|
"grad_norm": 3.343609571456909,
|
|
"learning_rate": 6.932773109243698e-06,
|
|
"loss": 1.1779,
|
|
"step": 330
|
|
},
|
|
{
|
|
"epoch": 0.17407310018406522,
|
|
"grad_norm": 3.4778082370758057,
|
|
"learning_rate": 6.9537815126050425e-06,
|
|
"loss": 1.2524,
|
|
"step": 331
|
|
},
|
|
{
|
|
"epoch": 0.1745990007888509,
|
|
"grad_norm": 3.3407654762268066,
|
|
"learning_rate": 6.974789915966387e-06,
|
|
"loss": 1.2195,
|
|
"step": 332
|
|
},
|
|
{
|
|
"epoch": 0.1751249013936366,
|
|
"grad_norm": 3.2673823833465576,
|
|
"learning_rate": 6.995798319327731e-06,
|
|
"loss": 1.1569,
|
|
"step": 333
|
|
},
|
|
{
|
|
"epoch": 0.17565080199842228,
|
|
"grad_norm": 3.6628923416137695,
|
|
"learning_rate": 7.016806722689076e-06,
|
|
"loss": 1.2327,
|
|
"step": 334
|
|
},
|
|
{
|
|
"epoch": 0.176176702603208,
|
|
"grad_norm": 3.3434464931488037,
|
|
"learning_rate": 7.037815126050421e-06,
|
|
"loss": 1.177,
|
|
"step": 335
|
|
},
|
|
{
|
|
"epoch": 0.1767026032079937,
|
|
"grad_norm": 3.778930902481079,
|
|
"learning_rate": 7.058823529411766e-06,
|
|
"loss": 1.2157,
|
|
"step": 336
|
|
},
|
|
{
|
|
"epoch": 0.17722850381277938,
|
|
"grad_norm": 3.465285301208496,
|
|
"learning_rate": 7.0798319327731104e-06,
|
|
"loss": 1.2491,
|
|
"step": 337
|
|
},
|
|
{
|
|
"epoch": 0.17775440441756507,
|
|
"grad_norm": 3.534600019454956,
|
|
"learning_rate": 7.100840336134454e-06,
|
|
"loss": 1.1841,
|
|
"step": 338
|
|
},
|
|
{
|
|
"epoch": 0.1782803050223508,
|
|
"grad_norm": 3.36071515083313,
|
|
"learning_rate": 7.121848739495799e-06,
|
|
"loss": 1.226,
|
|
"step": 339
|
|
},
|
|
{
|
|
"epoch": 0.17880620562713648,
|
|
"grad_norm": 3.8877899646759033,
|
|
"learning_rate": 7.1428571428571436e-06,
|
|
"loss": 1.1769,
|
|
"step": 340
|
|
},
|
|
{
|
|
"epoch": 0.17933210623192217,
|
|
"grad_norm": 3.6745994091033936,
|
|
"learning_rate": 7.163865546218488e-06,
|
|
"loss": 1.1598,
|
|
"step": 341
|
|
},
|
|
{
|
|
"epoch": 0.17985800683670786,
|
|
"grad_norm": 3.72076416015625,
|
|
"learning_rate": 7.184873949579833e-06,
|
|
"loss": 1.3382,
|
|
"step": 342
|
|
},
|
|
{
|
|
"epoch": 0.18038390744149355,
|
|
"grad_norm": 3.49186635017395,
|
|
"learning_rate": 7.205882352941177e-06,
|
|
"loss": 1.1782,
|
|
"step": 343
|
|
},
|
|
{
|
|
"epoch": 0.18090980804627926,
|
|
"grad_norm": 3.5564091205596924,
|
|
"learning_rate": 7.226890756302521e-06,
|
|
"loss": 1.2838,
|
|
"step": 344
|
|
},
|
|
{
|
|
"epoch": 0.18143570865106495,
|
|
"grad_norm": 3.4774675369262695,
|
|
"learning_rate": 7.247899159663866e-06,
|
|
"loss": 1.1848,
|
|
"step": 345
|
|
},
|
|
{
|
|
"epoch": 0.18196160925585064,
|
|
"grad_norm": 3.3904049396514893,
|
|
"learning_rate": 7.268907563025211e-06,
|
|
"loss": 1.1794,
|
|
"step": 346
|
|
},
|
|
{
|
|
"epoch": 0.18248750986063633,
|
|
"grad_norm": 3.2446486949920654,
|
|
"learning_rate": 7.2899159663865545e-06,
|
|
"loss": 1.2067,
|
|
"step": 347
|
|
},
|
|
{
|
|
"epoch": 0.18301341046542202,
|
|
"grad_norm": 3.391934633255005,
|
|
"learning_rate": 7.310924369747899e-06,
|
|
"loss": 1.3455,
|
|
"step": 348
|
|
},
|
|
{
|
|
"epoch": 0.18353931107020774,
|
|
"grad_norm": 3.606053352355957,
|
|
"learning_rate": 7.331932773109244e-06,
|
|
"loss": 1.2721,
|
|
"step": 349
|
|
},
|
|
{
|
|
"epoch": 0.18406521167499343,
|
|
"grad_norm": 3.6179702281951904,
|
|
"learning_rate": 7.352941176470589e-06,
|
|
"loss": 1.322,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.18459111227977912,
|
|
"grad_norm": 3.1939680576324463,
|
|
"learning_rate": 7.373949579831934e-06,
|
|
"loss": 1.2703,
|
|
"step": 351
|
|
},
|
|
{
|
|
"epoch": 0.1851170128845648,
|
|
"grad_norm": 3.524625062942505,
|
|
"learning_rate": 7.394957983193279e-06,
|
|
"loss": 1.2863,
|
|
"step": 352
|
|
},
|
|
{
|
|
"epoch": 0.18564291348935053,
|
|
"grad_norm": 3.3166604042053223,
|
|
"learning_rate": 7.415966386554622e-06,
|
|
"loss": 1.3238,
|
|
"step": 353
|
|
},
|
|
{
|
|
"epoch": 0.18616881409413621,
|
|
"grad_norm": 3.383056163787842,
|
|
"learning_rate": 7.436974789915967e-06,
|
|
"loss": 1.1503,
|
|
"step": 354
|
|
},
|
|
{
|
|
"epoch": 0.1866947146989219,
|
|
"grad_norm": 3.3399603366851807,
|
|
"learning_rate": 7.457983193277312e-06,
|
|
"loss": 1.2167,
|
|
"step": 355
|
|
},
|
|
{
|
|
"epoch": 0.1872206153037076,
|
|
"grad_norm": 3.477342128753662,
|
|
"learning_rate": 7.478991596638656e-06,
|
|
"loss": 1.1822,
|
|
"step": 356
|
|
},
|
|
{
|
|
"epoch": 0.18774651590849328,
|
|
"grad_norm": 3.3987176418304443,
|
|
"learning_rate": 7.500000000000001e-06,
|
|
"loss": 1.2895,
|
|
"step": 357
|
|
},
|
|
{
|
|
"epoch": 0.188272416513279,
|
|
"grad_norm": 3.5033538341522217,
|
|
"learning_rate": 7.521008403361345e-06,
|
|
"loss": 1.2019,
|
|
"step": 358
|
|
},
|
|
{
|
|
"epoch": 0.1887983171180647,
|
|
"grad_norm": 3.4427967071533203,
|
|
"learning_rate": 7.5420168067226895e-06,
|
|
"loss": 1.1933,
|
|
"step": 359
|
|
},
|
|
{
|
|
"epoch": 0.18932421772285038,
|
|
"grad_norm": 3.594771385192871,
|
|
"learning_rate": 7.563025210084034e-06,
|
|
"loss": 1.3297,
|
|
"step": 360
|
|
},
|
|
{
|
|
"epoch": 0.18985011832763607,
|
|
"grad_norm": 3.6362686157226562,
|
|
"learning_rate": 7.584033613445379e-06,
|
|
"loss": 1.2112,
|
|
"step": 361
|
|
},
|
|
{
|
|
"epoch": 0.19037601893242176,
|
|
"grad_norm": 3.255256414413452,
|
|
"learning_rate": 7.605042016806723e-06,
|
|
"loss": 1.1939,
|
|
"step": 362
|
|
},
|
|
{
|
|
"epoch": 0.19090191953720748,
|
|
"grad_norm": 3.1901960372924805,
|
|
"learning_rate": 7.626050420168067e-06,
|
|
"loss": 1.1966,
|
|
"step": 363
|
|
},
|
|
{
|
|
"epoch": 0.19142782014199317,
|
|
"grad_norm": 3.1330695152282715,
|
|
"learning_rate": 7.647058823529411e-06,
|
|
"loss": 1.1844,
|
|
"step": 364
|
|
},
|
|
{
|
|
"epoch": 0.19195372074677886,
|
|
"grad_norm": 3.4831624031066895,
|
|
"learning_rate": 7.668067226890757e-06,
|
|
"loss": 1.1647,
|
|
"step": 365
|
|
},
|
|
{
|
|
"epoch": 0.19247962135156454,
|
|
"grad_norm": 3.4756598472595215,
|
|
"learning_rate": 7.689075630252102e-06,
|
|
"loss": 1.1864,
|
|
"step": 366
|
|
},
|
|
{
|
|
"epoch": 0.19300552195635026,
|
|
"grad_norm": 3.6858303546905518,
|
|
"learning_rate": 7.710084033613446e-06,
|
|
"loss": 1.2697,
|
|
"step": 367
|
|
},
|
|
{
|
|
"epoch": 0.19353142256113595,
|
|
"grad_norm": 3.6222240924835205,
|
|
"learning_rate": 7.731092436974791e-06,
|
|
"loss": 1.2917,
|
|
"step": 368
|
|
},
|
|
{
|
|
"epoch": 0.19405732316592164,
|
|
"grad_norm": 3.7319860458374023,
|
|
"learning_rate": 7.752100840336135e-06,
|
|
"loss": 1.3782,
|
|
"step": 369
|
|
},
|
|
{
|
|
"epoch": 0.19458322377070733,
|
|
"grad_norm": 3.536344289779663,
|
|
"learning_rate": 7.773109243697479e-06,
|
|
"loss": 1.2504,
|
|
"step": 370
|
|
},
|
|
{
|
|
"epoch": 0.19510912437549302,
|
|
"grad_norm": 3.349506378173828,
|
|
"learning_rate": 7.794117647058825e-06,
|
|
"loss": 1.1946,
|
|
"step": 371
|
|
},
|
|
{
|
|
"epoch": 0.19563502498027874,
|
|
"grad_norm": 3.3223843574523926,
|
|
"learning_rate": 7.815126050420168e-06,
|
|
"loss": 1.1858,
|
|
"step": 372
|
|
},
|
|
{
|
|
"epoch": 0.19616092558506443,
|
|
"grad_norm": 3.455443859100342,
|
|
"learning_rate": 7.836134453781514e-06,
|
|
"loss": 1.2998,
|
|
"step": 373
|
|
},
|
|
{
|
|
"epoch": 0.19668682618985012,
|
|
"grad_norm": 3.517239570617676,
|
|
"learning_rate": 7.857142857142858e-06,
|
|
"loss": 1.2731,
|
|
"step": 374
|
|
},
|
|
{
|
|
"epoch": 0.1972127267946358,
|
|
"grad_norm": 3.516794443130493,
|
|
"learning_rate": 7.878151260504201e-06,
|
|
"loss": 1.3466,
|
|
"step": 375
|
|
},
|
|
{
|
|
"epoch": 0.1977386273994215,
|
|
"grad_norm": 3.724583148956299,
|
|
"learning_rate": 7.899159663865547e-06,
|
|
"loss": 1.188,
|
|
"step": 376
|
|
},
|
|
{
|
|
"epoch": 0.1982645280042072,
|
|
"grad_norm": 3.3913514614105225,
|
|
"learning_rate": 7.92016806722689e-06,
|
|
"loss": 1.2229,
|
|
"step": 377
|
|
},
|
|
{
|
|
"epoch": 0.1987904286089929,
|
|
"grad_norm": 3.2986233234405518,
|
|
"learning_rate": 7.941176470588236e-06,
|
|
"loss": 1.1746,
|
|
"step": 378
|
|
},
|
|
{
|
|
"epoch": 0.1993163292137786,
|
|
"grad_norm": 3.4680469036102295,
|
|
"learning_rate": 7.96218487394958e-06,
|
|
"loss": 1.2391,
|
|
"step": 379
|
|
},
|
|
{
|
|
"epoch": 0.19984222981856428,
|
|
"grad_norm": 3.447463035583496,
|
|
"learning_rate": 7.983193277310926e-06,
|
|
"loss": 1.2872,
|
|
"step": 380
|
|
},
|
|
{
|
|
"epoch": 0.20036813042335,
|
|
"grad_norm": 3.3691208362579346,
|
|
"learning_rate": 8.00420168067227e-06,
|
|
"loss": 1.2778,
|
|
"step": 381
|
|
},
|
|
{
|
|
"epoch": 0.2008940310281357,
|
|
"grad_norm": 3.437351942062378,
|
|
"learning_rate": 8.025210084033615e-06,
|
|
"loss": 1.233,
|
|
"step": 382
|
|
},
|
|
{
|
|
"epoch": 0.20141993163292138,
|
|
"grad_norm": 3.3677237033843994,
|
|
"learning_rate": 8.046218487394959e-06,
|
|
"loss": 1.2617,
|
|
"step": 383
|
|
},
|
|
{
|
|
"epoch": 0.20194583223770707,
|
|
"grad_norm": 3.234267473220825,
|
|
"learning_rate": 8.067226890756303e-06,
|
|
"loss": 1.3127,
|
|
"step": 384
|
|
},
|
|
{
|
|
"epoch": 0.20247173284249276,
|
|
"grad_norm": 3.4119865894317627,
|
|
"learning_rate": 8.088235294117648e-06,
|
|
"loss": 1.2321,
|
|
"step": 385
|
|
},
|
|
{
|
|
"epoch": 0.20299763344727847,
|
|
"grad_norm": 3.3688788414001465,
|
|
"learning_rate": 8.109243697478992e-06,
|
|
"loss": 1.2763,
|
|
"step": 386
|
|
},
|
|
{
|
|
"epoch": 0.20352353405206416,
|
|
"grad_norm": 3.3807713985443115,
|
|
"learning_rate": 8.130252100840337e-06,
|
|
"loss": 1.2447,
|
|
"step": 387
|
|
},
|
|
{
|
|
"epoch": 0.20404943465684985,
|
|
"grad_norm": 3.346285820007324,
|
|
"learning_rate": 8.151260504201681e-06,
|
|
"loss": 1.3624,
|
|
"step": 388
|
|
},
|
|
{
|
|
"epoch": 0.20457533526163554,
|
|
"grad_norm": 3.453068256378174,
|
|
"learning_rate": 8.172268907563025e-06,
|
|
"loss": 1.3239,
|
|
"step": 389
|
|
},
|
|
{
|
|
"epoch": 0.20510123586642123,
|
|
"grad_norm": 3.154751777648926,
|
|
"learning_rate": 8.19327731092437e-06,
|
|
"loss": 1.1587,
|
|
"step": 390
|
|
},
|
|
{
|
|
"epoch": 0.20562713647120695,
|
|
"grad_norm": 3.279787302017212,
|
|
"learning_rate": 8.214285714285714e-06,
|
|
"loss": 1.1492,
|
|
"step": 391
|
|
},
|
|
{
|
|
"epoch": 0.20615303707599264,
|
|
"grad_norm": 3.5313329696655273,
|
|
"learning_rate": 8.23529411764706e-06,
|
|
"loss": 1.2254,
|
|
"step": 392
|
|
},
|
|
{
|
|
"epoch": 0.20667893768077833,
|
|
"grad_norm": 3.2370212078094482,
|
|
"learning_rate": 8.256302521008404e-06,
|
|
"loss": 1.2273,
|
|
"step": 393
|
|
},
|
|
{
|
|
"epoch": 0.20720483828556402,
|
|
"grad_norm": 3.2094857692718506,
|
|
"learning_rate": 8.277310924369747e-06,
|
|
"loss": 1.1828,
|
|
"step": 394
|
|
},
|
|
{
|
|
"epoch": 0.20773073889034974,
|
|
"grad_norm": 3.7691197395324707,
|
|
"learning_rate": 8.298319327731093e-06,
|
|
"loss": 1.1902,
|
|
"step": 395
|
|
},
|
|
{
|
|
"epoch": 0.20825663949513543,
|
|
"grad_norm": 3.344576835632324,
|
|
"learning_rate": 8.319327731092438e-06,
|
|
"loss": 1.3294,
|
|
"step": 396
|
|
},
|
|
{
|
|
"epoch": 0.20878254009992112,
|
|
"grad_norm": 3.6031830310821533,
|
|
"learning_rate": 8.340336134453782e-06,
|
|
"loss": 1.2922,
|
|
"step": 397
|
|
},
|
|
{
|
|
"epoch": 0.2093084407047068,
|
|
"grad_norm": 3.801748752593994,
|
|
"learning_rate": 8.361344537815128e-06,
|
|
"loss": 1.1932,
|
|
"step": 398
|
|
},
|
|
{
|
|
"epoch": 0.2098343413094925,
|
|
"grad_norm": 3.2574076652526855,
|
|
"learning_rate": 8.382352941176472e-06,
|
|
"loss": 1.1781,
|
|
"step": 399
|
|
},
|
|
{
|
|
"epoch": 0.2103602419142782,
|
|
"grad_norm": 3.436522960662842,
|
|
"learning_rate": 8.403361344537815e-06,
|
|
"loss": 1.287,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 0.2108861425190639,
|
|
"grad_norm": 3.4246935844421387,
|
|
"learning_rate": 8.424369747899161e-06,
|
|
"loss": 1.2457,
|
|
"step": 401
|
|
},
|
|
{
|
|
"epoch": 0.2114120431238496,
|
|
"grad_norm": 3.5879364013671875,
|
|
"learning_rate": 8.445378151260505e-06,
|
|
"loss": 1.2463,
|
|
"step": 402
|
|
},
|
|
{
|
|
"epoch": 0.21193794372863528,
|
|
"grad_norm": 3.6634533405303955,
|
|
"learning_rate": 8.46638655462185e-06,
|
|
"loss": 1.2632,
|
|
"step": 403
|
|
},
|
|
{
|
|
"epoch": 0.21246384433342097,
|
|
"grad_norm": 3.1841490268707275,
|
|
"learning_rate": 8.487394957983194e-06,
|
|
"loss": 1.2535,
|
|
"step": 404
|
|
},
|
|
{
|
|
"epoch": 0.2129897449382067,
|
|
"grad_norm": 3.436417579650879,
|
|
"learning_rate": 8.508403361344538e-06,
|
|
"loss": 1.3369,
|
|
"step": 405
|
|
},
|
|
{
|
|
"epoch": 0.21351564554299238,
|
|
"grad_norm": 3.61454176902771,
|
|
"learning_rate": 8.529411764705883e-06,
|
|
"loss": 1.3573,
|
|
"step": 406
|
|
},
|
|
{
|
|
"epoch": 0.21404154614777807,
|
|
"grad_norm": 3.713686466217041,
|
|
"learning_rate": 8.550420168067227e-06,
|
|
"loss": 1.2536,
|
|
"step": 407
|
|
},
|
|
{
|
|
"epoch": 0.21456744675256376,
|
|
"grad_norm": 3.719069004058838,
|
|
"learning_rate": 8.571428571428571e-06,
|
|
"loss": 1.278,
|
|
"step": 408
|
|
},
|
|
{
|
|
"epoch": 0.21509334735734947,
|
|
"grad_norm": 3.3540775775909424,
|
|
"learning_rate": 8.592436974789916e-06,
|
|
"loss": 1.2699,
|
|
"step": 409
|
|
},
|
|
{
|
|
"epoch": 0.21561924796213516,
|
|
"grad_norm": 3.3411834239959717,
|
|
"learning_rate": 8.613445378151262e-06,
|
|
"loss": 1.2923,
|
|
"step": 410
|
|
},
|
|
{
|
|
"epoch": 0.21614514856692085,
|
|
"grad_norm": 3.4757919311523438,
|
|
"learning_rate": 8.634453781512606e-06,
|
|
"loss": 1.1606,
|
|
"step": 411
|
|
},
|
|
{
|
|
"epoch": 0.21667104917170654,
|
|
"grad_norm": 3.543704032897949,
|
|
"learning_rate": 8.655462184873951e-06,
|
|
"loss": 1.2258,
|
|
"step": 412
|
|
},
|
|
{
|
|
"epoch": 0.21719694977649223,
|
|
"grad_norm": 3.6404592990875244,
|
|
"learning_rate": 8.676470588235295e-06,
|
|
"loss": 1.3106,
|
|
"step": 413
|
|
},
|
|
{
|
|
"epoch": 0.21772285038127795,
|
|
"grad_norm": 3.4733400344848633,
|
|
"learning_rate": 8.697478991596639e-06,
|
|
"loss": 1.2736,
|
|
"step": 414
|
|
},
|
|
{
|
|
"epoch": 0.21824875098606364,
|
|
"grad_norm": 3.543753147125244,
|
|
"learning_rate": 8.718487394957984e-06,
|
|
"loss": 1.2985,
|
|
"step": 415
|
|
},
|
|
{
|
|
"epoch": 0.21877465159084933,
|
|
"grad_norm": 3.145627737045288,
|
|
"learning_rate": 8.739495798319328e-06,
|
|
"loss": 1.3169,
|
|
"step": 416
|
|
},
|
|
{
|
|
"epoch": 0.21930055219563502,
|
|
"grad_norm": 3.7013349533081055,
|
|
"learning_rate": 8.760504201680674e-06,
|
|
"loss": 1.3639,
|
|
"step": 417
|
|
},
|
|
{
|
|
"epoch": 0.2198264528004207,
|
|
"grad_norm": 3.2583837509155273,
|
|
"learning_rate": 8.781512605042017e-06,
|
|
"loss": 1.2682,
|
|
"step": 418
|
|
},
|
|
{
|
|
"epoch": 0.22035235340520642,
|
|
"grad_norm": 3.2886593341827393,
|
|
"learning_rate": 8.802521008403361e-06,
|
|
"loss": 1.1349,
|
|
"step": 419
|
|
},
|
|
{
|
|
"epoch": 0.22087825400999211,
|
|
"grad_norm": 3.4134280681610107,
|
|
"learning_rate": 8.823529411764707e-06,
|
|
"loss": 1.2764,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 0.2214041546147778,
|
|
"grad_norm": 3.5238420963287354,
|
|
"learning_rate": 8.84453781512605e-06,
|
|
"loss": 1.2083,
|
|
"step": 421
|
|
},
|
|
{
|
|
"epoch": 0.2219300552195635,
|
|
"grad_norm": 3.4028427600860596,
|
|
"learning_rate": 8.865546218487396e-06,
|
|
"loss": 1.3196,
|
|
"step": 422
|
|
},
|
|
{
|
|
"epoch": 0.2224559558243492,
|
|
"grad_norm": 3.3646979331970215,
|
|
"learning_rate": 8.88655462184874e-06,
|
|
"loss": 1.3946,
|
|
"step": 423
|
|
},
|
|
{
|
|
"epoch": 0.2229818564291349,
|
|
"grad_norm": 3.406067371368408,
|
|
"learning_rate": 8.907563025210085e-06,
|
|
"loss": 1.2762,
|
|
"step": 424
|
|
},
|
|
{
|
|
"epoch": 0.2235077570339206,
|
|
"grad_norm": 3.1912147998809814,
|
|
"learning_rate": 8.92857142857143e-06,
|
|
"loss": 1.2401,
|
|
"step": 425
|
|
},
|
|
{
|
|
"epoch": 0.22403365763870628,
|
|
"grad_norm": 3.3375768661499023,
|
|
"learning_rate": 8.949579831932775e-06,
|
|
"loss": 1.2039,
|
|
"step": 426
|
|
},
|
|
{
|
|
"epoch": 0.22455955824349197,
|
|
"grad_norm": 3.25075626373291,
|
|
"learning_rate": 8.970588235294119e-06,
|
|
"loss": 1.3374,
|
|
"step": 427
|
|
},
|
|
{
|
|
"epoch": 0.2250854588482777,
|
|
"grad_norm": 3.7262144088745117,
|
|
"learning_rate": 8.991596638655462e-06,
|
|
"loss": 1.3661,
|
|
"step": 428
|
|
},
|
|
{
|
|
"epoch": 0.22561135945306338,
|
|
"grad_norm": 3.6752355098724365,
|
|
"learning_rate": 9.012605042016808e-06,
|
|
"loss": 1.3766,
|
|
"step": 429
|
|
},
|
|
{
|
|
"epoch": 0.22613726005784907,
|
|
"grad_norm": 3.316227912902832,
|
|
"learning_rate": 9.033613445378152e-06,
|
|
"loss": 1.0969,
|
|
"step": 430
|
|
},
|
|
{
|
|
"epoch": 0.22666316066263476,
|
|
"grad_norm": 3.606106996536255,
|
|
"learning_rate": 9.054621848739497e-06,
|
|
"loss": 1.2944,
|
|
"step": 431
|
|
},
|
|
{
|
|
"epoch": 0.22718906126742044,
|
|
"grad_norm": 3.3386571407318115,
|
|
"learning_rate": 9.075630252100841e-06,
|
|
"loss": 1.2981,
|
|
"step": 432
|
|
},
|
|
{
|
|
"epoch": 0.22771496187220616,
|
|
"grad_norm": 3.1603779792785645,
|
|
"learning_rate": 9.096638655462185e-06,
|
|
"loss": 1.2257,
|
|
"step": 433
|
|
},
|
|
{
|
|
"epoch": 0.22824086247699185,
|
|
"grad_norm": 3.2775022983551025,
|
|
"learning_rate": 9.11764705882353e-06,
|
|
"loss": 1.1883,
|
|
"step": 434
|
|
},
|
|
{
|
|
"epoch": 0.22876676308177754,
|
|
"grad_norm": 3.324110984802246,
|
|
"learning_rate": 9.138655462184874e-06,
|
|
"loss": 1.2139,
|
|
"step": 435
|
|
},
|
|
{
|
|
"epoch": 0.22929266368656323,
|
|
"grad_norm": 3.3305015563964844,
|
|
"learning_rate": 9.15966386554622e-06,
|
|
"loss": 1.2369,
|
|
"step": 436
|
|
},
|
|
{
|
|
"epoch": 0.22981856429134895,
|
|
"grad_norm": 3.4254801273345947,
|
|
"learning_rate": 9.180672268907563e-06,
|
|
"loss": 1.2791,
|
|
"step": 437
|
|
},
|
|
{
|
|
"epoch": 0.23034446489613464,
|
|
"grad_norm": 3.2646000385284424,
|
|
"learning_rate": 9.201680672268907e-06,
|
|
"loss": 1.2385,
|
|
"step": 438
|
|
},
|
|
{
|
|
"epoch": 0.23087036550092033,
|
|
"grad_norm": 3.2213170528411865,
|
|
"learning_rate": 9.222689075630253e-06,
|
|
"loss": 1.2588,
|
|
"step": 439
|
|
},
|
|
{
|
|
"epoch": 0.23139626610570602,
|
|
"grad_norm": 3.165525436401367,
|
|
"learning_rate": 9.243697478991598e-06,
|
|
"loss": 1.2506,
|
|
"step": 440
|
|
},
|
|
{
|
|
"epoch": 0.2319221667104917,
|
|
"grad_norm": 3.1039962768554688,
|
|
"learning_rate": 9.264705882352942e-06,
|
|
"loss": 1.1679,
|
|
"step": 441
|
|
},
|
|
{
|
|
"epoch": 0.23244806731527742,
|
|
"grad_norm": 3.4077186584472656,
|
|
"learning_rate": 9.285714285714288e-06,
|
|
"loss": 1.1349,
|
|
"step": 442
|
|
},
|
|
{
|
|
"epoch": 0.2329739679200631,
|
|
"grad_norm": 3.3018462657928467,
|
|
"learning_rate": 9.306722689075631e-06,
|
|
"loss": 1.2525,
|
|
"step": 443
|
|
},
|
|
{
|
|
"epoch": 0.2334998685248488,
|
|
"grad_norm": 3.18839955329895,
|
|
"learning_rate": 9.327731092436975e-06,
|
|
"loss": 1.2263,
|
|
"step": 444
|
|
},
|
|
{
|
|
"epoch": 0.2340257691296345,
|
|
"grad_norm": 3.305591344833374,
|
|
"learning_rate": 9.34873949579832e-06,
|
|
"loss": 1.2796,
|
|
"step": 445
|
|
},
|
|
{
|
|
"epoch": 0.23455166973442018,
|
|
"grad_norm": 3.390716314315796,
|
|
"learning_rate": 9.369747899159664e-06,
|
|
"loss": 1.2414,
|
|
"step": 446
|
|
},
|
|
{
|
|
"epoch": 0.2350775703392059,
|
|
"grad_norm": 3.4202969074249268,
|
|
"learning_rate": 9.390756302521008e-06,
|
|
"loss": 1.2461,
|
|
"step": 447
|
|
},
|
|
{
|
|
"epoch": 0.2356034709439916,
|
|
"grad_norm": 3.480309009552002,
|
|
"learning_rate": 9.411764705882354e-06,
|
|
"loss": 1.2265,
|
|
"step": 448
|
|
},
|
|
{
|
|
"epoch": 0.23612937154877728,
|
|
"grad_norm": 3.4652767181396484,
|
|
"learning_rate": 9.432773109243698e-06,
|
|
"loss": 1.3006,
|
|
"step": 449
|
|
},
|
|
{
|
|
"epoch": 0.23665527215356297,
|
|
"grad_norm": 3.5682034492492676,
|
|
"learning_rate": 9.453781512605043e-06,
|
|
"loss": 1.3058,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 0.23718117275834869,
|
|
"grad_norm": 3.51560115814209,
|
|
"learning_rate": 9.474789915966387e-06,
|
|
"loss": 1.3212,
|
|
"step": 451
|
|
},
|
|
{
|
|
"epoch": 0.23770707336313437,
|
|
"grad_norm": 3.389254093170166,
|
|
"learning_rate": 9.49579831932773e-06,
|
|
"loss": 1.2214,
|
|
"step": 452
|
|
},
|
|
{
|
|
"epoch": 0.23823297396792006,
|
|
"grad_norm": 3.2306952476501465,
|
|
"learning_rate": 9.516806722689076e-06,
|
|
"loss": 1.3258,
|
|
"step": 453
|
|
},
|
|
{
|
|
"epoch": 0.23875887457270575,
|
|
"grad_norm": 3.1801748275756836,
|
|
"learning_rate": 9.537815126050422e-06,
|
|
"loss": 1.1598,
|
|
"step": 454
|
|
},
|
|
{
|
|
"epoch": 0.23928477517749144,
|
|
"grad_norm": 3.3290438652038574,
|
|
"learning_rate": 9.558823529411766e-06,
|
|
"loss": 1.2457,
|
|
"step": 455
|
|
},
|
|
{
|
|
"epoch": 0.23981067578227716,
|
|
"grad_norm": 3.3275296688079834,
|
|
"learning_rate": 9.579831932773111e-06,
|
|
"loss": 1.1944,
|
|
"step": 456
|
|
},
|
|
{
|
|
"epoch": 0.24033657638706285,
|
|
"grad_norm": 3.212510108947754,
|
|
"learning_rate": 9.600840336134455e-06,
|
|
"loss": 1.2681,
|
|
"step": 457
|
|
},
|
|
{
|
|
"epoch": 0.24086247699184854,
|
|
"grad_norm": 3.6425256729125977,
|
|
"learning_rate": 9.621848739495799e-06,
|
|
"loss": 1.3541,
|
|
"step": 458
|
|
},
|
|
{
|
|
"epoch": 0.24138837759663423,
|
|
"grad_norm": 3.6957900524139404,
|
|
"learning_rate": 9.642857142857144e-06,
|
|
"loss": 1.292,
|
|
"step": 459
|
|
},
|
|
{
|
|
"epoch": 0.24191427820141992,
|
|
"grad_norm": 3.473247766494751,
|
|
"learning_rate": 9.663865546218488e-06,
|
|
"loss": 1.3963,
|
|
"step": 460
|
|
},
|
|
{
|
|
"epoch": 0.24244017880620564,
|
|
"grad_norm": 3.1998612880706787,
|
|
"learning_rate": 9.684873949579834e-06,
|
|
"loss": 1.2584,
|
|
"step": 461
|
|
},
|
|
{
|
|
"epoch": 0.24296607941099133,
|
|
"grad_norm": 3.608579635620117,
|
|
"learning_rate": 9.705882352941177e-06,
|
|
"loss": 1.1799,
|
|
"step": 462
|
|
},
|
|
{
|
|
"epoch": 0.24349198001577702,
|
|
"grad_norm": 3.45597243309021,
|
|
"learning_rate": 9.726890756302521e-06,
|
|
"loss": 1.3952,
|
|
"step": 463
|
|
},
|
|
{
|
|
"epoch": 0.2440178806205627,
|
|
"grad_norm": 3.7488176822662354,
|
|
"learning_rate": 9.747899159663867e-06,
|
|
"loss": 1.2042,
|
|
"step": 464
|
|
},
|
|
{
|
|
"epoch": 0.24454378122534842,
|
|
"grad_norm": 3.3650524616241455,
|
|
"learning_rate": 9.76890756302521e-06,
|
|
"loss": 1.2851,
|
|
"step": 465
|
|
},
|
|
{
|
|
"epoch": 0.2450696818301341,
|
|
"grad_norm": 3.1787731647491455,
|
|
"learning_rate": 9.789915966386554e-06,
|
|
"loss": 1.2049,
|
|
"step": 466
|
|
},
|
|
{
|
|
"epoch": 0.2455955824349198,
|
|
"grad_norm": 3.3141655921936035,
|
|
"learning_rate": 9.8109243697479e-06,
|
|
"loss": 1.2836,
|
|
"step": 467
|
|
},
|
|
{
|
|
"epoch": 0.2461214830397055,
|
|
"grad_norm": 3.3473637104034424,
|
|
"learning_rate": 9.831932773109244e-06,
|
|
"loss": 1.1765,
|
|
"step": 468
|
|
},
|
|
{
|
|
"epoch": 0.24664738364449118,
|
|
"grad_norm": 3.2358179092407227,
|
|
"learning_rate": 9.852941176470589e-06,
|
|
"loss": 1.1996,
|
|
"step": 469
|
|
},
|
|
{
|
|
"epoch": 0.2471732842492769,
|
|
"grad_norm": 3.335753917694092,
|
|
"learning_rate": 9.873949579831935e-06,
|
|
"loss": 1.2594,
|
|
"step": 470
|
|
},
|
|
{
|
|
"epoch": 0.2476991848540626,
|
|
"grad_norm": 3.253922462463379,
|
|
"learning_rate": 9.894957983193278e-06,
|
|
"loss": 1.2598,
|
|
"step": 471
|
|
},
|
|
{
|
|
"epoch": 0.24822508545884828,
|
|
"grad_norm": 3.15269136428833,
|
|
"learning_rate": 9.915966386554622e-06,
|
|
"loss": 1.2652,
|
|
"step": 472
|
|
},
|
|
{
|
|
"epoch": 0.24875098606363397,
|
|
"grad_norm": 3.415088653564453,
|
|
"learning_rate": 9.936974789915968e-06,
|
|
"loss": 1.2718,
|
|
"step": 473
|
|
},
|
|
{
|
|
"epoch": 0.24927688666841966,
|
|
"grad_norm": 3.5668835639953613,
|
|
"learning_rate": 9.957983193277312e-06,
|
|
"loss": 1.3977,
|
|
"step": 474
|
|
},
|
|
{
|
|
"epoch": 0.24980278727320537,
|
|
"grad_norm": 3.307189464569092,
|
|
"learning_rate": 9.978991596638657e-06,
|
|
"loss": 1.2344,
|
|
"step": 475
|
|
},
|
|
{
|
|
"epoch": 0.25032868787799106,
|
|
"grad_norm": 3.240340232849121,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1339,
|
|
"step": 476
|
|
},
|
|
{
|
|
"epoch": 0.25085458848277675,
|
|
"grad_norm": 3.509732246398926,
|
|
"learning_rate": 9.999999697336226e-06,
|
|
"loss": 1.2654,
|
|
"step": 477
|
|
},
|
|
{
|
|
"epoch": 0.25138048908756244,
|
|
"grad_norm": 3.858966827392578,
|
|
"learning_rate": 9.99999878934494e-06,
|
|
"loss": 1.2723,
|
|
"step": 478
|
|
},
|
|
{
|
|
"epoch": 0.25190638969234813,
|
|
"grad_norm": 3.1328506469726562,
|
|
"learning_rate": 9.999997276026253e-06,
|
|
"loss": 1.3726,
|
|
"step": 479
|
|
},
|
|
{
|
|
"epoch": 0.2524322902971338,
|
|
"grad_norm": 3.1664280891418457,
|
|
"learning_rate": 9.999995157380345e-06,
|
|
"loss": 1.2383,
|
|
"step": 480
|
|
},
|
|
{
|
|
"epoch": 0.2529581909019195,
|
|
"grad_norm": 3.2397758960723877,
|
|
"learning_rate": 9.999992433407474e-06,
|
|
"loss": 1.2049,
|
|
"step": 481
|
|
},
|
|
{
|
|
"epoch": 0.25348409150670526,
|
|
"grad_norm": 3.769228458404541,
|
|
"learning_rate": 9.999989104107972e-06,
|
|
"loss": 1.2404,
|
|
"step": 482
|
|
},
|
|
{
|
|
"epoch": 0.25400999211149095,
|
|
"grad_norm": 3.559947967529297,
|
|
"learning_rate": 9.99998516948224e-06,
|
|
"loss": 1.2424,
|
|
"step": 483
|
|
},
|
|
{
|
|
"epoch": 0.25453589271627663,
|
|
"grad_norm": 3.129535436630249,
|
|
"learning_rate": 9.999980629530754e-06,
|
|
"loss": 1.3003,
|
|
"step": 484
|
|
},
|
|
{
|
|
"epoch": 0.2550617933210623,
|
|
"grad_norm": 3.1546337604522705,
|
|
"learning_rate": 9.999975484254065e-06,
|
|
"loss": 1.2336,
|
|
"step": 485
|
|
},
|
|
{
|
|
"epoch": 0.255587693925848,
|
|
"grad_norm": 3.3545641899108887,
|
|
"learning_rate": 9.999969733652794e-06,
|
|
"loss": 1.3105,
|
|
"step": 486
|
|
},
|
|
{
|
|
"epoch": 0.2561135945306337,
|
|
"grad_norm": 3.8369863033294678,
|
|
"learning_rate": 9.999963377727639e-06,
|
|
"loss": 1.2203,
|
|
"step": 487
|
|
},
|
|
{
|
|
"epoch": 0.2566394951354194,
|
|
"grad_norm": 3.4755983352661133,
|
|
"learning_rate": 9.99995641647937e-06,
|
|
"loss": 1.2224,
|
|
"step": 488
|
|
},
|
|
{
|
|
"epoch": 0.2571653957402051,
|
|
"grad_norm": 3.3242459297180176,
|
|
"learning_rate": 9.99994884990883e-06,
|
|
"loss": 1.2871,
|
|
"step": 489
|
|
},
|
|
{
|
|
"epoch": 0.2576912963449908,
|
|
"grad_norm": 3.08255934715271,
|
|
"learning_rate": 9.99994067801693e-06,
|
|
"loss": 1.1898,
|
|
"step": 490
|
|
},
|
|
{
|
|
"epoch": 0.2582171969497765,
|
|
"grad_norm": 3.3641436100006104,
|
|
"learning_rate": 9.999931900804668e-06,
|
|
"loss": 1.393,
|
|
"step": 491
|
|
},
|
|
{
|
|
"epoch": 0.2587430975545622,
|
|
"grad_norm": 3.1735382080078125,
|
|
"learning_rate": 9.999922518273098e-06,
|
|
"loss": 1.1969,
|
|
"step": 492
|
|
},
|
|
{
|
|
"epoch": 0.2592689981593479,
|
|
"grad_norm": 3.380448341369629,
|
|
"learning_rate": 9.99991253042336e-06,
|
|
"loss": 1.2476,
|
|
"step": 493
|
|
},
|
|
{
|
|
"epoch": 0.2597948987641336,
|
|
"grad_norm": 3.2164065837860107,
|
|
"learning_rate": 9.999901937256662e-06,
|
|
"loss": 1.2268,
|
|
"step": 494
|
|
},
|
|
{
|
|
"epoch": 0.2603207993689193,
|
|
"grad_norm": 3.311929702758789,
|
|
"learning_rate": 9.99989073877429e-06,
|
|
"loss": 1.1993,
|
|
"step": 495
|
|
},
|
|
{
|
|
"epoch": 0.26084669997370497,
|
|
"grad_norm": 3.2050387859344482,
|
|
"learning_rate": 9.999878934977595e-06,
|
|
"loss": 1.3001,
|
|
"step": 496
|
|
},
|
|
{
|
|
"epoch": 0.26137260057849065,
|
|
"grad_norm": 3.2873027324676514,
|
|
"learning_rate": 9.999866525868009e-06,
|
|
"loss": 1.2619,
|
|
"step": 497
|
|
},
|
|
{
|
|
"epoch": 0.26189850118327634,
|
|
"grad_norm": 3.77085280418396,
|
|
"learning_rate": 9.999853511447034e-06,
|
|
"loss": 1.3853,
|
|
"step": 498
|
|
},
|
|
{
|
|
"epoch": 0.26242440178806203,
|
|
"grad_norm": 3.8282227516174316,
|
|
"learning_rate": 9.999839891716243e-06,
|
|
"loss": 1.4078,
|
|
"step": 499
|
|
},
|
|
{
|
|
"epoch": 0.2629503023928478,
|
|
"grad_norm": 3.237232208251953,
|
|
"learning_rate": 9.99982566667729e-06,
|
|
"loss": 1.2915,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.2629503023928478,
|
|
"eval_loss": 1.2809544801712036,
|
|
"eval_runtime": 8.9514,
|
|
"eval_samples_per_second": 44.686,
|
|
"eval_steps_per_second": 2.793,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.26347620299763347,
|
|
"grad_norm": 3.2364449501037598,
|
|
"learning_rate": 9.999810836331892e-06,
|
|
"loss": 1.3638,
|
|
"step": 501
|
|
},
|
|
{
|
|
"epoch": 0.26400210360241916,
|
|
"grad_norm": 3.2302019596099854,
|
|
"learning_rate": 9.999795400681846e-06,
|
|
"loss": 1.2739,
|
|
"step": 502
|
|
},
|
|
{
|
|
"epoch": 0.26452800420720485,
|
|
"grad_norm": 3.262401580810547,
|
|
"learning_rate": 9.999779359729022e-06,
|
|
"loss": 1.215,
|
|
"step": 503
|
|
},
|
|
{
|
|
"epoch": 0.26505390481199054,
|
|
"grad_norm": 3.0523831844329834,
|
|
"learning_rate": 9.99976271347536e-06,
|
|
"loss": 1.2443,
|
|
"step": 504
|
|
},
|
|
{
|
|
"epoch": 0.2655798054167762,
|
|
"grad_norm": 3.1379828453063965,
|
|
"learning_rate": 9.999745461922879e-06,
|
|
"loss": 1.2486,
|
|
"step": 505
|
|
},
|
|
{
|
|
"epoch": 0.2661057060215619,
|
|
"grad_norm": 3.3893792629241943,
|
|
"learning_rate": 9.999727605073663e-06,
|
|
"loss": 1.3172,
|
|
"step": 506
|
|
},
|
|
{
|
|
"epoch": 0.2666316066263476,
|
|
"grad_norm": 3.3273253440856934,
|
|
"learning_rate": 9.999709142929876e-06,
|
|
"loss": 1.2517,
|
|
"step": 507
|
|
},
|
|
{
|
|
"epoch": 0.2671575072311333,
|
|
"grad_norm": 3.272167444229126,
|
|
"learning_rate": 9.999690075493754e-06,
|
|
"loss": 1.2469,
|
|
"step": 508
|
|
},
|
|
{
|
|
"epoch": 0.267683407835919,
|
|
"grad_norm": 3.1925384998321533,
|
|
"learning_rate": 9.999670402767603e-06,
|
|
"loss": 1.2799,
|
|
"step": 509
|
|
},
|
|
{
|
|
"epoch": 0.26820930844070473,
|
|
"grad_norm": 3.2661190032958984,
|
|
"learning_rate": 9.999650124753808e-06,
|
|
"loss": 1.2304,
|
|
"step": 510
|
|
},
|
|
{
|
|
"epoch": 0.2687352090454904,
|
|
"grad_norm": 3.5036795139312744,
|
|
"learning_rate": 9.99962924145482e-06,
|
|
"loss": 1.2772,
|
|
"step": 511
|
|
},
|
|
{
|
|
"epoch": 0.2692611096502761,
|
|
"grad_norm": 3.3797895908355713,
|
|
"learning_rate": 9.99960775287317e-06,
|
|
"loss": 1.2108,
|
|
"step": 512
|
|
},
|
|
{
|
|
"epoch": 0.2697870102550618,
|
|
"grad_norm": 3.1981654167175293,
|
|
"learning_rate": 9.999585659011461e-06,
|
|
"loss": 1.252,
|
|
"step": 513
|
|
},
|
|
{
|
|
"epoch": 0.2703129108598475,
|
|
"grad_norm": 3.440767765045166,
|
|
"learning_rate": 9.999562959872363e-06,
|
|
"loss": 1.3283,
|
|
"step": 514
|
|
},
|
|
{
|
|
"epoch": 0.2708388114646332,
|
|
"grad_norm": 3.5900278091430664,
|
|
"learning_rate": 9.999539655458628e-06,
|
|
"loss": 1.2728,
|
|
"step": 515
|
|
},
|
|
{
|
|
"epoch": 0.27136471206941887,
|
|
"grad_norm": 3.429997682571411,
|
|
"learning_rate": 9.999515745773076e-06,
|
|
"loss": 1.3023,
|
|
"step": 516
|
|
},
|
|
{
|
|
"epoch": 0.27189061267420456,
|
|
"grad_norm": 3.3940064907073975,
|
|
"learning_rate": 9.999491230818601e-06,
|
|
"loss": 1.3859,
|
|
"step": 517
|
|
},
|
|
{
|
|
"epoch": 0.27241651327899025,
|
|
"grad_norm": 3.3305585384368896,
|
|
"learning_rate": 9.999466110598173e-06,
|
|
"loss": 1.298,
|
|
"step": 518
|
|
},
|
|
{
|
|
"epoch": 0.272942413883776,
|
|
"grad_norm": 3.0065155029296875,
|
|
"learning_rate": 9.999440385114831e-06,
|
|
"loss": 1.3264,
|
|
"step": 519
|
|
},
|
|
{
|
|
"epoch": 0.2734683144885617,
|
|
"grad_norm": 3.489729166030884,
|
|
"learning_rate": 9.999414054371691e-06,
|
|
"loss": 1.3436,
|
|
"step": 520
|
|
},
|
|
{
|
|
"epoch": 0.27399421509334737,
|
|
"grad_norm": 3.15972638130188,
|
|
"learning_rate": 9.999387118371939e-06,
|
|
"loss": 1.1329,
|
|
"step": 521
|
|
},
|
|
{
|
|
"epoch": 0.27452011569813306,
|
|
"grad_norm": 3.1717405319213867,
|
|
"learning_rate": 9.999359577118838e-06,
|
|
"loss": 1.3082,
|
|
"step": 522
|
|
},
|
|
{
|
|
"epoch": 0.27504601630291875,
|
|
"grad_norm": 3.3579068183898926,
|
|
"learning_rate": 9.999331430615722e-06,
|
|
"loss": 1.3825,
|
|
"step": 523
|
|
},
|
|
{
|
|
"epoch": 0.27557191690770444,
|
|
"grad_norm": 3.2819440364837646,
|
|
"learning_rate": 9.999302678865997e-06,
|
|
"loss": 1.2226,
|
|
"step": 524
|
|
},
|
|
{
|
|
"epoch": 0.27609781751249013,
|
|
"grad_norm": 3.188291311264038,
|
|
"learning_rate": 9.999273321873144e-06,
|
|
"loss": 1.2235,
|
|
"step": 525
|
|
},
|
|
{
|
|
"epoch": 0.2766237181172758,
|
|
"grad_norm": 3.1913392543792725,
|
|
"learning_rate": 9.99924335964072e-06,
|
|
"loss": 1.2653,
|
|
"step": 526
|
|
},
|
|
{
|
|
"epoch": 0.2771496187220615,
|
|
"grad_norm": 3.599456548690796,
|
|
"learning_rate": 9.999212792172348e-06,
|
|
"loss": 1.2974,
|
|
"step": 527
|
|
},
|
|
{
|
|
"epoch": 0.27767551932684725,
|
|
"grad_norm": 4.070490837097168,
|
|
"learning_rate": 9.999181619471732e-06,
|
|
"loss": 1.2584,
|
|
"step": 528
|
|
},
|
|
{
|
|
"epoch": 0.27820141993163294,
|
|
"grad_norm": 3.550588846206665,
|
|
"learning_rate": 9.999149841542645e-06,
|
|
"loss": 1.3468,
|
|
"step": 529
|
|
},
|
|
{
|
|
"epoch": 0.27872732053641863,
|
|
"grad_norm": 3.383667230606079,
|
|
"learning_rate": 9.999117458388936e-06,
|
|
"loss": 1.19,
|
|
"step": 530
|
|
},
|
|
{
|
|
"epoch": 0.2792532211412043,
|
|
"grad_norm": 3.0106310844421387,
|
|
"learning_rate": 9.999084470014522e-06,
|
|
"loss": 1.1344,
|
|
"step": 531
|
|
},
|
|
{
|
|
"epoch": 0.27977912174599,
|
|
"grad_norm": 3.5307071208953857,
|
|
"learning_rate": 9.999050876423397e-06,
|
|
"loss": 1.2794,
|
|
"step": 532
|
|
},
|
|
{
|
|
"epoch": 0.2803050223507757,
|
|
"grad_norm": 3.0677614212036133,
|
|
"learning_rate": 9.999016677619631e-06,
|
|
"loss": 1.1847,
|
|
"step": 533
|
|
},
|
|
{
|
|
"epoch": 0.2808309229555614,
|
|
"grad_norm": 3.036958694458008,
|
|
"learning_rate": 9.998981873607363e-06,
|
|
"loss": 1.3003,
|
|
"step": 534
|
|
},
|
|
{
|
|
"epoch": 0.2813568235603471,
|
|
"grad_norm": 3.19941782951355,
|
|
"learning_rate": 9.998946464390805e-06,
|
|
"loss": 1.2622,
|
|
"step": 535
|
|
},
|
|
{
|
|
"epoch": 0.28188272416513277,
|
|
"grad_norm": 2.786508560180664,
|
|
"learning_rate": 9.998910449974246e-06,
|
|
"loss": 1.2186,
|
|
"step": 536
|
|
},
|
|
{
|
|
"epoch": 0.28240862476991846,
|
|
"grad_norm": 3.2854580879211426,
|
|
"learning_rate": 9.998873830362044e-06,
|
|
"loss": 1.3212,
|
|
"step": 537
|
|
},
|
|
{
|
|
"epoch": 0.2829345253747042,
|
|
"grad_norm": 3.095620632171631,
|
|
"learning_rate": 9.998836605558635e-06,
|
|
"loss": 1.2473,
|
|
"step": 538
|
|
},
|
|
{
|
|
"epoch": 0.2834604259794899,
|
|
"grad_norm": 3.0324721336364746,
|
|
"learning_rate": 9.998798775568523e-06,
|
|
"loss": 1.2585,
|
|
"step": 539
|
|
},
|
|
{
|
|
"epoch": 0.2839863265842756,
|
|
"grad_norm": 3.6475131511688232,
|
|
"learning_rate": 9.99876034039629e-06,
|
|
"loss": 1.4038,
|
|
"step": 540
|
|
},
|
|
{
|
|
"epoch": 0.2845122271890613,
|
|
"grad_norm": 3.099637269973755,
|
|
"learning_rate": 9.998721300046589e-06,
|
|
"loss": 1.3211,
|
|
"step": 541
|
|
},
|
|
{
|
|
"epoch": 0.28503812779384696,
|
|
"grad_norm": 3.1605703830718994,
|
|
"learning_rate": 9.998681654524145e-06,
|
|
"loss": 1.3125,
|
|
"step": 542
|
|
},
|
|
{
|
|
"epoch": 0.28556402839863265,
|
|
"grad_norm": 3.5903327465057373,
|
|
"learning_rate": 9.998641403833757e-06,
|
|
"loss": 1.3652,
|
|
"step": 543
|
|
},
|
|
{
|
|
"epoch": 0.28608992900341834,
|
|
"grad_norm": 3.3279531002044678,
|
|
"learning_rate": 9.998600547980302e-06,
|
|
"loss": 1.2058,
|
|
"step": 544
|
|
},
|
|
{
|
|
"epoch": 0.28661582960820403,
|
|
"grad_norm": 3.234729766845703,
|
|
"learning_rate": 9.998559086968722e-06,
|
|
"loss": 1.2589,
|
|
"step": 545
|
|
},
|
|
{
|
|
"epoch": 0.2871417302129897,
|
|
"grad_norm": 3.1457557678222656,
|
|
"learning_rate": 9.99851702080404e-06,
|
|
"loss": 1.1593,
|
|
"step": 546
|
|
},
|
|
{
|
|
"epoch": 0.28766763081777547,
|
|
"grad_norm": 3.183593273162842,
|
|
"learning_rate": 9.998474349491345e-06,
|
|
"loss": 1.2526,
|
|
"step": 547
|
|
},
|
|
{
|
|
"epoch": 0.28819353142256116,
|
|
"grad_norm": 3.6854939460754395,
|
|
"learning_rate": 9.998431073035804e-06,
|
|
"loss": 1.3036,
|
|
"step": 548
|
|
},
|
|
{
|
|
"epoch": 0.28871943202734685,
|
|
"grad_norm": 3.4136545658111572,
|
|
"learning_rate": 9.99838719144266e-06,
|
|
"loss": 1.1874,
|
|
"step": 549
|
|
},
|
|
{
|
|
"epoch": 0.28924533263213253,
|
|
"grad_norm": 3.091783046722412,
|
|
"learning_rate": 9.998342704717221e-06,
|
|
"loss": 1.1433,
|
|
"step": 550
|
|
},
|
|
{
|
|
"epoch": 0.2897712332369182,
|
|
"grad_norm": 2.959890365600586,
|
|
"learning_rate": 9.998297612864874e-06,
|
|
"loss": 1.2287,
|
|
"step": 551
|
|
},
|
|
{
|
|
"epoch": 0.2902971338417039,
|
|
"grad_norm": 3.221503973007202,
|
|
"learning_rate": 9.99825191589108e-06,
|
|
"loss": 1.2151,
|
|
"step": 552
|
|
},
|
|
{
|
|
"epoch": 0.2908230344464896,
|
|
"grad_norm": 3.3817384243011475,
|
|
"learning_rate": 9.998205613801369e-06,
|
|
"loss": 1.2449,
|
|
"step": 553
|
|
},
|
|
{
|
|
"epoch": 0.2913489350512753,
|
|
"grad_norm": 3.083874225616455,
|
|
"learning_rate": 9.998158706601348e-06,
|
|
"loss": 1.2687,
|
|
"step": 554
|
|
},
|
|
{
|
|
"epoch": 0.291874835656061,
|
|
"grad_norm": 3.409015655517578,
|
|
"learning_rate": 9.998111194296695e-06,
|
|
"loss": 1.2165,
|
|
"step": 555
|
|
},
|
|
{
|
|
"epoch": 0.2924007362608467,
|
|
"grad_norm": 3.4030258655548096,
|
|
"learning_rate": 9.998063076893164e-06,
|
|
"loss": 1.3351,
|
|
"step": 556
|
|
},
|
|
{
|
|
"epoch": 0.2929266368656324,
|
|
"grad_norm": 3.1987433433532715,
|
|
"learning_rate": 9.998014354396577e-06,
|
|
"loss": 1.2701,
|
|
"step": 557
|
|
},
|
|
{
|
|
"epoch": 0.2934525374704181,
|
|
"grad_norm": 3.245910882949829,
|
|
"learning_rate": 9.997965026812837e-06,
|
|
"loss": 1.2656,
|
|
"step": 558
|
|
},
|
|
{
|
|
"epoch": 0.2939784380752038,
|
|
"grad_norm": 3.811640977859497,
|
|
"learning_rate": 9.997915094147913e-06,
|
|
"loss": 1.2144,
|
|
"step": 559
|
|
},
|
|
{
|
|
"epoch": 0.2945043386799895,
|
|
"grad_norm": 3.237927198410034,
|
|
"learning_rate": 9.997864556407848e-06,
|
|
"loss": 1.2638,
|
|
"step": 560
|
|
},
|
|
{
|
|
"epoch": 0.2950302392847752,
|
|
"grad_norm": 3.274254083633423,
|
|
"learning_rate": 9.997813413598763e-06,
|
|
"loss": 1.1867,
|
|
"step": 561
|
|
},
|
|
{
|
|
"epoch": 0.29555613988956086,
|
|
"grad_norm": 3.227024555206299,
|
|
"learning_rate": 9.99776166572685e-06,
|
|
"loss": 1.2588,
|
|
"step": 562
|
|
},
|
|
{
|
|
"epoch": 0.29608204049434655,
|
|
"grad_norm": 3.414257764816284,
|
|
"learning_rate": 9.997709312798373e-06,
|
|
"loss": 1.2756,
|
|
"step": 563
|
|
},
|
|
{
|
|
"epoch": 0.29660794109913224,
|
|
"grad_norm": 3.4572484493255615,
|
|
"learning_rate": 9.997656354819674e-06,
|
|
"loss": 1.2641,
|
|
"step": 564
|
|
},
|
|
{
|
|
"epoch": 0.29713384170391793,
|
|
"grad_norm": 3.258253812789917,
|
|
"learning_rate": 9.997602791797156e-06,
|
|
"loss": 1.2917,
|
|
"step": 565
|
|
},
|
|
{
|
|
"epoch": 0.2976597423087037,
|
|
"grad_norm": 4.144758701324463,
|
|
"learning_rate": 9.997548623737312e-06,
|
|
"loss": 1.3437,
|
|
"step": 566
|
|
},
|
|
{
|
|
"epoch": 0.29818564291348937,
|
|
"grad_norm": 3.1915879249572754,
|
|
"learning_rate": 9.997493850646696e-06,
|
|
"loss": 1.216,
|
|
"step": 567
|
|
},
|
|
{
|
|
"epoch": 0.29871154351827506,
|
|
"grad_norm": 3.2920475006103516,
|
|
"learning_rate": 9.997438472531937e-06,
|
|
"loss": 1.0558,
|
|
"step": 568
|
|
},
|
|
{
|
|
"epoch": 0.29923744412306075,
|
|
"grad_norm": 3.3217265605926514,
|
|
"learning_rate": 9.997382489399744e-06,
|
|
"loss": 1.3056,
|
|
"step": 569
|
|
},
|
|
{
|
|
"epoch": 0.29976334472784644,
|
|
"grad_norm": 3.278029441833496,
|
|
"learning_rate": 9.997325901256891e-06,
|
|
"loss": 1.3306,
|
|
"step": 570
|
|
},
|
|
{
|
|
"epoch": 0.3002892453326321,
|
|
"grad_norm": 3.1918156147003174,
|
|
"learning_rate": 9.997268708110232e-06,
|
|
"loss": 1.3185,
|
|
"step": 571
|
|
},
|
|
{
|
|
"epoch": 0.3008151459374178,
|
|
"grad_norm": 2.9795260429382324,
|
|
"learning_rate": 9.99721090996669e-06,
|
|
"loss": 1.1595,
|
|
"step": 572
|
|
},
|
|
{
|
|
"epoch": 0.3013410465422035,
|
|
"grad_norm": 3.442984104156494,
|
|
"learning_rate": 9.99715250683326e-06,
|
|
"loss": 1.2518,
|
|
"step": 573
|
|
},
|
|
{
|
|
"epoch": 0.3018669471469892,
|
|
"grad_norm": 2.9818856716156006,
|
|
"learning_rate": 9.997093498717016e-06,
|
|
"loss": 1.2446,
|
|
"step": 574
|
|
},
|
|
{
|
|
"epoch": 0.30239284775177494,
|
|
"grad_norm": 2.9826998710632324,
|
|
"learning_rate": 9.9970338856251e-06,
|
|
"loss": 1.1874,
|
|
"step": 575
|
|
},
|
|
{
|
|
"epoch": 0.30291874835656063,
|
|
"grad_norm": 3.0807487964630127,
|
|
"learning_rate": 9.996973667564728e-06,
|
|
"loss": 1.2509,
|
|
"step": 576
|
|
},
|
|
{
|
|
"epoch": 0.3034446489613463,
|
|
"grad_norm": 2.9740095138549805,
|
|
"learning_rate": 9.996912844543192e-06,
|
|
"loss": 1.1973,
|
|
"step": 577
|
|
},
|
|
{
|
|
"epoch": 0.303970549566132,
|
|
"grad_norm": 3.0660805702209473,
|
|
"learning_rate": 9.996851416567856e-06,
|
|
"loss": 1.1904,
|
|
"step": 578
|
|
},
|
|
{
|
|
"epoch": 0.3044964501709177,
|
|
"grad_norm": 3.129523754119873,
|
|
"learning_rate": 9.996789383646157e-06,
|
|
"loss": 1.1967,
|
|
"step": 579
|
|
},
|
|
{
|
|
"epoch": 0.3050223507757034,
|
|
"grad_norm": 3.330491065979004,
|
|
"learning_rate": 9.996726745785605e-06,
|
|
"loss": 1.3176,
|
|
"step": 580
|
|
},
|
|
{
|
|
"epoch": 0.3055482513804891,
|
|
"grad_norm": 3.359264612197876,
|
|
"learning_rate": 9.99666350299378e-06,
|
|
"loss": 1.2787,
|
|
"step": 581
|
|
},
|
|
{
|
|
"epoch": 0.30607415198527477,
|
|
"grad_norm": 3.1598410606384277,
|
|
"learning_rate": 9.996599655278341e-06,
|
|
"loss": 1.1704,
|
|
"step": 582
|
|
},
|
|
{
|
|
"epoch": 0.30660005259006046,
|
|
"grad_norm": 3.4885876178741455,
|
|
"learning_rate": 9.99653520264702e-06,
|
|
"loss": 1.2532,
|
|
"step": 583
|
|
},
|
|
{
|
|
"epoch": 0.3071259531948462,
|
|
"grad_norm": 3.4063632488250732,
|
|
"learning_rate": 9.996470145107616e-06,
|
|
"loss": 1.403,
|
|
"step": 584
|
|
},
|
|
{
|
|
"epoch": 0.3076518537996319,
|
|
"grad_norm": 3.2507660388946533,
|
|
"learning_rate": 9.996404482668007e-06,
|
|
"loss": 1.2313,
|
|
"step": 585
|
|
},
|
|
{
|
|
"epoch": 0.3081777544044176,
|
|
"grad_norm": 3.1777806282043457,
|
|
"learning_rate": 9.996338215336143e-06,
|
|
"loss": 1.281,
|
|
"step": 586
|
|
},
|
|
{
|
|
"epoch": 0.30870365500920327,
|
|
"grad_norm": 3.1093013286590576,
|
|
"learning_rate": 9.996271343120045e-06,
|
|
"loss": 1.2614,
|
|
"step": 587
|
|
},
|
|
{
|
|
"epoch": 0.30922955561398896,
|
|
"grad_norm": 3.150723934173584,
|
|
"learning_rate": 9.996203866027811e-06,
|
|
"loss": 1.2264,
|
|
"step": 588
|
|
},
|
|
{
|
|
"epoch": 0.30975545621877465,
|
|
"grad_norm": 3.1520981788635254,
|
|
"learning_rate": 9.99613578406761e-06,
|
|
"loss": 1.2726,
|
|
"step": 589
|
|
},
|
|
{
|
|
"epoch": 0.31028135682356034,
|
|
"grad_norm": 3.0869760513305664,
|
|
"learning_rate": 9.996067097247682e-06,
|
|
"loss": 1.2402,
|
|
"step": 590
|
|
},
|
|
{
|
|
"epoch": 0.31080725742834603,
|
|
"grad_norm": 3.2703399658203125,
|
|
"learning_rate": 9.995997805576345e-06,
|
|
"loss": 1.3074,
|
|
"step": 591
|
|
},
|
|
{
|
|
"epoch": 0.3113331580331317,
|
|
"grad_norm": 3.0706095695495605,
|
|
"learning_rate": 9.995927909061987e-06,
|
|
"loss": 1.172,
|
|
"step": 592
|
|
},
|
|
{
|
|
"epoch": 0.3118590586379174,
|
|
"grad_norm": 3.130279779434204,
|
|
"learning_rate": 9.99585740771307e-06,
|
|
"loss": 1.1334,
|
|
"step": 593
|
|
},
|
|
{
|
|
"epoch": 0.31238495924270315,
|
|
"grad_norm": 3.230865478515625,
|
|
"learning_rate": 9.995786301538129e-06,
|
|
"loss": 1.3068,
|
|
"step": 594
|
|
},
|
|
{
|
|
"epoch": 0.31291085984748884,
|
|
"grad_norm": 3.528940439224243,
|
|
"learning_rate": 9.995714590545773e-06,
|
|
"loss": 1.2647,
|
|
"step": 595
|
|
},
|
|
{
|
|
"epoch": 0.31343676045227453,
|
|
"grad_norm": 3.3343887329101562,
|
|
"learning_rate": 9.995642274744684e-06,
|
|
"loss": 1.2635,
|
|
"step": 596
|
|
},
|
|
{
|
|
"epoch": 0.3139626610570602,
|
|
"grad_norm": 3.4318997859954834,
|
|
"learning_rate": 9.995569354143615e-06,
|
|
"loss": 1.2613,
|
|
"step": 597
|
|
},
|
|
{
|
|
"epoch": 0.3144885616618459,
|
|
"grad_norm": 3.155257225036621,
|
|
"learning_rate": 9.995495828751399e-06,
|
|
"loss": 1.224,
|
|
"step": 598
|
|
},
|
|
{
|
|
"epoch": 0.3150144622666316,
|
|
"grad_norm": 3.1119210720062256,
|
|
"learning_rate": 9.99542169857693e-06,
|
|
"loss": 1.2364,
|
|
"step": 599
|
|
},
|
|
{
|
|
"epoch": 0.3155403628714173,
|
|
"grad_norm": 3.0402979850769043,
|
|
"learning_rate": 9.995346963629189e-06,
|
|
"loss": 1.2431,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 0.316066263476203,
|
|
"grad_norm": 3.1695780754089355,
|
|
"learning_rate": 9.99527162391722e-06,
|
|
"loss": 1.2827,
|
|
"step": 601
|
|
},
|
|
{
|
|
"epoch": 0.31659216408098867,
|
|
"grad_norm": 2.9504966735839844,
|
|
"learning_rate": 9.995195679450147e-06,
|
|
"loss": 1.1776,
|
|
"step": 602
|
|
},
|
|
{
|
|
"epoch": 0.3171180646857744,
|
|
"grad_norm": 3.2845983505249023,
|
|
"learning_rate": 9.995119130237162e-06,
|
|
"loss": 1.1915,
|
|
"step": 603
|
|
},
|
|
{
|
|
"epoch": 0.3176439652905601,
|
|
"grad_norm": 3.1545157432556152,
|
|
"learning_rate": 9.995041976287534e-06,
|
|
"loss": 1.342,
|
|
"step": 604
|
|
},
|
|
{
|
|
"epoch": 0.3181698658953458,
|
|
"grad_norm": 3.204342842102051,
|
|
"learning_rate": 9.994964217610602e-06,
|
|
"loss": 1.1737,
|
|
"step": 605
|
|
},
|
|
{
|
|
"epoch": 0.3186957665001315,
|
|
"grad_norm": 3.186814069747925,
|
|
"learning_rate": 9.994885854215782e-06,
|
|
"loss": 1.2916,
|
|
"step": 606
|
|
},
|
|
{
|
|
"epoch": 0.3192216671049172,
|
|
"grad_norm": 3.2076594829559326,
|
|
"learning_rate": 9.994806886112558e-06,
|
|
"loss": 1.3616,
|
|
"step": 607
|
|
},
|
|
{
|
|
"epoch": 0.31974756770970286,
|
|
"grad_norm": 3.1973483562469482,
|
|
"learning_rate": 9.994727313310494e-06,
|
|
"loss": 1.262,
|
|
"step": 608
|
|
},
|
|
{
|
|
"epoch": 0.32027346831448855,
|
|
"grad_norm": 3.2398462295532227,
|
|
"learning_rate": 9.994647135819222e-06,
|
|
"loss": 1.1339,
|
|
"step": 609
|
|
},
|
|
{
|
|
"epoch": 0.32079936891927424,
|
|
"grad_norm": 3.128938913345337,
|
|
"learning_rate": 9.994566353648447e-06,
|
|
"loss": 1.2077,
|
|
"step": 610
|
|
},
|
|
{
|
|
"epoch": 0.32132526952405993,
|
|
"grad_norm": 3.08506441116333,
|
|
"learning_rate": 9.994484966807951e-06,
|
|
"loss": 1.2335,
|
|
"step": 611
|
|
},
|
|
{
|
|
"epoch": 0.3218511701288456,
|
|
"grad_norm": 3.307058095932007,
|
|
"learning_rate": 9.994402975307587e-06,
|
|
"loss": 1.1676,
|
|
"step": 612
|
|
},
|
|
{
|
|
"epoch": 0.32237707073363137,
|
|
"grad_norm": 3.0307862758636475,
|
|
"learning_rate": 9.99432037915728e-06,
|
|
"loss": 1.1794,
|
|
"step": 613
|
|
},
|
|
{
|
|
"epoch": 0.32290297133841706,
|
|
"grad_norm": 3.436113119125366,
|
|
"learning_rate": 9.994237178367032e-06,
|
|
"loss": 1.3283,
|
|
"step": 614
|
|
},
|
|
{
|
|
"epoch": 0.32342887194320274,
|
|
"grad_norm": 3.5635712146759033,
|
|
"learning_rate": 9.994153372946912e-06,
|
|
"loss": 1.2156,
|
|
"step": 615
|
|
},
|
|
{
|
|
"epoch": 0.32395477254798843,
|
|
"grad_norm": 3.0984458923339844,
|
|
"learning_rate": 9.99406896290707e-06,
|
|
"loss": 1.2919,
|
|
"step": 616
|
|
},
|
|
{
|
|
"epoch": 0.3244806731527741,
|
|
"grad_norm": 3.1586906909942627,
|
|
"learning_rate": 9.993983948257724e-06,
|
|
"loss": 1.2819,
|
|
"step": 617
|
|
},
|
|
{
|
|
"epoch": 0.3250065737575598,
|
|
"grad_norm": 3.1414153575897217,
|
|
"learning_rate": 9.993898329009164e-06,
|
|
"loss": 1.3246,
|
|
"step": 618
|
|
},
|
|
{
|
|
"epoch": 0.3255324743623455,
|
|
"grad_norm": 3.3155205249786377,
|
|
"learning_rate": 9.993812105171758e-06,
|
|
"loss": 1.3451,
|
|
"step": 619
|
|
},
|
|
{
|
|
"epoch": 0.3260583749671312,
|
|
"grad_norm": 3.141777515411377,
|
|
"learning_rate": 9.993725276755944e-06,
|
|
"loss": 1.268,
|
|
"step": 620
|
|
},
|
|
{
|
|
"epoch": 0.3265842755719169,
|
|
"grad_norm": 3.215632200241089,
|
|
"learning_rate": 9.993637843772233e-06,
|
|
"loss": 1.2924,
|
|
"step": 621
|
|
},
|
|
{
|
|
"epoch": 0.3271101761767026,
|
|
"grad_norm": 3.0507137775421143,
|
|
"learning_rate": 9.99354980623121e-06,
|
|
"loss": 1.2642,
|
|
"step": 622
|
|
},
|
|
{
|
|
"epoch": 0.3276360767814883,
|
|
"grad_norm": 3.095705270767212,
|
|
"learning_rate": 9.993461164143537e-06,
|
|
"loss": 1.1749,
|
|
"step": 623
|
|
},
|
|
{
|
|
"epoch": 0.328161977386274,
|
|
"grad_norm": 3.139889717102051,
|
|
"learning_rate": 9.993371917519941e-06,
|
|
"loss": 1.2199,
|
|
"step": 624
|
|
},
|
|
{
|
|
"epoch": 0.3286878779910597,
|
|
"grad_norm": 3.1935482025146484,
|
|
"learning_rate": 9.993282066371229e-06,
|
|
"loss": 1.2563,
|
|
"step": 625
|
|
},
|
|
{
|
|
"epoch": 0.3292137785958454,
|
|
"grad_norm": 3.1031148433685303,
|
|
"learning_rate": 9.99319161070828e-06,
|
|
"loss": 1.3186,
|
|
"step": 626
|
|
},
|
|
{
|
|
"epoch": 0.3297396792006311,
|
|
"grad_norm": 3.002609968185425,
|
|
"learning_rate": 9.993100550542042e-06,
|
|
"loss": 1.3023,
|
|
"step": 627
|
|
},
|
|
{
|
|
"epoch": 0.33026557980541676,
|
|
"grad_norm": 3.0571510791778564,
|
|
"learning_rate": 9.993008885883539e-06,
|
|
"loss": 1.2056,
|
|
"step": 628
|
|
},
|
|
{
|
|
"epoch": 0.33079148041020245,
|
|
"grad_norm": 3.1616172790527344,
|
|
"learning_rate": 9.99291661674387e-06,
|
|
"loss": 1.2668,
|
|
"step": 629
|
|
},
|
|
{
|
|
"epoch": 0.33131738101498814,
|
|
"grad_norm": 3.3690426349639893,
|
|
"learning_rate": 9.992823743134209e-06,
|
|
"loss": 1.2518,
|
|
"step": 630
|
|
},
|
|
{
|
|
"epoch": 0.3318432816197739,
|
|
"grad_norm": 3.3865299224853516,
|
|
"learning_rate": 9.992730265065793e-06,
|
|
"loss": 1.3286,
|
|
"step": 631
|
|
},
|
|
{
|
|
"epoch": 0.3323691822245596,
|
|
"grad_norm": 3.290006637573242,
|
|
"learning_rate": 9.992636182549944e-06,
|
|
"loss": 1.2331,
|
|
"step": 632
|
|
},
|
|
{
|
|
"epoch": 0.33289508282934527,
|
|
"grad_norm": 3.0594630241394043,
|
|
"learning_rate": 9.992541495598051e-06,
|
|
"loss": 1.2704,
|
|
"step": 633
|
|
},
|
|
{
|
|
"epoch": 0.33342098343413096,
|
|
"grad_norm": 3.0419819355010986,
|
|
"learning_rate": 9.992446204221577e-06,
|
|
"loss": 1.1968,
|
|
"step": 634
|
|
},
|
|
{
|
|
"epoch": 0.33394688403891665,
|
|
"grad_norm": 3.124880313873291,
|
|
"learning_rate": 9.992350308432058e-06,
|
|
"loss": 1.2575,
|
|
"step": 635
|
|
},
|
|
{
|
|
"epoch": 0.33447278464370234,
|
|
"grad_norm": 3.35003399848938,
|
|
"learning_rate": 9.992253808241104e-06,
|
|
"loss": 1.2935,
|
|
"step": 636
|
|
},
|
|
{
|
|
"epoch": 0.334998685248488,
|
|
"grad_norm": 3.0908632278442383,
|
|
"learning_rate": 9.992156703660397e-06,
|
|
"loss": 1.3128,
|
|
"step": 637
|
|
},
|
|
{
|
|
"epoch": 0.3355245858532737,
|
|
"grad_norm": 3.111051559448242,
|
|
"learning_rate": 9.992058994701694e-06,
|
|
"loss": 1.1062,
|
|
"step": 638
|
|
},
|
|
{
|
|
"epoch": 0.3360504864580594,
|
|
"grad_norm": 3.4956271648406982,
|
|
"learning_rate": 9.991960681376827e-06,
|
|
"loss": 1.2558,
|
|
"step": 639
|
|
},
|
|
{
|
|
"epoch": 0.3365763870628451,
|
|
"grad_norm": 3.39172625541687,
|
|
"learning_rate": 9.991861763697692e-06,
|
|
"loss": 1.3707,
|
|
"step": 640
|
|
},
|
|
{
|
|
"epoch": 0.33710228766763084,
|
|
"grad_norm": 3.094860792160034,
|
|
"learning_rate": 9.99176224167627e-06,
|
|
"loss": 1.3354,
|
|
"step": 641
|
|
},
|
|
{
|
|
"epoch": 0.33762818827241653,
|
|
"grad_norm": 3.3545854091644287,
|
|
"learning_rate": 9.991662115324607e-06,
|
|
"loss": 1.3797,
|
|
"step": 642
|
|
},
|
|
{
|
|
"epoch": 0.3381540888772022,
|
|
"grad_norm": 3.178557872772217,
|
|
"learning_rate": 9.991561384654825e-06,
|
|
"loss": 1.2928,
|
|
"step": 643
|
|
},
|
|
{
|
|
"epoch": 0.3386799894819879,
|
|
"grad_norm": 3.08526611328125,
|
|
"learning_rate": 9.99146004967912e-06,
|
|
"loss": 1.1733,
|
|
"step": 644
|
|
},
|
|
{
|
|
"epoch": 0.3392058900867736,
|
|
"grad_norm": 3.3157966136932373,
|
|
"learning_rate": 9.991358110409761e-06,
|
|
"loss": 1.2552,
|
|
"step": 645
|
|
},
|
|
{
|
|
"epoch": 0.3397317906915593,
|
|
"grad_norm": 3.2351930141448975,
|
|
"learning_rate": 9.991255566859086e-06,
|
|
"loss": 1.3581,
|
|
"step": 646
|
|
},
|
|
{
|
|
"epoch": 0.340257691296345,
|
|
"grad_norm": 2.8856968879699707,
|
|
"learning_rate": 9.991152419039514e-06,
|
|
"loss": 1.1648,
|
|
"step": 647
|
|
},
|
|
{
|
|
"epoch": 0.34078359190113067,
|
|
"grad_norm": 2.8800039291381836,
|
|
"learning_rate": 9.991048666963527e-06,
|
|
"loss": 1.2132,
|
|
"step": 648
|
|
},
|
|
{
|
|
"epoch": 0.34130949250591636,
|
|
"grad_norm": 3.1628317832946777,
|
|
"learning_rate": 9.99094431064369e-06,
|
|
"loss": 1.3036,
|
|
"step": 649
|
|
},
|
|
{
|
|
"epoch": 0.3418353931107021,
|
|
"grad_norm": 3.1099801063537598,
|
|
"learning_rate": 9.990839350092638e-06,
|
|
"loss": 1.2091,
|
|
"step": 650
|
|
},
|
|
{
|
|
"epoch": 0.3423612937154878,
|
|
"grad_norm": 3.030766010284424,
|
|
"learning_rate": 9.990733785323074e-06,
|
|
"loss": 1.2926,
|
|
"step": 651
|
|
},
|
|
{
|
|
"epoch": 0.3428871943202735,
|
|
"grad_norm": 3.0072193145751953,
|
|
"learning_rate": 9.99062761634778e-06,
|
|
"loss": 1.2235,
|
|
"step": 652
|
|
},
|
|
{
|
|
"epoch": 0.34341309492505917,
|
|
"grad_norm": 3.19268536567688,
|
|
"learning_rate": 9.990520843179611e-06,
|
|
"loss": 1.3277,
|
|
"step": 653
|
|
},
|
|
{
|
|
"epoch": 0.34393899552984486,
|
|
"grad_norm": 2.97922945022583,
|
|
"learning_rate": 9.990413465831493e-06,
|
|
"loss": 1.1911,
|
|
"step": 654
|
|
},
|
|
{
|
|
"epoch": 0.34446489613463055,
|
|
"grad_norm": 3.189166784286499,
|
|
"learning_rate": 9.990305484316424e-06,
|
|
"loss": 1.3475,
|
|
"step": 655
|
|
},
|
|
{
|
|
"epoch": 0.34499079673941624,
|
|
"grad_norm": 3.045102119445801,
|
|
"learning_rate": 9.990196898647478e-06,
|
|
"loss": 1.2649,
|
|
"step": 656
|
|
},
|
|
{
|
|
"epoch": 0.34551669734420193,
|
|
"grad_norm": 2.9332075119018555,
|
|
"learning_rate": 9.9900877088378e-06,
|
|
"loss": 1.1504,
|
|
"step": 657
|
|
},
|
|
{
|
|
"epoch": 0.3460425979489876,
|
|
"grad_norm": 3.2756752967834473,
|
|
"learning_rate": 9.98997791490061e-06,
|
|
"loss": 1.2597,
|
|
"step": 658
|
|
},
|
|
{
|
|
"epoch": 0.34656849855377336,
|
|
"grad_norm": 3.4546256065368652,
|
|
"learning_rate": 9.989867516849198e-06,
|
|
"loss": 1.2669,
|
|
"step": 659
|
|
},
|
|
{
|
|
"epoch": 0.34709439915855905,
|
|
"grad_norm": 2.9611923694610596,
|
|
"learning_rate": 9.989756514696936e-06,
|
|
"loss": 1.1428,
|
|
"step": 660
|
|
},
|
|
{
|
|
"epoch": 0.34762029976334474,
|
|
"grad_norm": 3.2649290561676025,
|
|
"learning_rate": 9.989644908457255e-06,
|
|
"loss": 1.3483,
|
|
"step": 661
|
|
},
|
|
{
|
|
"epoch": 0.34814620036813043,
|
|
"grad_norm": 2.9122157096862793,
|
|
"learning_rate": 9.98953269814367e-06,
|
|
"loss": 1.2001,
|
|
"step": 662
|
|
},
|
|
{
|
|
"epoch": 0.3486721009729161,
|
|
"grad_norm": 3.5231406688690186,
|
|
"learning_rate": 9.989419883769767e-06,
|
|
"loss": 1.2508,
|
|
"step": 663
|
|
},
|
|
{
|
|
"epoch": 0.3491980015777018,
|
|
"grad_norm": 2.958273410797119,
|
|
"learning_rate": 9.9893064653492e-06,
|
|
"loss": 1.2359,
|
|
"step": 664
|
|
},
|
|
{
|
|
"epoch": 0.3497239021824875,
|
|
"grad_norm": 3.169466495513916,
|
|
"learning_rate": 9.989192442895704e-06,
|
|
"loss": 1.2287,
|
|
"step": 665
|
|
},
|
|
{
|
|
"epoch": 0.3502498027872732,
|
|
"grad_norm": 3.058429479598999,
|
|
"learning_rate": 9.989077816423083e-06,
|
|
"loss": 1.3188,
|
|
"step": 666
|
|
},
|
|
{
|
|
"epoch": 0.3507757033920589,
|
|
"grad_norm": 2.9665424823760986,
|
|
"learning_rate": 9.98896258594521e-06,
|
|
"loss": 1.2053,
|
|
"step": 667
|
|
},
|
|
{
|
|
"epoch": 0.35130160399684457,
|
|
"grad_norm": 2.9990475177764893,
|
|
"learning_rate": 9.98884675147604e-06,
|
|
"loss": 1.2354,
|
|
"step": 668
|
|
},
|
|
{
|
|
"epoch": 0.3518275046016303,
|
|
"grad_norm": 3.090883731842041,
|
|
"learning_rate": 9.988730313029596e-06,
|
|
"loss": 1.2832,
|
|
"step": 669
|
|
},
|
|
{
|
|
"epoch": 0.352353405206416,
|
|
"grad_norm": 3.1802048683166504,
|
|
"learning_rate": 9.988613270619973e-06,
|
|
"loss": 1.1666,
|
|
"step": 670
|
|
},
|
|
{
|
|
"epoch": 0.3528793058112017,
|
|
"grad_norm": 2.9730775356292725,
|
|
"learning_rate": 9.988495624261341e-06,
|
|
"loss": 1.1778,
|
|
"step": 671
|
|
},
|
|
{
|
|
"epoch": 0.3534052064159874,
|
|
"grad_norm": 2.845747470855713,
|
|
"learning_rate": 9.988377373967944e-06,
|
|
"loss": 1.0983,
|
|
"step": 672
|
|
},
|
|
{
|
|
"epoch": 0.3539311070207731,
|
|
"grad_norm": 3.422753095626831,
|
|
"learning_rate": 9.988258519754096e-06,
|
|
"loss": 1.3652,
|
|
"step": 673
|
|
},
|
|
{
|
|
"epoch": 0.35445700762555876,
|
|
"grad_norm": 3.291478157043457,
|
|
"learning_rate": 9.988139061634188e-06,
|
|
"loss": 1.4272,
|
|
"step": 674
|
|
},
|
|
{
|
|
"epoch": 0.35498290823034445,
|
|
"grad_norm": 3.193990707397461,
|
|
"learning_rate": 9.988018999622683e-06,
|
|
"loss": 1.3671,
|
|
"step": 675
|
|
},
|
|
{
|
|
"epoch": 0.35550880883513014,
|
|
"grad_norm": 3.050693988800049,
|
|
"learning_rate": 9.987898333734114e-06,
|
|
"loss": 1.2825,
|
|
"step": 676
|
|
},
|
|
{
|
|
"epoch": 0.35603470943991583,
|
|
"grad_norm": 2.9304311275482178,
|
|
"learning_rate": 9.98777706398309e-06,
|
|
"loss": 1.1859,
|
|
"step": 677
|
|
},
|
|
{
|
|
"epoch": 0.3565606100447016,
|
|
"grad_norm": 2.9854581356048584,
|
|
"learning_rate": 9.987655190384295e-06,
|
|
"loss": 1.2926,
|
|
"step": 678
|
|
},
|
|
{
|
|
"epoch": 0.35708651064948727,
|
|
"grad_norm": 3.269815683364868,
|
|
"learning_rate": 9.987532712952483e-06,
|
|
"loss": 1.3136,
|
|
"step": 679
|
|
},
|
|
{
|
|
"epoch": 0.35761241125427295,
|
|
"grad_norm": 2.8265762329101562,
|
|
"learning_rate": 9.987409631702478e-06,
|
|
"loss": 1.207,
|
|
"step": 680
|
|
},
|
|
{
|
|
"epoch": 0.35813831185905864,
|
|
"grad_norm": 3.4381608963012695,
|
|
"learning_rate": 9.987285946649186e-06,
|
|
"loss": 1.3513,
|
|
"step": 681
|
|
},
|
|
{
|
|
"epoch": 0.35866421246384433,
|
|
"grad_norm": 2.863412380218506,
|
|
"learning_rate": 9.987161657807576e-06,
|
|
"loss": 1.2084,
|
|
"step": 682
|
|
},
|
|
{
|
|
"epoch": 0.35919011306863,
|
|
"grad_norm": 3.1376121044158936,
|
|
"learning_rate": 9.9870367651927e-06,
|
|
"loss": 1.4213,
|
|
"step": 683
|
|
},
|
|
{
|
|
"epoch": 0.3597160136734157,
|
|
"grad_norm": 2.957144021987915,
|
|
"learning_rate": 9.986911268819674e-06,
|
|
"loss": 1.2207,
|
|
"step": 684
|
|
},
|
|
{
|
|
"epoch": 0.3602419142782014,
|
|
"grad_norm": 3.1863062381744385,
|
|
"learning_rate": 9.986785168703694e-06,
|
|
"loss": 1.2583,
|
|
"step": 685
|
|
},
|
|
{
|
|
"epoch": 0.3607678148829871,
|
|
"grad_norm": 3.0562474727630615,
|
|
"learning_rate": 9.986658464860027e-06,
|
|
"loss": 1.2713,
|
|
"step": 686
|
|
},
|
|
{
|
|
"epoch": 0.36129371548777284,
|
|
"grad_norm": 3.1078624725341797,
|
|
"learning_rate": 9.98653115730401e-06,
|
|
"loss": 1.2265,
|
|
"step": 687
|
|
},
|
|
{
|
|
"epoch": 0.3618196160925585,
|
|
"grad_norm": 2.9854729175567627,
|
|
"learning_rate": 9.986403246051055e-06,
|
|
"loss": 1.217,
|
|
"step": 688
|
|
},
|
|
{
|
|
"epoch": 0.3623455166973442,
|
|
"grad_norm": 3.185779571533203,
|
|
"learning_rate": 9.986274731116651e-06,
|
|
"loss": 1.2584,
|
|
"step": 689
|
|
},
|
|
{
|
|
"epoch": 0.3628714173021299,
|
|
"grad_norm": 3.004950761795044,
|
|
"learning_rate": 9.986145612516355e-06,
|
|
"loss": 1.2757,
|
|
"step": 690
|
|
},
|
|
{
|
|
"epoch": 0.3633973179069156,
|
|
"grad_norm": 3.20649790763855,
|
|
"learning_rate": 9.986015890265799e-06,
|
|
"loss": 1.2702,
|
|
"step": 691
|
|
},
|
|
{
|
|
"epoch": 0.3639232185117013,
|
|
"grad_norm": 3.234217882156372,
|
|
"learning_rate": 9.985885564380686e-06,
|
|
"loss": 1.2481,
|
|
"step": 692
|
|
},
|
|
{
|
|
"epoch": 0.364449119116487,
|
|
"grad_norm": 3.4999547004699707,
|
|
"learning_rate": 9.985754634876796e-06,
|
|
"loss": 1.33,
|
|
"step": 693
|
|
},
|
|
{
|
|
"epoch": 0.36497501972127266,
|
|
"grad_norm": 3.371229410171509,
|
|
"learning_rate": 9.985623101769981e-06,
|
|
"loss": 1.2411,
|
|
"step": 694
|
|
},
|
|
{
|
|
"epoch": 0.36550092032605835,
|
|
"grad_norm": 2.8583662509918213,
|
|
"learning_rate": 9.985490965076164e-06,
|
|
"loss": 1.2422,
|
|
"step": 695
|
|
},
|
|
{
|
|
"epoch": 0.36602682093084404,
|
|
"grad_norm": 3.007049322128296,
|
|
"learning_rate": 9.985358224811341e-06,
|
|
"loss": 1.2716,
|
|
"step": 696
|
|
},
|
|
{
|
|
"epoch": 0.3665527215356298,
|
|
"grad_norm": 2.933000087738037,
|
|
"learning_rate": 9.985224880991582e-06,
|
|
"loss": 1.2607,
|
|
"step": 697
|
|
},
|
|
{
|
|
"epoch": 0.3670786221404155,
|
|
"grad_norm": 3.2839972972869873,
|
|
"learning_rate": 9.985090933633033e-06,
|
|
"loss": 1.2323,
|
|
"step": 698
|
|
},
|
|
{
|
|
"epoch": 0.36760452274520117,
|
|
"grad_norm": 2.8908584117889404,
|
|
"learning_rate": 9.98495638275191e-06,
|
|
"loss": 1.2437,
|
|
"step": 699
|
|
},
|
|
{
|
|
"epoch": 0.36813042334998686,
|
|
"grad_norm": 3.0887835025787354,
|
|
"learning_rate": 9.9848212283645e-06,
|
|
"loss": 1.2571,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 0.36865632395477255,
|
|
"grad_norm": 3.034158229827881,
|
|
"learning_rate": 9.984685470487168e-06,
|
|
"loss": 1.3003,
|
|
"step": 701
|
|
},
|
|
{
|
|
"epoch": 0.36918222455955824,
|
|
"grad_norm": 3.0524158477783203,
|
|
"learning_rate": 9.984549109136346e-06,
|
|
"loss": 1.308,
|
|
"step": 702
|
|
},
|
|
{
|
|
"epoch": 0.3697081251643439,
|
|
"grad_norm": 3.096510887145996,
|
|
"learning_rate": 9.984412144328548e-06,
|
|
"loss": 1.2798,
|
|
"step": 703
|
|
},
|
|
{
|
|
"epoch": 0.3702340257691296,
|
|
"grad_norm": 2.8943581581115723,
|
|
"learning_rate": 9.984274576080352e-06,
|
|
"loss": 1.267,
|
|
"step": 704
|
|
},
|
|
{
|
|
"epoch": 0.3707599263739153,
|
|
"grad_norm": 3.066819667816162,
|
|
"learning_rate": 9.984136404408413e-06,
|
|
"loss": 1.2832,
|
|
"step": 705
|
|
},
|
|
{
|
|
"epoch": 0.37128582697870105,
|
|
"grad_norm": 3.172398328781128,
|
|
"learning_rate": 9.983997629329462e-06,
|
|
"loss": 1.2629,
|
|
"step": 706
|
|
},
|
|
{
|
|
"epoch": 0.37181172758348674,
|
|
"grad_norm": 3.133291721343994,
|
|
"learning_rate": 9.983858250860296e-06,
|
|
"loss": 1.2572,
|
|
"step": 707
|
|
},
|
|
{
|
|
"epoch": 0.37233762818827243,
|
|
"grad_norm": 3.1537201404571533,
|
|
"learning_rate": 9.983718269017791e-06,
|
|
"loss": 1.3283,
|
|
"step": 708
|
|
},
|
|
{
|
|
"epoch": 0.3728635287930581,
|
|
"grad_norm": 3.136389970779419,
|
|
"learning_rate": 9.983577683818891e-06,
|
|
"loss": 1.2711,
|
|
"step": 709
|
|
},
|
|
{
|
|
"epoch": 0.3733894293978438,
|
|
"grad_norm": 3.1949572563171387,
|
|
"learning_rate": 9.98343649528062e-06,
|
|
"loss": 1.3013,
|
|
"step": 710
|
|
},
|
|
{
|
|
"epoch": 0.3739153300026295,
|
|
"grad_norm": 3.0228097438812256,
|
|
"learning_rate": 9.98329470342007e-06,
|
|
"loss": 1.37,
|
|
"step": 711
|
|
},
|
|
{
|
|
"epoch": 0.3744412306074152,
|
|
"grad_norm": 3.0644755363464355,
|
|
"learning_rate": 9.983152308254405e-06,
|
|
"loss": 1.2303,
|
|
"step": 712
|
|
},
|
|
{
|
|
"epoch": 0.3749671312122009,
|
|
"grad_norm": 2.9789505004882812,
|
|
"learning_rate": 9.983009309800866e-06,
|
|
"loss": 1.3372,
|
|
"step": 713
|
|
},
|
|
{
|
|
"epoch": 0.37549303181698657,
|
|
"grad_norm": 2.9281303882598877,
|
|
"learning_rate": 9.982865708076765e-06,
|
|
"loss": 1.2347,
|
|
"step": 714
|
|
},
|
|
{
|
|
"epoch": 0.3760189324217723,
|
|
"grad_norm": 3.1057536602020264,
|
|
"learning_rate": 9.982721503099489e-06,
|
|
"loss": 1.3438,
|
|
"step": 715
|
|
},
|
|
{
|
|
"epoch": 0.376544833026558,
|
|
"grad_norm": 3.012782573699951,
|
|
"learning_rate": 9.982576694886491e-06,
|
|
"loss": 1.0957,
|
|
"step": 716
|
|
},
|
|
{
|
|
"epoch": 0.3770707336313437,
|
|
"grad_norm": 3.0080039501190186,
|
|
"learning_rate": 9.98243128345531e-06,
|
|
"loss": 1.2677,
|
|
"step": 717
|
|
},
|
|
{
|
|
"epoch": 0.3775966342361294,
|
|
"grad_norm": 3.079721450805664,
|
|
"learning_rate": 9.982285268823541e-06,
|
|
"loss": 1.2497,
|
|
"step": 718
|
|
},
|
|
{
|
|
"epoch": 0.37812253484091507,
|
|
"grad_norm": 3.0270872116088867,
|
|
"learning_rate": 9.98213865100887e-06,
|
|
"loss": 1.3338,
|
|
"step": 719
|
|
},
|
|
{
|
|
"epoch": 0.37864843544570076,
|
|
"grad_norm": 3.1811351776123047,
|
|
"learning_rate": 9.981991430029041e-06,
|
|
"loss": 1.2605,
|
|
"step": 720
|
|
},
|
|
{
|
|
"epoch": 0.37917433605048645,
|
|
"grad_norm": 3.1121339797973633,
|
|
"learning_rate": 9.981843605901882e-06,
|
|
"loss": 1.2003,
|
|
"step": 721
|
|
},
|
|
{
|
|
"epoch": 0.37970023665527214,
|
|
"grad_norm": 2.9571330547332764,
|
|
"learning_rate": 9.981695178645286e-06,
|
|
"loss": 1.2758,
|
|
"step": 722
|
|
},
|
|
{
|
|
"epoch": 0.38022613726005783,
|
|
"grad_norm": 3.3723056316375732,
|
|
"learning_rate": 9.981546148277224e-06,
|
|
"loss": 1.3089,
|
|
"step": 723
|
|
},
|
|
{
|
|
"epoch": 0.3807520378648435,
|
|
"grad_norm": 3.1867928504943848,
|
|
"learning_rate": 9.981396514815739e-06,
|
|
"loss": 1.3123,
|
|
"step": 724
|
|
},
|
|
{
|
|
"epoch": 0.38127793846962926,
|
|
"grad_norm": 3.0516862869262695,
|
|
"learning_rate": 9.981246278278944e-06,
|
|
"loss": 1.3021,
|
|
"step": 725
|
|
},
|
|
{
|
|
"epoch": 0.38180383907441495,
|
|
"grad_norm": 3.1751708984375,
|
|
"learning_rate": 9.981095438685031e-06,
|
|
"loss": 1.2328,
|
|
"step": 726
|
|
},
|
|
{
|
|
"epoch": 0.38232973967920064,
|
|
"grad_norm": 3.1216859817504883,
|
|
"learning_rate": 9.980943996052259e-06,
|
|
"loss": 1.266,
|
|
"step": 727
|
|
},
|
|
{
|
|
"epoch": 0.38285564028398633,
|
|
"grad_norm": 3.1756913661956787,
|
|
"learning_rate": 9.980791950398964e-06,
|
|
"loss": 1.2808,
|
|
"step": 728
|
|
},
|
|
{
|
|
"epoch": 0.383381540888772,
|
|
"grad_norm": 3.1451830863952637,
|
|
"learning_rate": 9.980639301743552e-06,
|
|
"loss": 1.397,
|
|
"step": 729
|
|
},
|
|
{
|
|
"epoch": 0.3839074414935577,
|
|
"grad_norm": 3.027385711669922,
|
|
"learning_rate": 9.980486050104503e-06,
|
|
"loss": 1.2311,
|
|
"step": 730
|
|
},
|
|
{
|
|
"epoch": 0.3844333420983434,
|
|
"grad_norm": 3.192145347595215,
|
|
"learning_rate": 9.980332195500374e-06,
|
|
"loss": 1.2895,
|
|
"step": 731
|
|
},
|
|
{
|
|
"epoch": 0.3849592427031291,
|
|
"grad_norm": 3.308378219604492,
|
|
"learning_rate": 9.980177737949786e-06,
|
|
"loss": 1.2738,
|
|
"step": 732
|
|
},
|
|
{
|
|
"epoch": 0.3854851433079148,
|
|
"grad_norm": 3.1952481269836426,
|
|
"learning_rate": 9.980022677471442e-06,
|
|
"loss": 1.3182,
|
|
"step": 733
|
|
},
|
|
{
|
|
"epoch": 0.3860110439127005,
|
|
"grad_norm": 3.0240139961242676,
|
|
"learning_rate": 9.979867014084115e-06,
|
|
"loss": 1.2431,
|
|
"step": 734
|
|
},
|
|
{
|
|
"epoch": 0.3865369445174862,
|
|
"grad_norm": 3.152616500854492,
|
|
"learning_rate": 9.97971074780665e-06,
|
|
"loss": 1.2004,
|
|
"step": 735
|
|
},
|
|
{
|
|
"epoch": 0.3870628451222719,
|
|
"grad_norm": 3.1367688179016113,
|
|
"learning_rate": 9.979553878657964e-06,
|
|
"loss": 1.3002,
|
|
"step": 736
|
|
},
|
|
{
|
|
"epoch": 0.3875887457270576,
|
|
"grad_norm": 3.0055739879608154,
|
|
"learning_rate": 9.979396406657049e-06,
|
|
"loss": 1.2732,
|
|
"step": 737
|
|
},
|
|
{
|
|
"epoch": 0.3881146463318433,
|
|
"grad_norm": 2.9753293991088867,
|
|
"learning_rate": 9.979238331822972e-06,
|
|
"loss": 1.316,
|
|
"step": 738
|
|
},
|
|
{
|
|
"epoch": 0.38864054693662897,
|
|
"grad_norm": 3.1822166442871094,
|
|
"learning_rate": 9.979079654174867e-06,
|
|
"loss": 1.2529,
|
|
"step": 739
|
|
},
|
|
{
|
|
"epoch": 0.38916644754141466,
|
|
"grad_norm": 3.071653366088867,
|
|
"learning_rate": 9.978920373731944e-06,
|
|
"loss": 1.2647,
|
|
"step": 740
|
|
},
|
|
{
|
|
"epoch": 0.38969234814620035,
|
|
"grad_norm": 3.002633571624756,
|
|
"learning_rate": 9.978760490513489e-06,
|
|
"loss": 1.1099,
|
|
"step": 741
|
|
},
|
|
{
|
|
"epoch": 0.39021824875098604,
|
|
"grad_norm": 3.0228371620178223,
|
|
"learning_rate": 9.978600004538858e-06,
|
|
"loss": 1.199,
|
|
"step": 742
|
|
},
|
|
{
|
|
"epoch": 0.3907441493557718,
|
|
"grad_norm": 3.238349199295044,
|
|
"learning_rate": 9.978438915827479e-06,
|
|
"loss": 1.4145,
|
|
"step": 743
|
|
},
|
|
{
|
|
"epoch": 0.3912700499605575,
|
|
"grad_norm": 3.190460443496704,
|
|
"learning_rate": 9.978277224398856e-06,
|
|
"loss": 1.3514,
|
|
"step": 744
|
|
},
|
|
{
|
|
"epoch": 0.39179595056534317,
|
|
"grad_norm": 3.142568349838257,
|
|
"learning_rate": 9.978114930272562e-06,
|
|
"loss": 1.2713,
|
|
"step": 745
|
|
},
|
|
{
|
|
"epoch": 0.39232185117012885,
|
|
"grad_norm": 2.9691591262817383,
|
|
"learning_rate": 9.977952033468247e-06,
|
|
"loss": 1.2301,
|
|
"step": 746
|
|
},
|
|
{
|
|
"epoch": 0.39284775177491454,
|
|
"grad_norm": 3.1980960369110107,
|
|
"learning_rate": 9.97778853400563e-06,
|
|
"loss": 1.2975,
|
|
"step": 747
|
|
},
|
|
{
|
|
"epoch": 0.39337365237970023,
|
|
"grad_norm": 3.1054530143737793,
|
|
"learning_rate": 9.977624431904509e-06,
|
|
"loss": 1.2619,
|
|
"step": 748
|
|
},
|
|
{
|
|
"epoch": 0.3938995529844859,
|
|
"grad_norm": 2.8866591453552246,
|
|
"learning_rate": 9.977459727184748e-06,
|
|
"loss": 1.1567,
|
|
"step": 749
|
|
},
|
|
{
|
|
"epoch": 0.3944254535892716,
|
|
"grad_norm": 2.911992073059082,
|
|
"learning_rate": 9.977294419866287e-06,
|
|
"loss": 1.1874,
|
|
"step": 750
|
|
},
|
|
{
|
|
"epoch": 0.3949513541940573,
|
|
"grad_norm": 3.1664133071899414,
|
|
"learning_rate": 9.97712850996914e-06,
|
|
"loss": 1.2302,
|
|
"step": 751
|
|
},
|
|
{
|
|
"epoch": 0.395477254798843,
|
|
"grad_norm": 3.130518674850464,
|
|
"learning_rate": 9.976961997513395e-06,
|
|
"loss": 1.2908,
|
|
"step": 752
|
|
},
|
|
{
|
|
"epoch": 0.39600315540362874,
|
|
"grad_norm": 2.9121437072753906,
|
|
"learning_rate": 9.976794882519207e-06,
|
|
"loss": 1.3136,
|
|
"step": 753
|
|
},
|
|
{
|
|
"epoch": 0.3965290560084144,
|
|
"grad_norm": 2.877485513687134,
|
|
"learning_rate": 9.97662716500681e-06,
|
|
"loss": 1.2313,
|
|
"step": 754
|
|
},
|
|
{
|
|
"epoch": 0.3970549566132001,
|
|
"grad_norm": 3.031449556350708,
|
|
"learning_rate": 9.976458844996507e-06,
|
|
"loss": 1.1943,
|
|
"step": 755
|
|
},
|
|
{
|
|
"epoch": 0.3975808572179858,
|
|
"grad_norm": 3.186206817626953,
|
|
"learning_rate": 9.97628992250868e-06,
|
|
"loss": 1.2426,
|
|
"step": 756
|
|
},
|
|
{
|
|
"epoch": 0.3981067578227715,
|
|
"grad_norm": 3.2916715145111084,
|
|
"learning_rate": 9.976120397563776e-06,
|
|
"loss": 1.2265,
|
|
"step": 757
|
|
},
|
|
{
|
|
"epoch": 0.3986326584275572,
|
|
"grad_norm": 2.8378243446350098,
|
|
"learning_rate": 9.97595027018232e-06,
|
|
"loss": 1.2717,
|
|
"step": 758
|
|
},
|
|
{
|
|
"epoch": 0.3991585590323429,
|
|
"grad_norm": 3.24177885055542,
|
|
"learning_rate": 9.975779540384907e-06,
|
|
"loss": 1.2853,
|
|
"step": 759
|
|
},
|
|
{
|
|
"epoch": 0.39968445963712856,
|
|
"grad_norm": 3.739731788635254,
|
|
"learning_rate": 9.975608208192206e-06,
|
|
"loss": 1.2507,
|
|
"step": 760
|
|
},
|
|
{
|
|
"epoch": 0.40021036024191425,
|
|
"grad_norm": 2.9595437049865723,
|
|
"learning_rate": 9.975436273624964e-06,
|
|
"loss": 1.233,
|
|
"step": 761
|
|
},
|
|
{
|
|
"epoch": 0.4007362608467,
|
|
"grad_norm": 2.9739890098571777,
|
|
"learning_rate": 9.975263736703992e-06,
|
|
"loss": 1.34,
|
|
"step": 762
|
|
},
|
|
{
|
|
"epoch": 0.4012621614514857,
|
|
"grad_norm": 3.131807327270508,
|
|
"learning_rate": 9.97509059745018e-06,
|
|
"loss": 1.2541,
|
|
"step": 763
|
|
},
|
|
{
|
|
"epoch": 0.4017880620562714,
|
|
"grad_norm": 2.9594075679779053,
|
|
"learning_rate": 9.974916855884488e-06,
|
|
"loss": 1.2259,
|
|
"step": 764
|
|
},
|
|
{
|
|
"epoch": 0.40231396266105707,
|
|
"grad_norm": 2.9331164360046387,
|
|
"learning_rate": 9.97474251202795e-06,
|
|
"loss": 1.2425,
|
|
"step": 765
|
|
},
|
|
{
|
|
"epoch": 0.40283986326584276,
|
|
"grad_norm": 3.0225133895874023,
|
|
"learning_rate": 9.974567565901676e-06,
|
|
"loss": 1.3187,
|
|
"step": 766
|
|
},
|
|
{
|
|
"epoch": 0.40336576387062845,
|
|
"grad_norm": 2.8529632091522217,
|
|
"learning_rate": 9.974392017526843e-06,
|
|
"loss": 1.2553,
|
|
"step": 767
|
|
},
|
|
{
|
|
"epoch": 0.40389166447541414,
|
|
"grad_norm": 2.9452273845672607,
|
|
"learning_rate": 9.974215866924706e-06,
|
|
"loss": 1.2173,
|
|
"step": 768
|
|
},
|
|
{
|
|
"epoch": 0.4044175650801998,
|
|
"grad_norm": 3.171651840209961,
|
|
"learning_rate": 9.974039114116586e-06,
|
|
"loss": 1.2636,
|
|
"step": 769
|
|
},
|
|
{
|
|
"epoch": 0.4049434656849855,
|
|
"grad_norm": 2.9483590126037598,
|
|
"learning_rate": 9.97386175912389e-06,
|
|
"loss": 1.1968,
|
|
"step": 770
|
|
},
|
|
{
|
|
"epoch": 0.40546936628977126,
|
|
"grad_norm": 3.1218364238739014,
|
|
"learning_rate": 9.973683801968081e-06,
|
|
"loss": 1.2941,
|
|
"step": 771
|
|
},
|
|
{
|
|
"epoch": 0.40599526689455695,
|
|
"grad_norm": 3.0449860095977783,
|
|
"learning_rate": 9.973505242670708e-06,
|
|
"loss": 1.3274,
|
|
"step": 772
|
|
},
|
|
{
|
|
"epoch": 0.40652116749934264,
|
|
"grad_norm": 2.885972023010254,
|
|
"learning_rate": 9.97332608125339e-06,
|
|
"loss": 1.2245,
|
|
"step": 773
|
|
},
|
|
{
|
|
"epoch": 0.40704706810412833,
|
|
"grad_norm": 2.974050521850586,
|
|
"learning_rate": 9.973146317737812e-06,
|
|
"loss": 1.2652,
|
|
"step": 774
|
|
},
|
|
{
|
|
"epoch": 0.407572968708914,
|
|
"grad_norm": 2.90822172164917,
|
|
"learning_rate": 9.972965952145742e-06,
|
|
"loss": 1.1369,
|
|
"step": 775
|
|
},
|
|
{
|
|
"epoch": 0.4080988693136997,
|
|
"grad_norm": 2.892280101776123,
|
|
"learning_rate": 9.97278498449901e-06,
|
|
"loss": 1.2155,
|
|
"step": 776
|
|
},
|
|
{
|
|
"epoch": 0.4086247699184854,
|
|
"grad_norm": 2.9581539630889893,
|
|
"learning_rate": 9.972603414819534e-06,
|
|
"loss": 1.2046,
|
|
"step": 777
|
|
},
|
|
{
|
|
"epoch": 0.4091506705232711,
|
|
"grad_norm": 3.164325714111328,
|
|
"learning_rate": 9.972421243129288e-06,
|
|
"loss": 1.2025,
|
|
"step": 778
|
|
},
|
|
{
|
|
"epoch": 0.4096765711280568,
|
|
"grad_norm": 3.383617877960205,
|
|
"learning_rate": 9.97223846945033e-06,
|
|
"loss": 1.2796,
|
|
"step": 779
|
|
},
|
|
{
|
|
"epoch": 0.41020247173284247,
|
|
"grad_norm": 3.4706368446350098,
|
|
"learning_rate": 9.972055093804788e-06,
|
|
"loss": 1.2597,
|
|
"step": 780
|
|
},
|
|
{
|
|
"epoch": 0.4107283723376282,
|
|
"grad_norm": 3.1070616245269775,
|
|
"learning_rate": 9.971871116214863e-06,
|
|
"loss": 1.2662,
|
|
"step": 781
|
|
},
|
|
{
|
|
"epoch": 0.4112542729424139,
|
|
"grad_norm": 3.1561152935028076,
|
|
"learning_rate": 9.971686536702823e-06,
|
|
"loss": 1.1869,
|
|
"step": 782
|
|
},
|
|
{
|
|
"epoch": 0.4117801735471996,
|
|
"grad_norm": 2.9988865852355957,
|
|
"learning_rate": 9.97150135529102e-06,
|
|
"loss": 1.2341,
|
|
"step": 783
|
|
},
|
|
{
|
|
"epoch": 0.4123060741519853,
|
|
"grad_norm": 3.0471668243408203,
|
|
"learning_rate": 9.971315572001871e-06,
|
|
"loss": 1.2429,
|
|
"step": 784
|
|
},
|
|
{
|
|
"epoch": 0.41283197475677097,
|
|
"grad_norm": 3.1780314445495605,
|
|
"learning_rate": 9.971129186857868e-06,
|
|
"loss": 1.207,
|
|
"step": 785
|
|
},
|
|
{
|
|
"epoch": 0.41335787536155666,
|
|
"grad_norm": 2.884211778640747,
|
|
"learning_rate": 9.970942199881577e-06,
|
|
"loss": 1.2734,
|
|
"step": 786
|
|
},
|
|
{
|
|
"epoch": 0.41388377596634235,
|
|
"grad_norm": 3.4129996299743652,
|
|
"learning_rate": 9.970754611095635e-06,
|
|
"loss": 1.3164,
|
|
"step": 787
|
|
},
|
|
{
|
|
"epoch": 0.41440967657112804,
|
|
"grad_norm": 2.9307498931884766,
|
|
"learning_rate": 9.970566420522752e-06,
|
|
"loss": 1.2373,
|
|
"step": 788
|
|
},
|
|
{
|
|
"epoch": 0.41493557717591373,
|
|
"grad_norm": 3.0829718112945557,
|
|
"learning_rate": 9.97037762818571e-06,
|
|
"loss": 1.3499,
|
|
"step": 789
|
|
},
|
|
{
|
|
"epoch": 0.4154614777806995,
|
|
"grad_norm": 2.9609909057617188,
|
|
"learning_rate": 9.970188234107368e-06,
|
|
"loss": 1.294,
|
|
"step": 790
|
|
},
|
|
{
|
|
"epoch": 0.41598737838548516,
|
|
"grad_norm": 2.9701831340789795,
|
|
"learning_rate": 9.969998238310654e-06,
|
|
"loss": 1.1427,
|
|
"step": 791
|
|
},
|
|
{
|
|
"epoch": 0.41651327899027085,
|
|
"grad_norm": 3.0786240100860596,
|
|
"learning_rate": 9.969807640818568e-06,
|
|
"loss": 1.27,
|
|
"step": 792
|
|
},
|
|
{
|
|
"epoch": 0.41703917959505654,
|
|
"grad_norm": 2.758620023727417,
|
|
"learning_rate": 9.969616441654189e-06,
|
|
"loss": 1.1461,
|
|
"step": 793
|
|
},
|
|
{
|
|
"epoch": 0.41756508019984223,
|
|
"grad_norm": 2.916839599609375,
|
|
"learning_rate": 9.969424640840661e-06,
|
|
"loss": 1.0873,
|
|
"step": 794
|
|
},
|
|
{
|
|
"epoch": 0.4180909808046279,
|
|
"grad_norm": 3.1802310943603516,
|
|
"learning_rate": 9.969232238401206e-06,
|
|
"loss": 1.3314,
|
|
"step": 795
|
|
},
|
|
{
|
|
"epoch": 0.4186168814094136,
|
|
"grad_norm": 3.031858444213867,
|
|
"learning_rate": 9.969039234359116e-06,
|
|
"loss": 1.2407,
|
|
"step": 796
|
|
},
|
|
{
|
|
"epoch": 0.4191427820141993,
|
|
"grad_norm": 3.199017286300659,
|
|
"learning_rate": 9.96884562873776e-06,
|
|
"loss": 1.2885,
|
|
"step": 797
|
|
},
|
|
{
|
|
"epoch": 0.419668682618985,
|
|
"grad_norm": 3.1345367431640625,
|
|
"learning_rate": 9.968651421560576e-06,
|
|
"loss": 1.2587,
|
|
"step": 798
|
|
},
|
|
{
|
|
"epoch": 0.42019458322377073,
|
|
"grad_norm": 2.9804677963256836,
|
|
"learning_rate": 9.968456612851073e-06,
|
|
"loss": 1.2336,
|
|
"step": 799
|
|
},
|
|
{
|
|
"epoch": 0.4207204838285564,
|
|
"grad_norm": 2.9228107929229736,
|
|
"learning_rate": 9.968261202632837e-06,
|
|
"loss": 1.161,
|
|
"step": 800
|
|
},
|
|
{
|
|
"epoch": 0.4212463844333421,
|
|
"grad_norm": 3.026862382888794,
|
|
"learning_rate": 9.968065190929527e-06,
|
|
"loss": 1.296,
|
|
"step": 801
|
|
},
|
|
{
|
|
"epoch": 0.4217722850381278,
|
|
"grad_norm": 3.0640735626220703,
|
|
"learning_rate": 9.967868577764872e-06,
|
|
"loss": 1.1995,
|
|
"step": 802
|
|
},
|
|
{
|
|
"epoch": 0.4222981856429135,
|
|
"grad_norm": 3.174865484237671,
|
|
"learning_rate": 9.967671363162674e-06,
|
|
"loss": 1.3055,
|
|
"step": 803
|
|
},
|
|
{
|
|
"epoch": 0.4228240862476992,
|
|
"grad_norm": 3.080817699432373,
|
|
"learning_rate": 9.967473547146813e-06,
|
|
"loss": 1.1705,
|
|
"step": 804
|
|
},
|
|
{
|
|
"epoch": 0.42334998685248487,
|
|
"grad_norm": 3.163001298904419,
|
|
"learning_rate": 9.967275129741231e-06,
|
|
"loss": 1.2241,
|
|
"step": 805
|
|
},
|
|
{
|
|
"epoch": 0.42387588745727056,
|
|
"grad_norm": 3.1753580570220947,
|
|
"learning_rate": 9.967076110969958e-06,
|
|
"loss": 1.2422,
|
|
"step": 806
|
|
},
|
|
{
|
|
"epoch": 0.42440178806205625,
|
|
"grad_norm": 2.8388352394104004,
|
|
"learning_rate": 9.96687649085708e-06,
|
|
"loss": 1.1937,
|
|
"step": 807
|
|
},
|
|
{
|
|
"epoch": 0.42492768866684194,
|
|
"grad_norm": 3.1188883781433105,
|
|
"learning_rate": 9.966676269426769e-06,
|
|
"loss": 1.2298,
|
|
"step": 808
|
|
},
|
|
{
|
|
"epoch": 0.4254535892716277,
|
|
"grad_norm": 2.7816848754882812,
|
|
"learning_rate": 9.966475446703264e-06,
|
|
"loss": 1.127,
|
|
"step": 809
|
|
},
|
|
{
|
|
"epoch": 0.4259794898764134,
|
|
"grad_norm": 2.9201889038085938,
|
|
"learning_rate": 9.966274022710878e-06,
|
|
"loss": 1.1823,
|
|
"step": 810
|
|
},
|
|
{
|
|
"epoch": 0.42650539048119906,
|
|
"grad_norm": 2.8770434856414795,
|
|
"learning_rate": 9.966071997473995e-06,
|
|
"loss": 1.1446,
|
|
"step": 811
|
|
},
|
|
{
|
|
"epoch": 0.42703129108598475,
|
|
"grad_norm": 3.0478477478027344,
|
|
"learning_rate": 9.965869371017074e-06,
|
|
"loss": 1.2428,
|
|
"step": 812
|
|
},
|
|
{
|
|
"epoch": 0.42755719169077044,
|
|
"grad_norm": 3.0313892364501953,
|
|
"learning_rate": 9.965666143364647e-06,
|
|
"loss": 1.1451,
|
|
"step": 813
|
|
},
|
|
{
|
|
"epoch": 0.42808309229555613,
|
|
"grad_norm": 3.265969753265381,
|
|
"learning_rate": 9.965462314541317e-06,
|
|
"loss": 1.2996,
|
|
"step": 814
|
|
},
|
|
{
|
|
"epoch": 0.4286089929003418,
|
|
"grad_norm": 3.115689754486084,
|
|
"learning_rate": 9.96525788457176e-06,
|
|
"loss": 1.1404,
|
|
"step": 815
|
|
},
|
|
{
|
|
"epoch": 0.4291348935051275,
|
|
"grad_norm": 2.886488199234009,
|
|
"learning_rate": 9.965052853480727e-06,
|
|
"loss": 1.1549,
|
|
"step": 816
|
|
},
|
|
{
|
|
"epoch": 0.4296607941099132,
|
|
"grad_norm": 3.151937484741211,
|
|
"learning_rate": 9.96484722129304e-06,
|
|
"loss": 1.3159,
|
|
"step": 817
|
|
},
|
|
{
|
|
"epoch": 0.43018669471469895,
|
|
"grad_norm": 2.8313939571380615,
|
|
"learning_rate": 9.964640988033591e-06,
|
|
"loss": 1.249,
|
|
"step": 818
|
|
},
|
|
{
|
|
"epoch": 0.43071259531948464,
|
|
"grad_norm": 3.2079062461853027,
|
|
"learning_rate": 9.964434153727351e-06,
|
|
"loss": 1.3028,
|
|
"step": 819
|
|
},
|
|
{
|
|
"epoch": 0.4312384959242703,
|
|
"grad_norm": 2.992985248565674,
|
|
"learning_rate": 9.96422671839936e-06,
|
|
"loss": 1.2351,
|
|
"step": 820
|
|
},
|
|
{
|
|
"epoch": 0.431764396529056,
|
|
"grad_norm": 2.8675076961517334,
|
|
"learning_rate": 9.964018682074732e-06,
|
|
"loss": 1.3008,
|
|
"step": 821
|
|
},
|
|
{
|
|
"epoch": 0.4322902971338417,
|
|
"grad_norm": 3.35284161567688,
|
|
"learning_rate": 9.963810044778651e-06,
|
|
"loss": 1.263,
|
|
"step": 822
|
|
},
|
|
{
|
|
"epoch": 0.4328161977386274,
|
|
"grad_norm": 4.117649555206299,
|
|
"learning_rate": 9.963600806536377e-06,
|
|
"loss": 1.2166,
|
|
"step": 823
|
|
},
|
|
{
|
|
"epoch": 0.4333420983434131,
|
|
"grad_norm": 3.150189161300659,
|
|
"learning_rate": 9.96339096737324e-06,
|
|
"loss": 1.2889,
|
|
"step": 824
|
|
},
|
|
{
|
|
"epoch": 0.4338679989481988,
|
|
"grad_norm": 2.931222677230835,
|
|
"learning_rate": 9.963180527314647e-06,
|
|
"loss": 1.0817,
|
|
"step": 825
|
|
},
|
|
{
|
|
"epoch": 0.43439389955298446,
|
|
"grad_norm": 3.080522060394287,
|
|
"learning_rate": 9.962969486386072e-06,
|
|
"loss": 1.2166,
|
|
"step": 826
|
|
},
|
|
{
|
|
"epoch": 0.4349198001577702,
|
|
"grad_norm": 3.139298439025879,
|
|
"learning_rate": 9.962757844613067e-06,
|
|
"loss": 1.3257,
|
|
"step": 827
|
|
},
|
|
{
|
|
"epoch": 0.4354457007625559,
|
|
"grad_norm": 2.840996265411377,
|
|
"learning_rate": 9.962545602021254e-06,
|
|
"loss": 1.3263,
|
|
"step": 828
|
|
},
|
|
{
|
|
"epoch": 0.4359716013673416,
|
|
"grad_norm": 3.0878891944885254,
|
|
"learning_rate": 9.962332758636327e-06,
|
|
"loss": 1.2426,
|
|
"step": 829
|
|
},
|
|
{
|
|
"epoch": 0.4364975019721273,
|
|
"grad_norm": 3.1815011501312256,
|
|
"learning_rate": 9.962119314484056e-06,
|
|
"loss": 1.3309,
|
|
"step": 830
|
|
},
|
|
{
|
|
"epoch": 0.43702340257691297,
|
|
"grad_norm": 2.936521053314209,
|
|
"learning_rate": 9.961905269590281e-06,
|
|
"loss": 1.2524,
|
|
"step": 831
|
|
},
|
|
{
|
|
"epoch": 0.43754930318169866,
|
|
"grad_norm": 2.8644607067108154,
|
|
"learning_rate": 9.961690623980916e-06,
|
|
"loss": 1.1897,
|
|
"step": 832
|
|
},
|
|
{
|
|
"epoch": 0.43807520378648435,
|
|
"grad_norm": 2.9414706230163574,
|
|
"learning_rate": 9.961475377681945e-06,
|
|
"loss": 1.256,
|
|
"step": 833
|
|
},
|
|
{
|
|
"epoch": 0.43860110439127004,
|
|
"grad_norm": 2.9504730701446533,
|
|
"learning_rate": 9.961259530719427e-06,
|
|
"loss": 1.1907,
|
|
"step": 834
|
|
},
|
|
{
|
|
"epoch": 0.4391270049960557,
|
|
"grad_norm": 2.7885990142822266,
|
|
"learning_rate": 9.961043083119497e-06,
|
|
"loss": 1.1486,
|
|
"step": 835
|
|
},
|
|
{
|
|
"epoch": 0.4396529056008414,
|
|
"grad_norm": 3.248070478439331,
|
|
"learning_rate": 9.960826034908358e-06,
|
|
"loss": 1.2224,
|
|
"step": 836
|
|
},
|
|
{
|
|
"epoch": 0.44017880620562716,
|
|
"grad_norm": 2.981091022491455,
|
|
"learning_rate": 9.960608386112284e-06,
|
|
"loss": 1.2389,
|
|
"step": 837
|
|
},
|
|
{
|
|
"epoch": 0.44070470681041285,
|
|
"grad_norm": 2.9134414196014404,
|
|
"learning_rate": 9.960390136757628e-06,
|
|
"loss": 1.2219,
|
|
"step": 838
|
|
},
|
|
{
|
|
"epoch": 0.44123060741519854,
|
|
"grad_norm": 3.0330445766448975,
|
|
"learning_rate": 9.960171286870812e-06,
|
|
"loss": 1.3405,
|
|
"step": 839
|
|
},
|
|
{
|
|
"epoch": 0.44175650801998423,
|
|
"grad_norm": 2.932924747467041,
|
|
"learning_rate": 9.95995183647833e-06,
|
|
"loss": 1.2009,
|
|
"step": 840
|
|
},
|
|
{
|
|
"epoch": 0.4422824086247699,
|
|
"grad_norm": 3.1174912452697754,
|
|
"learning_rate": 9.959731785606752e-06,
|
|
"loss": 1.2324,
|
|
"step": 841
|
|
},
|
|
{
|
|
"epoch": 0.4428083092295556,
|
|
"grad_norm": 2.873262405395508,
|
|
"learning_rate": 9.959511134282715e-06,
|
|
"loss": 1.144,
|
|
"step": 842
|
|
},
|
|
{
|
|
"epoch": 0.4433342098343413,
|
|
"grad_norm": 3.006535053253174,
|
|
"learning_rate": 9.959289882532936e-06,
|
|
"loss": 1.2353,
|
|
"step": 843
|
|
},
|
|
{
|
|
"epoch": 0.443860110439127,
|
|
"grad_norm": 2.93619441986084,
|
|
"learning_rate": 9.9590680303842e-06,
|
|
"loss": 1.1873,
|
|
"step": 844
|
|
},
|
|
{
|
|
"epoch": 0.4443860110439127,
|
|
"grad_norm": 3.1838221549987793,
|
|
"learning_rate": 9.958845577863363e-06,
|
|
"loss": 1.2236,
|
|
"step": 845
|
|
},
|
|
{
|
|
"epoch": 0.4449119116486984,
|
|
"grad_norm": 2.975998878479004,
|
|
"learning_rate": 9.95862252499736e-06,
|
|
"loss": 1.2451,
|
|
"step": 846
|
|
},
|
|
{
|
|
"epoch": 0.4454378122534841,
|
|
"grad_norm": 2.9895670413970947,
|
|
"learning_rate": 9.958398871813193e-06,
|
|
"loss": 1.1731,
|
|
"step": 847
|
|
},
|
|
{
|
|
"epoch": 0.4459637128582698,
|
|
"grad_norm": 3.083936929702759,
|
|
"learning_rate": 9.958174618337939e-06,
|
|
"loss": 1.2755,
|
|
"step": 848
|
|
},
|
|
{
|
|
"epoch": 0.4464896134630555,
|
|
"grad_norm": 3.013084650039673,
|
|
"learning_rate": 9.957949764598748e-06,
|
|
"loss": 1.2174,
|
|
"step": 849
|
|
},
|
|
{
|
|
"epoch": 0.4470155140678412,
|
|
"grad_norm": 2.9400880336761475,
|
|
"learning_rate": 9.95772431062284e-06,
|
|
"loss": 1.1937,
|
|
"step": 850
|
|
},
|
|
{
|
|
"epoch": 0.44754141467262687,
|
|
"grad_norm": 2.8958284854888916,
|
|
"learning_rate": 9.957498256437512e-06,
|
|
"loss": 1.1335,
|
|
"step": 851
|
|
},
|
|
{
|
|
"epoch": 0.44806731527741256,
|
|
"grad_norm": 3.0661416053771973,
|
|
"learning_rate": 9.95727160207013e-06,
|
|
"loss": 1.3293,
|
|
"step": 852
|
|
},
|
|
{
|
|
"epoch": 0.44859321588219825,
|
|
"grad_norm": 3.210169553756714,
|
|
"learning_rate": 9.957044347548137e-06,
|
|
"loss": 1.2766,
|
|
"step": 853
|
|
},
|
|
{
|
|
"epoch": 0.44911911648698394,
|
|
"grad_norm": 3.3500049114227295,
|
|
"learning_rate": 9.95681649289904e-06,
|
|
"loss": 1.35,
|
|
"step": 854
|
|
},
|
|
{
|
|
"epoch": 0.4496450170917697,
|
|
"grad_norm": 2.807283878326416,
|
|
"learning_rate": 9.95658803815043e-06,
|
|
"loss": 1.2103,
|
|
"step": 855
|
|
},
|
|
{
|
|
"epoch": 0.4501709176965554,
|
|
"grad_norm": 3.1486129760742188,
|
|
"learning_rate": 9.956358983329961e-06,
|
|
"loss": 1.2392,
|
|
"step": 856
|
|
},
|
|
{
|
|
"epoch": 0.45069681830134106,
|
|
"grad_norm": 3.1016509532928467,
|
|
"learning_rate": 9.956129328465364e-06,
|
|
"loss": 1.2919,
|
|
"step": 857
|
|
},
|
|
{
|
|
"epoch": 0.45122271890612675,
|
|
"grad_norm": 2.969515800476074,
|
|
"learning_rate": 9.955899073584447e-06,
|
|
"loss": 1.1831,
|
|
"step": 858
|
|
},
|
|
{
|
|
"epoch": 0.45174861951091244,
|
|
"grad_norm": 2.671034336090088,
|
|
"learning_rate": 9.95566821871508e-06,
|
|
"loss": 1.1031,
|
|
"step": 859
|
|
},
|
|
{
|
|
"epoch": 0.45227452011569813,
|
|
"grad_norm": 2.760820150375366,
|
|
"learning_rate": 9.955436763885216e-06,
|
|
"loss": 1.1702,
|
|
"step": 860
|
|
},
|
|
{
|
|
"epoch": 0.4528004207204838,
|
|
"grad_norm": 3.2095282077789307,
|
|
"learning_rate": 9.955204709122871e-06,
|
|
"loss": 1.2493,
|
|
"step": 861
|
|
},
|
|
{
|
|
"epoch": 0.4533263213252695,
|
|
"grad_norm": 2.858408212661743,
|
|
"learning_rate": 9.954972054456145e-06,
|
|
"loss": 1.2333,
|
|
"step": 862
|
|
},
|
|
{
|
|
"epoch": 0.4538522219300552,
|
|
"grad_norm": 3.050116539001465,
|
|
"learning_rate": 9.9547387999132e-06,
|
|
"loss": 1.2789,
|
|
"step": 863
|
|
},
|
|
{
|
|
"epoch": 0.4543781225348409,
|
|
"grad_norm": 2.9362905025482178,
|
|
"learning_rate": 9.954504945522276e-06,
|
|
"loss": 1.2777,
|
|
"step": 864
|
|
},
|
|
{
|
|
"epoch": 0.45490402313962663,
|
|
"grad_norm": 3.0969936847686768,
|
|
"learning_rate": 9.954270491311686e-06,
|
|
"loss": 1.2488,
|
|
"step": 865
|
|
},
|
|
{
|
|
"epoch": 0.4554299237444123,
|
|
"grad_norm": 3.110858917236328,
|
|
"learning_rate": 9.954035437309813e-06,
|
|
"loss": 1.2194,
|
|
"step": 866
|
|
},
|
|
{
|
|
"epoch": 0.455955824349198,
|
|
"grad_norm": 3.125241756439209,
|
|
"learning_rate": 9.953799783545116e-06,
|
|
"loss": 1.1444,
|
|
"step": 867
|
|
},
|
|
{
|
|
"epoch": 0.4564817249539837,
|
|
"grad_norm": 2.9375815391540527,
|
|
"learning_rate": 9.953563530046121e-06,
|
|
"loss": 1.2435,
|
|
"step": 868
|
|
},
|
|
{
|
|
"epoch": 0.4570076255587694,
|
|
"grad_norm": 3.3166725635528564,
|
|
"learning_rate": 9.953326676841434e-06,
|
|
"loss": 1.2668,
|
|
"step": 869
|
|
},
|
|
{
|
|
"epoch": 0.4575335261635551,
|
|
"grad_norm": 2.905985116958618,
|
|
"learning_rate": 9.953089223959727e-06,
|
|
"loss": 1.2538,
|
|
"step": 870
|
|
},
|
|
{
|
|
"epoch": 0.45805942676834077,
|
|
"grad_norm": 3.008389711380005,
|
|
"learning_rate": 9.95285117142975e-06,
|
|
"loss": 1.2578,
|
|
"step": 871
|
|
},
|
|
{
|
|
"epoch": 0.45858532737312646,
|
|
"grad_norm": 2.9201221466064453,
|
|
"learning_rate": 9.952612519280318e-06,
|
|
"loss": 1.3282,
|
|
"step": 872
|
|
},
|
|
{
|
|
"epoch": 0.45911122797791215,
|
|
"grad_norm": 3.358569860458374,
|
|
"learning_rate": 9.952373267540328e-06,
|
|
"loss": 1.2764,
|
|
"step": 873
|
|
},
|
|
{
|
|
"epoch": 0.4596371285826979,
|
|
"grad_norm": 3.0456926822662354,
|
|
"learning_rate": 9.952133416238746e-06,
|
|
"loss": 1.3115,
|
|
"step": 874
|
|
},
|
|
{
|
|
"epoch": 0.4601630291874836,
|
|
"grad_norm": 2.8996191024780273,
|
|
"learning_rate": 9.951892965404604e-06,
|
|
"loss": 1.2433,
|
|
"step": 875
|
|
},
|
|
{
|
|
"epoch": 0.4606889297922693,
|
|
"grad_norm": 3.1872613430023193,
|
|
"learning_rate": 9.951651915067019e-06,
|
|
"loss": 1.2303,
|
|
"step": 876
|
|
},
|
|
{
|
|
"epoch": 0.46121483039705496,
|
|
"grad_norm": 2.7677419185638428,
|
|
"learning_rate": 9.95141026525517e-06,
|
|
"loss": 1.238,
|
|
"step": 877
|
|
},
|
|
{
|
|
"epoch": 0.46174073100184065,
|
|
"grad_norm": 3.096921443939209,
|
|
"learning_rate": 9.951168015998312e-06,
|
|
"loss": 1.2062,
|
|
"step": 878
|
|
},
|
|
{
|
|
"epoch": 0.46226663160662634,
|
|
"grad_norm": 2.939069986343384,
|
|
"learning_rate": 9.950925167325775e-06,
|
|
"loss": 1.3059,
|
|
"step": 879
|
|
},
|
|
{
|
|
"epoch": 0.46279253221141203,
|
|
"grad_norm": 3.0082857608795166,
|
|
"learning_rate": 9.95068171926696e-06,
|
|
"loss": 1.2663,
|
|
"step": 880
|
|
},
|
|
{
|
|
"epoch": 0.4633184328161977,
|
|
"grad_norm": 2.807844400405884,
|
|
"learning_rate": 9.950437671851338e-06,
|
|
"loss": 1.2506,
|
|
"step": 881
|
|
},
|
|
{
|
|
"epoch": 0.4638443334209834,
|
|
"grad_norm": 2.9095351696014404,
|
|
"learning_rate": 9.950193025108456e-06,
|
|
"loss": 1.2238,
|
|
"step": 882
|
|
},
|
|
{
|
|
"epoch": 0.4643702340257691,
|
|
"grad_norm": 3.0612833499908447,
|
|
"learning_rate": 9.949947779067933e-06,
|
|
"loss": 1.1836,
|
|
"step": 883
|
|
},
|
|
{
|
|
"epoch": 0.46489613463055485,
|
|
"grad_norm": 3.0842068195343018,
|
|
"learning_rate": 9.94970193375946e-06,
|
|
"loss": 1.2899,
|
|
"step": 884
|
|
},
|
|
{
|
|
"epoch": 0.46542203523534054,
|
|
"grad_norm": 3.1199519634246826,
|
|
"learning_rate": 9.949455489212797e-06,
|
|
"loss": 1.3016,
|
|
"step": 885
|
|
},
|
|
{
|
|
"epoch": 0.4659479358401262,
|
|
"grad_norm": 2.807921886444092,
|
|
"learning_rate": 9.949208445457781e-06,
|
|
"loss": 1.1841,
|
|
"step": 886
|
|
},
|
|
{
|
|
"epoch": 0.4664738364449119,
|
|
"grad_norm": 2.8469009399414062,
|
|
"learning_rate": 9.948960802524323e-06,
|
|
"loss": 1.1406,
|
|
"step": 887
|
|
},
|
|
{
|
|
"epoch": 0.4669997370496976,
|
|
"grad_norm": 3.2436530590057373,
|
|
"learning_rate": 9.948712560442404e-06,
|
|
"loss": 1.2746,
|
|
"step": 888
|
|
},
|
|
{
|
|
"epoch": 0.4675256376544833,
|
|
"grad_norm": 3.005619764328003,
|
|
"learning_rate": 9.948463719242076e-06,
|
|
"loss": 1.3046,
|
|
"step": 889
|
|
},
|
|
{
|
|
"epoch": 0.468051538259269,
|
|
"grad_norm": 3.224093198776245,
|
|
"learning_rate": 9.948214278953465e-06,
|
|
"loss": 1.2271,
|
|
"step": 890
|
|
},
|
|
{
|
|
"epoch": 0.4685774388640547,
|
|
"grad_norm": 3.0279593467712402,
|
|
"learning_rate": 9.94796423960677e-06,
|
|
"loss": 1.2389,
|
|
"step": 891
|
|
},
|
|
{
|
|
"epoch": 0.46910333946884036,
|
|
"grad_norm": 2.887617349624634,
|
|
"learning_rate": 9.94771360123226e-06,
|
|
"loss": 1.1549,
|
|
"step": 892
|
|
},
|
|
{
|
|
"epoch": 0.4696292400736261,
|
|
"grad_norm": 3.0618772506713867,
|
|
"learning_rate": 9.947462363860284e-06,
|
|
"loss": 1.1072,
|
|
"step": 893
|
|
},
|
|
{
|
|
"epoch": 0.4701551406784118,
|
|
"grad_norm": 2.83664608001709,
|
|
"learning_rate": 9.947210527521253e-06,
|
|
"loss": 1.1414,
|
|
"step": 894
|
|
},
|
|
{
|
|
"epoch": 0.4706810412831975,
|
|
"grad_norm": 3.0461323261260986,
|
|
"learning_rate": 9.946958092245659e-06,
|
|
"loss": 1.2174,
|
|
"step": 895
|
|
},
|
|
{
|
|
"epoch": 0.4712069418879832,
|
|
"grad_norm": 3.0457630157470703,
|
|
"learning_rate": 9.94670505806406e-06,
|
|
"loss": 1.2898,
|
|
"step": 896
|
|
},
|
|
{
|
|
"epoch": 0.47173284249276887,
|
|
"grad_norm": 2.8451483249664307,
|
|
"learning_rate": 9.946451425007094e-06,
|
|
"loss": 1.2437,
|
|
"step": 897
|
|
},
|
|
{
|
|
"epoch": 0.47225874309755456,
|
|
"grad_norm": 2.9501264095306396,
|
|
"learning_rate": 9.946197193105463e-06,
|
|
"loss": 1.2158,
|
|
"step": 898
|
|
},
|
|
{
|
|
"epoch": 0.47278464370234025,
|
|
"grad_norm": 2.8322365283966064,
|
|
"learning_rate": 9.945942362389949e-06,
|
|
"loss": 1.2123,
|
|
"step": 899
|
|
},
|
|
{
|
|
"epoch": 0.47331054430712594,
|
|
"grad_norm": 2.9531044960021973,
|
|
"learning_rate": 9.9456869328914e-06,
|
|
"loss": 1.2506,
|
|
"step": 900
|
|
},
|
|
{
|
|
"epoch": 0.4738364449119116,
|
|
"grad_norm": 2.968024253845215,
|
|
"learning_rate": 9.945430904640742e-06,
|
|
"loss": 1.2246,
|
|
"step": 901
|
|
},
|
|
{
|
|
"epoch": 0.47436234551669737,
|
|
"grad_norm": 2.8673524856567383,
|
|
"learning_rate": 9.945174277668972e-06,
|
|
"loss": 1.3838,
|
|
"step": 902
|
|
},
|
|
{
|
|
"epoch": 0.47488824612148306,
|
|
"grad_norm": 2.744797706604004,
|
|
"learning_rate": 9.944917052007157e-06,
|
|
"loss": 1.233,
|
|
"step": 903
|
|
},
|
|
{
|
|
"epoch": 0.47541414672626875,
|
|
"grad_norm": 2.8587570190429688,
|
|
"learning_rate": 9.944659227686437e-06,
|
|
"loss": 1.2703,
|
|
"step": 904
|
|
},
|
|
{
|
|
"epoch": 0.47594004733105444,
|
|
"grad_norm": 2.8735241889953613,
|
|
"learning_rate": 9.944400804738028e-06,
|
|
"loss": 1.2435,
|
|
"step": 905
|
|
},
|
|
{
|
|
"epoch": 0.47646594793584013,
|
|
"grad_norm": 2.6896915435791016,
|
|
"learning_rate": 9.944141783193215e-06,
|
|
"loss": 1.2544,
|
|
"step": 906
|
|
},
|
|
{
|
|
"epoch": 0.4769918485406258,
|
|
"grad_norm": 2.9248757362365723,
|
|
"learning_rate": 9.943882163083355e-06,
|
|
"loss": 1.3248,
|
|
"step": 907
|
|
},
|
|
{
|
|
"epoch": 0.4775177491454115,
|
|
"grad_norm": 2.87561297416687,
|
|
"learning_rate": 9.943621944439884e-06,
|
|
"loss": 1.4035,
|
|
"step": 908
|
|
},
|
|
{
|
|
"epoch": 0.4780436497501972,
|
|
"grad_norm": 2.8148603439331055,
|
|
"learning_rate": 9.9433611272943e-06,
|
|
"loss": 1.1682,
|
|
"step": 909
|
|
},
|
|
{
|
|
"epoch": 0.4785695503549829,
|
|
"grad_norm": 3.271822214126587,
|
|
"learning_rate": 9.943099711678184e-06,
|
|
"loss": 1.2668,
|
|
"step": 910
|
|
},
|
|
{
|
|
"epoch": 0.4790954509597686,
|
|
"grad_norm": 2.841844320297241,
|
|
"learning_rate": 9.94283769762318e-06,
|
|
"loss": 1.2493,
|
|
"step": 911
|
|
},
|
|
{
|
|
"epoch": 0.4796213515645543,
|
|
"grad_norm": 2.9426422119140625,
|
|
"learning_rate": 9.94257508516101e-06,
|
|
"loss": 1.1437,
|
|
"step": 912
|
|
},
|
|
{
|
|
"epoch": 0.48014725216934,
|
|
"grad_norm": 3.0486366748809814,
|
|
"learning_rate": 9.942311874323468e-06,
|
|
"loss": 1.2868,
|
|
"step": 913
|
|
},
|
|
{
|
|
"epoch": 0.4806731527741257,
|
|
"grad_norm": 2.993549346923828,
|
|
"learning_rate": 9.94204806514242e-06,
|
|
"loss": 1.1966,
|
|
"step": 914
|
|
},
|
|
{
|
|
"epoch": 0.4811990533789114,
|
|
"grad_norm": 3.0795860290527344,
|
|
"learning_rate": 9.941783657649803e-06,
|
|
"loss": 1.309,
|
|
"step": 915
|
|
},
|
|
{
|
|
"epoch": 0.4817249539836971,
|
|
"grad_norm": 3.236253261566162,
|
|
"learning_rate": 9.941518651877629e-06,
|
|
"loss": 1.3184,
|
|
"step": 916
|
|
},
|
|
{
|
|
"epoch": 0.48225085458848277,
|
|
"grad_norm": 3.206937313079834,
|
|
"learning_rate": 9.94125304785798e-06,
|
|
"loss": 1.2586,
|
|
"step": 917
|
|
},
|
|
{
|
|
"epoch": 0.48277675519326846,
|
|
"grad_norm": 3.1810033321380615,
|
|
"learning_rate": 9.940986845623014e-06,
|
|
"loss": 1.3264,
|
|
"step": 918
|
|
},
|
|
{
|
|
"epoch": 0.48330265579805415,
|
|
"grad_norm": 3.0482616424560547,
|
|
"learning_rate": 9.940720045204955e-06,
|
|
"loss": 1.2768,
|
|
"step": 919
|
|
},
|
|
{
|
|
"epoch": 0.48382855640283984,
|
|
"grad_norm": 3.1699299812316895,
|
|
"learning_rate": 9.940452646636105e-06,
|
|
"loss": 1.3509,
|
|
"step": 920
|
|
},
|
|
{
|
|
"epoch": 0.4843544570076256,
|
|
"grad_norm": 2.8135862350463867,
|
|
"learning_rate": 9.940184649948838e-06,
|
|
"loss": 1.281,
|
|
"step": 921
|
|
},
|
|
{
|
|
"epoch": 0.4848803576124113,
|
|
"grad_norm": 2.8988375663757324,
|
|
"learning_rate": 9.939916055175597e-06,
|
|
"loss": 1.1683,
|
|
"step": 922
|
|
},
|
|
{
|
|
"epoch": 0.48540625821719696,
|
|
"grad_norm": 3.0824379920959473,
|
|
"learning_rate": 9.939646862348902e-06,
|
|
"loss": 1.3405,
|
|
"step": 923
|
|
},
|
|
{
|
|
"epoch": 0.48593215882198265,
|
|
"grad_norm": 2.9967381954193115,
|
|
"learning_rate": 9.939377071501341e-06,
|
|
"loss": 1.2182,
|
|
"step": 924
|
|
},
|
|
{
|
|
"epoch": 0.48645805942676834,
|
|
"grad_norm": 3.0528340339660645,
|
|
"learning_rate": 9.939106682665576e-06,
|
|
"loss": 1.3204,
|
|
"step": 925
|
|
},
|
|
{
|
|
"epoch": 0.48698396003155403,
|
|
"grad_norm": 2.9739091396331787,
|
|
"learning_rate": 9.938835695874346e-06,
|
|
"loss": 1.1754,
|
|
"step": 926
|
|
},
|
|
{
|
|
"epoch": 0.4875098606363397,
|
|
"grad_norm": 3.0073466300964355,
|
|
"learning_rate": 9.938564111160452e-06,
|
|
"loss": 1.2551,
|
|
"step": 927
|
|
},
|
|
{
|
|
"epoch": 0.4880357612411254,
|
|
"grad_norm": 2.7729580402374268,
|
|
"learning_rate": 9.938291928556777e-06,
|
|
"loss": 1.1424,
|
|
"step": 928
|
|
},
|
|
{
|
|
"epoch": 0.4885616618459111,
|
|
"grad_norm": 3.06596040725708,
|
|
"learning_rate": 9.938019148096273e-06,
|
|
"loss": 1.1933,
|
|
"step": 929
|
|
},
|
|
{
|
|
"epoch": 0.48908756245069684,
|
|
"grad_norm": 3.1965761184692383,
|
|
"learning_rate": 9.937745769811963e-06,
|
|
"loss": 1.3018,
|
|
"step": 930
|
|
},
|
|
{
|
|
"epoch": 0.48961346305548253,
|
|
"grad_norm": 3.0823941230773926,
|
|
"learning_rate": 9.937471793736946e-06,
|
|
"loss": 1.2486,
|
|
"step": 931
|
|
},
|
|
{
|
|
"epoch": 0.4901393636602682,
|
|
"grad_norm": 2.9435348510742188,
|
|
"learning_rate": 9.937197219904387e-06,
|
|
"loss": 1.1406,
|
|
"step": 932
|
|
},
|
|
{
|
|
"epoch": 0.4906652642650539,
|
|
"grad_norm": 2.833984375,
|
|
"learning_rate": 9.936922048347531e-06,
|
|
"loss": 1.2166,
|
|
"step": 933
|
|
},
|
|
{
|
|
"epoch": 0.4911911648698396,
|
|
"grad_norm": 2.9711062908172607,
|
|
"learning_rate": 9.93664627909969e-06,
|
|
"loss": 1.3395,
|
|
"step": 934
|
|
},
|
|
{
|
|
"epoch": 0.4917170654746253,
|
|
"grad_norm": 2.806893825531006,
|
|
"learning_rate": 9.936369912194252e-06,
|
|
"loss": 1.2512,
|
|
"step": 935
|
|
},
|
|
{
|
|
"epoch": 0.492242966079411,
|
|
"grad_norm": 2.7795886993408203,
|
|
"learning_rate": 9.936092947664676e-06,
|
|
"loss": 1.2353,
|
|
"step": 936
|
|
},
|
|
{
|
|
"epoch": 0.49276886668419667,
|
|
"grad_norm": 2.767132520675659,
|
|
"learning_rate": 9.935815385544488e-06,
|
|
"loss": 1.1851,
|
|
"step": 937
|
|
},
|
|
{
|
|
"epoch": 0.49329476728898236,
|
|
"grad_norm": 2.8482043743133545,
|
|
"learning_rate": 9.935537225867296e-06,
|
|
"loss": 1.1853,
|
|
"step": 938
|
|
},
|
|
{
|
|
"epoch": 0.49382066789376805,
|
|
"grad_norm": 2.9351413249969482,
|
|
"learning_rate": 9.935258468666772e-06,
|
|
"loss": 1.2628,
|
|
"step": 939
|
|
},
|
|
{
|
|
"epoch": 0.4943465684985538,
|
|
"grad_norm": 2.939232110977173,
|
|
"learning_rate": 9.934979113976667e-06,
|
|
"loss": 1.2575,
|
|
"step": 940
|
|
},
|
|
{
|
|
"epoch": 0.4948724691033395,
|
|
"grad_norm": 3.1346426010131836,
|
|
"learning_rate": 9.9346991618308e-06,
|
|
"loss": 1.2376,
|
|
"step": 941
|
|
},
|
|
{
|
|
"epoch": 0.4953983697081252,
|
|
"grad_norm": 2.889117479324341,
|
|
"learning_rate": 9.934418612263063e-06,
|
|
"loss": 1.2921,
|
|
"step": 942
|
|
},
|
|
{
|
|
"epoch": 0.49592427031291086,
|
|
"grad_norm": 2.8371429443359375,
|
|
"learning_rate": 9.934137465307422e-06,
|
|
"loss": 1.3728,
|
|
"step": 943
|
|
},
|
|
{
|
|
"epoch": 0.49645017091769655,
|
|
"grad_norm": 2.8687610626220703,
|
|
"learning_rate": 9.933855720997914e-06,
|
|
"loss": 1.1322,
|
|
"step": 944
|
|
},
|
|
{
|
|
"epoch": 0.49697607152248224,
|
|
"grad_norm": 3.0074832439422607,
|
|
"learning_rate": 9.933573379368648e-06,
|
|
"loss": 1.2062,
|
|
"step": 945
|
|
},
|
|
{
|
|
"epoch": 0.49750197212726793,
|
|
"grad_norm": 3.1922459602355957,
|
|
"learning_rate": 9.933290440453805e-06,
|
|
"loss": 1.2081,
|
|
"step": 946
|
|
},
|
|
{
|
|
"epoch": 0.4980278727320536,
|
|
"grad_norm": 2.95231556892395,
|
|
"learning_rate": 9.93300690428764e-06,
|
|
"loss": 1.188,
|
|
"step": 947
|
|
},
|
|
{
|
|
"epoch": 0.4985537733368393,
|
|
"grad_norm": 2.916290760040283,
|
|
"learning_rate": 9.932722770904481e-06,
|
|
"loss": 1.1826,
|
|
"step": 948
|
|
},
|
|
{
|
|
"epoch": 0.49907967394162506,
|
|
"grad_norm": 3.086010694503784,
|
|
"learning_rate": 9.932438040338725e-06,
|
|
"loss": 1.214,
|
|
"step": 949
|
|
},
|
|
{
|
|
"epoch": 0.49960557454641075,
|
|
"grad_norm": 3.0533430576324463,
|
|
"learning_rate": 9.932152712624843e-06,
|
|
"loss": 1.2604,
|
|
"step": 950
|
|
},
|
|
{
|
|
"epoch": 0.5001314751511964,
|
|
"grad_norm": 3.181468963623047,
|
|
"learning_rate": 9.931866787797378e-06,
|
|
"loss": 1.2222,
|
|
"step": 951
|
|
},
|
|
{
|
|
"epoch": 0.5006573757559821,
|
|
"grad_norm": 2.907658576965332,
|
|
"learning_rate": 9.931580265890947e-06,
|
|
"loss": 1.2326,
|
|
"step": 952
|
|
},
|
|
{
|
|
"epoch": 0.5011832763607678,
|
|
"grad_norm": 3.101773738861084,
|
|
"learning_rate": 9.931293146940237e-06,
|
|
"loss": 1.1863,
|
|
"step": 953
|
|
},
|
|
{
|
|
"epoch": 0.5017091769655535,
|
|
"grad_norm": 2.9280619621276855,
|
|
"learning_rate": 9.93100543098001e-06,
|
|
"loss": 1.2357,
|
|
"step": 954
|
|
},
|
|
{
|
|
"epoch": 0.5022350775703393,
|
|
"grad_norm": 3.179316520690918,
|
|
"learning_rate": 9.930717118045094e-06,
|
|
"loss": 1.3179,
|
|
"step": 955
|
|
},
|
|
{
|
|
"epoch": 0.5027609781751249,
|
|
"grad_norm": 2.9885926246643066,
|
|
"learning_rate": 9.930428208170399e-06,
|
|
"loss": 1.2149,
|
|
"step": 956
|
|
},
|
|
{
|
|
"epoch": 0.5032868787799106,
|
|
"grad_norm": 3.2049219608306885,
|
|
"learning_rate": 9.9301387013909e-06,
|
|
"loss": 1.2761,
|
|
"step": 957
|
|
},
|
|
{
|
|
"epoch": 0.5038127793846963,
|
|
"grad_norm": 2.97983717918396,
|
|
"learning_rate": 9.929848597741644e-06,
|
|
"loss": 1.3022,
|
|
"step": 958
|
|
},
|
|
{
|
|
"epoch": 0.504338679989482,
|
|
"grad_norm": 2.879096269607544,
|
|
"learning_rate": 9.929557897257755e-06,
|
|
"loss": 1.1945,
|
|
"step": 959
|
|
},
|
|
{
|
|
"epoch": 0.5048645805942676,
|
|
"grad_norm": 2.9068946838378906,
|
|
"learning_rate": 9.929266599974428e-06,
|
|
"loss": 1.159,
|
|
"step": 960
|
|
},
|
|
{
|
|
"epoch": 0.5053904811990534,
|
|
"grad_norm": 3.254563808441162,
|
|
"learning_rate": 9.928974705926926e-06,
|
|
"loss": 1.2491,
|
|
"step": 961
|
|
},
|
|
{
|
|
"epoch": 0.505916381803839,
|
|
"grad_norm": 2.9189701080322266,
|
|
"learning_rate": 9.92868221515059e-06,
|
|
"loss": 1.2964,
|
|
"step": 962
|
|
},
|
|
{
|
|
"epoch": 0.5064422824086248,
|
|
"grad_norm": 2.986999273300171,
|
|
"learning_rate": 9.928389127680828e-06,
|
|
"loss": 1.1935,
|
|
"step": 963
|
|
},
|
|
{
|
|
"epoch": 0.5069681830134105,
|
|
"grad_norm": 2.89886474609375,
|
|
"learning_rate": 9.928095443553123e-06,
|
|
"loss": 1.2834,
|
|
"step": 964
|
|
},
|
|
{
|
|
"epoch": 0.5074940836181961,
|
|
"grad_norm": 2.9614715576171875,
|
|
"learning_rate": 9.927801162803032e-06,
|
|
"loss": 1.2448,
|
|
"step": 965
|
|
},
|
|
{
|
|
"epoch": 0.5080199842229819,
|
|
"grad_norm": 2.8559587001800537,
|
|
"learning_rate": 9.927506285466182e-06,
|
|
"loss": 1.1104,
|
|
"step": 966
|
|
},
|
|
{
|
|
"epoch": 0.5085458848277675,
|
|
"grad_norm": 3.017094612121582,
|
|
"learning_rate": 9.927210811578272e-06,
|
|
"loss": 1.2436,
|
|
"step": 967
|
|
},
|
|
{
|
|
"epoch": 0.5090717854325533,
|
|
"grad_norm": 2.961728811264038,
|
|
"learning_rate": 9.926914741175074e-06,
|
|
"loss": 1.2477,
|
|
"step": 968
|
|
},
|
|
{
|
|
"epoch": 0.5095976860373389,
|
|
"grad_norm": 3.0190367698669434,
|
|
"learning_rate": 9.92661807429243e-06,
|
|
"loss": 1.2278,
|
|
"step": 969
|
|
},
|
|
{
|
|
"epoch": 0.5101235866421246,
|
|
"grad_norm": 2.853564977645874,
|
|
"learning_rate": 9.926320810966257e-06,
|
|
"loss": 1.2242,
|
|
"step": 970
|
|
},
|
|
{
|
|
"epoch": 0.5106494872469103,
|
|
"grad_norm": 3.081231117248535,
|
|
"learning_rate": 9.926022951232546e-06,
|
|
"loss": 1.3948,
|
|
"step": 971
|
|
},
|
|
{
|
|
"epoch": 0.511175387851696,
|
|
"grad_norm": 2.8687729835510254,
|
|
"learning_rate": 9.925724495127354e-06,
|
|
"loss": 1.128,
|
|
"step": 972
|
|
},
|
|
{
|
|
"epoch": 0.5117012884564818,
|
|
"grad_norm": 2.9628713130950928,
|
|
"learning_rate": 9.925425442686815e-06,
|
|
"loss": 1.2154,
|
|
"step": 973
|
|
},
|
|
{
|
|
"epoch": 0.5122271890612674,
|
|
"grad_norm": 2.757115125656128,
|
|
"learning_rate": 9.925125793947135e-06,
|
|
"loss": 1.2026,
|
|
"step": 974
|
|
},
|
|
{
|
|
"epoch": 0.5127530896660532,
|
|
"grad_norm": 2.830552816390991,
|
|
"learning_rate": 9.92482554894459e-06,
|
|
"loss": 1.1256,
|
|
"step": 975
|
|
},
|
|
{
|
|
"epoch": 0.5132789902708388,
|
|
"grad_norm": 2.969594717025757,
|
|
"learning_rate": 9.924524707715528e-06,
|
|
"loss": 1.2604,
|
|
"step": 976
|
|
},
|
|
{
|
|
"epoch": 0.5138048908756245,
|
|
"grad_norm": 2.9754223823547363,
|
|
"learning_rate": 9.924223270296372e-06,
|
|
"loss": 1.2597,
|
|
"step": 977
|
|
},
|
|
{
|
|
"epoch": 0.5143307914804102,
|
|
"grad_norm": 2.771519184112549,
|
|
"learning_rate": 9.923921236723617e-06,
|
|
"loss": 1.1591,
|
|
"step": 978
|
|
},
|
|
{
|
|
"epoch": 0.5148566920851959,
|
|
"grad_norm": 3.076617479324341,
|
|
"learning_rate": 9.923618607033828e-06,
|
|
"loss": 1.3434,
|
|
"step": 979
|
|
},
|
|
{
|
|
"epoch": 0.5153825926899815,
|
|
"grad_norm": 2.892606258392334,
|
|
"learning_rate": 9.923315381263643e-06,
|
|
"loss": 1.2299,
|
|
"step": 980
|
|
},
|
|
{
|
|
"epoch": 0.5159084932947673,
|
|
"grad_norm": 2.798617362976074,
|
|
"learning_rate": 9.92301155944977e-06,
|
|
"loss": 1.2357,
|
|
"step": 981
|
|
},
|
|
{
|
|
"epoch": 0.516434393899553,
|
|
"grad_norm": 2.75593900680542,
|
|
"learning_rate": 9.922707141628993e-06,
|
|
"loss": 1.2567,
|
|
"step": 982
|
|
},
|
|
{
|
|
"epoch": 0.5169602945043387,
|
|
"grad_norm": 3.0316884517669678,
|
|
"learning_rate": 9.922402127838167e-06,
|
|
"loss": 1.2629,
|
|
"step": 983
|
|
},
|
|
{
|
|
"epoch": 0.5174861951091244,
|
|
"grad_norm": 3.0661323070526123,
|
|
"learning_rate": 9.922096518114218e-06,
|
|
"loss": 1.2841,
|
|
"step": 984
|
|
},
|
|
{
|
|
"epoch": 0.51801209571391,
|
|
"grad_norm": 2.964831829071045,
|
|
"learning_rate": 9.921790312494146e-06,
|
|
"loss": 1.2118,
|
|
"step": 985
|
|
},
|
|
{
|
|
"epoch": 0.5185379963186958,
|
|
"grad_norm": 2.878042221069336,
|
|
"learning_rate": 9.92148351101502e-06,
|
|
"loss": 1.1554,
|
|
"step": 986
|
|
},
|
|
{
|
|
"epoch": 0.5190638969234814,
|
|
"grad_norm": 3.0366384983062744,
|
|
"learning_rate": 9.921176113713984e-06,
|
|
"loss": 1.2452,
|
|
"step": 987
|
|
},
|
|
{
|
|
"epoch": 0.5195897975282672,
|
|
"grad_norm": 3.0179765224456787,
|
|
"learning_rate": 9.920868120628253e-06,
|
|
"loss": 1.2441,
|
|
"step": 988
|
|
},
|
|
{
|
|
"epoch": 0.5201156981330528,
|
|
"grad_norm": 3.1251392364501953,
|
|
"learning_rate": 9.920559531795117e-06,
|
|
"loss": 1.2749,
|
|
"step": 989
|
|
},
|
|
{
|
|
"epoch": 0.5206415987378386,
|
|
"grad_norm": 2.6859333515167236,
|
|
"learning_rate": 9.920250347251931e-06,
|
|
"loss": 1.2021,
|
|
"step": 990
|
|
},
|
|
{
|
|
"epoch": 0.5211674993426243,
|
|
"grad_norm": 2.921152114868164,
|
|
"learning_rate": 9.919940567036128e-06,
|
|
"loss": 1.1141,
|
|
"step": 991
|
|
},
|
|
{
|
|
"epoch": 0.5216933999474099,
|
|
"grad_norm": 2.8585832118988037,
|
|
"learning_rate": 9.919630191185214e-06,
|
|
"loss": 1.2234,
|
|
"step": 992
|
|
},
|
|
{
|
|
"epoch": 0.5222193005521957,
|
|
"grad_norm": 2.9292125701904297,
|
|
"learning_rate": 9.919319219736762e-06,
|
|
"loss": 1.2447,
|
|
"step": 993
|
|
},
|
|
{
|
|
"epoch": 0.5227452011569813,
|
|
"grad_norm": 2.972038984298706,
|
|
"learning_rate": 9.919007652728423e-06,
|
|
"loss": 1.2909,
|
|
"step": 994
|
|
},
|
|
{
|
|
"epoch": 0.523271101761767,
|
|
"grad_norm": 2.7918806076049805,
|
|
"learning_rate": 9.918695490197913e-06,
|
|
"loss": 1.1797,
|
|
"step": 995
|
|
},
|
|
{
|
|
"epoch": 0.5237970023665527,
|
|
"grad_norm": 2.951974868774414,
|
|
"learning_rate": 9.918382732183027e-06,
|
|
"loss": 1.2365,
|
|
"step": 996
|
|
},
|
|
{
|
|
"epoch": 0.5243229029713384,
|
|
"grad_norm": 2.9659509658813477,
|
|
"learning_rate": 9.918069378721628e-06,
|
|
"loss": 1.192,
|
|
"step": 997
|
|
},
|
|
{
|
|
"epoch": 0.5248488035761241,
|
|
"grad_norm": 2.7874913215637207,
|
|
"learning_rate": 9.917755429851652e-06,
|
|
"loss": 1.2065,
|
|
"step": 998
|
|
},
|
|
{
|
|
"epoch": 0.5253747041809098,
|
|
"grad_norm": 3.3350911140441895,
|
|
"learning_rate": 9.917440885611109e-06,
|
|
"loss": 1.2008,
|
|
"step": 999
|
|
},
|
|
{
|
|
"epoch": 0.5259006047856956,
|
|
"grad_norm": 3.1266896724700928,
|
|
"learning_rate": 9.917125746038079e-06,
|
|
"loss": 1.2899,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 0.5259006047856956,
|
|
"eval_loss": 1.2596983909606934,
|
|
"eval_runtime": 8.6001,
|
|
"eval_samples_per_second": 46.511,
|
|
"eval_steps_per_second": 2.907,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 0.5264265053904812,
|
|
"grad_norm": 2.9641273021698,
|
|
"learning_rate": 9.916810011170715e-06,
|
|
"loss": 1.2221,
|
|
"step": 1001
|
|
},
|
|
{
|
|
"epoch": 0.5269524059952669,
|
|
"grad_norm": 3.1696155071258545,
|
|
"learning_rate": 9.916493681047239e-06,
|
|
"loss": 1.2751,
|
|
"step": 1002
|
|
},
|
|
{
|
|
"epoch": 0.5274783066000526,
|
|
"grad_norm": 2.80365252494812,
|
|
"learning_rate": 9.916176755705949e-06,
|
|
"loss": 1.141,
|
|
"step": 1003
|
|
},
|
|
{
|
|
"epoch": 0.5280042072048383,
|
|
"grad_norm": 2.916433811187744,
|
|
"learning_rate": 9.915859235185214e-06,
|
|
"loss": 1.1923,
|
|
"step": 1004
|
|
},
|
|
{
|
|
"epoch": 0.528530107809624,
|
|
"grad_norm": 2.982666254043579,
|
|
"learning_rate": 9.915541119523476e-06,
|
|
"loss": 1.1305,
|
|
"step": 1005
|
|
},
|
|
{
|
|
"epoch": 0.5290560084144097,
|
|
"grad_norm": 2.8673219680786133,
|
|
"learning_rate": 9.915222408759247e-06,
|
|
"loss": 1.1509,
|
|
"step": 1006
|
|
},
|
|
{
|
|
"epoch": 0.5295819090191953,
|
|
"grad_norm": 2.876580238342285,
|
|
"learning_rate": 9.91490310293111e-06,
|
|
"loss": 1.2904,
|
|
"step": 1007
|
|
},
|
|
{
|
|
"epoch": 0.5301078096239811,
|
|
"grad_norm": 2.899583101272583,
|
|
"learning_rate": 9.914583202077726e-06,
|
|
"loss": 1.2692,
|
|
"step": 1008
|
|
},
|
|
{
|
|
"epoch": 0.5306337102287667,
|
|
"grad_norm": 2.9418885707855225,
|
|
"learning_rate": 9.914262706237818e-06,
|
|
"loss": 1.3333,
|
|
"step": 1009
|
|
},
|
|
{
|
|
"epoch": 0.5311596108335525,
|
|
"grad_norm": 3.02520751953125,
|
|
"learning_rate": 9.913941615450193e-06,
|
|
"loss": 1.2756,
|
|
"step": 1010
|
|
},
|
|
{
|
|
"epoch": 0.5316855114383382,
|
|
"grad_norm": 2.913227081298828,
|
|
"learning_rate": 9.91361992975372e-06,
|
|
"loss": 1.3411,
|
|
"step": 1011
|
|
},
|
|
{
|
|
"epoch": 0.5322114120431238,
|
|
"grad_norm": 2.950021505355835,
|
|
"learning_rate": 9.913297649187347e-06,
|
|
"loss": 1.2381,
|
|
"step": 1012
|
|
},
|
|
{
|
|
"epoch": 0.5327373126479096,
|
|
"grad_norm": 2.712833881378174,
|
|
"learning_rate": 9.912974773790088e-06,
|
|
"loss": 1.1345,
|
|
"step": 1013
|
|
},
|
|
{
|
|
"epoch": 0.5332632132526952,
|
|
"grad_norm": 2.916837453842163,
|
|
"learning_rate": 9.912651303601035e-06,
|
|
"loss": 1.2185,
|
|
"step": 1014
|
|
},
|
|
{
|
|
"epoch": 0.533789113857481,
|
|
"grad_norm": 3.0292999744415283,
|
|
"learning_rate": 9.912327238659346e-06,
|
|
"loss": 1.1939,
|
|
"step": 1015
|
|
},
|
|
{
|
|
"epoch": 0.5343150144622666,
|
|
"grad_norm": 3.1598141193389893,
|
|
"learning_rate": 9.912002579004258e-06,
|
|
"loss": 1.3663,
|
|
"step": 1016
|
|
},
|
|
{
|
|
"epoch": 0.5348409150670523,
|
|
"grad_norm": 2.945422649383545,
|
|
"learning_rate": 9.911677324675073e-06,
|
|
"loss": 1.2348,
|
|
"step": 1017
|
|
},
|
|
{
|
|
"epoch": 0.535366815671838,
|
|
"grad_norm": 3.046398401260376,
|
|
"learning_rate": 9.911351475711168e-06,
|
|
"loss": 1.2072,
|
|
"step": 1018
|
|
},
|
|
{
|
|
"epoch": 0.5358927162766237,
|
|
"grad_norm": 3.056072473526001,
|
|
"learning_rate": 9.911025032151996e-06,
|
|
"loss": 1.3004,
|
|
"step": 1019
|
|
},
|
|
{
|
|
"epoch": 0.5364186168814095,
|
|
"grad_norm": 3.106613874435425,
|
|
"learning_rate": 9.910697994037072e-06,
|
|
"loss": 1.2453,
|
|
"step": 1020
|
|
},
|
|
{
|
|
"epoch": 0.5369445174861951,
|
|
"grad_norm": 3.106867790222168,
|
|
"learning_rate": 9.910370361405993e-06,
|
|
"loss": 1.3487,
|
|
"step": 1021
|
|
},
|
|
{
|
|
"epoch": 0.5374704180909808,
|
|
"grad_norm": 3.0166573524475098,
|
|
"learning_rate": 9.910042134298424e-06,
|
|
"loss": 1.3458,
|
|
"step": 1022
|
|
},
|
|
{
|
|
"epoch": 0.5379963186957665,
|
|
"grad_norm": 2.987884521484375,
|
|
"learning_rate": 9.9097133127541e-06,
|
|
"loss": 1.291,
|
|
"step": 1023
|
|
},
|
|
{
|
|
"epoch": 0.5385222193005522,
|
|
"grad_norm": 2.9285943508148193,
|
|
"learning_rate": 9.909383896812832e-06,
|
|
"loss": 1.2307,
|
|
"step": 1024
|
|
},
|
|
{
|
|
"epoch": 0.5390481199053379,
|
|
"grad_norm": 2.913969039916992,
|
|
"learning_rate": 9.9090538865145e-06,
|
|
"loss": 1.259,
|
|
"step": 1025
|
|
},
|
|
{
|
|
"epoch": 0.5395740205101236,
|
|
"grad_norm": 2.925361156463623,
|
|
"learning_rate": 9.908723281899055e-06,
|
|
"loss": 1.2351,
|
|
"step": 1026
|
|
},
|
|
{
|
|
"epoch": 0.5400999211149092,
|
|
"grad_norm": 3.014766216278076,
|
|
"learning_rate": 9.908392083006526e-06,
|
|
"loss": 1.1491,
|
|
"step": 1027
|
|
},
|
|
{
|
|
"epoch": 0.540625821719695,
|
|
"grad_norm": 2.886103391647339,
|
|
"learning_rate": 9.908060289877006e-06,
|
|
"loss": 1.2604,
|
|
"step": 1028
|
|
},
|
|
{
|
|
"epoch": 0.5411517223244807,
|
|
"grad_norm": 2.980923891067505,
|
|
"learning_rate": 9.907727902550665e-06,
|
|
"loss": 1.2124,
|
|
"step": 1029
|
|
},
|
|
{
|
|
"epoch": 0.5416776229292664,
|
|
"grad_norm": 2.9844229221343994,
|
|
"learning_rate": 9.907394921067745e-06,
|
|
"loss": 1.2689,
|
|
"step": 1030
|
|
},
|
|
{
|
|
"epoch": 0.5422035235340521,
|
|
"grad_norm": 3.162384510040283,
|
|
"learning_rate": 9.907061345468558e-06,
|
|
"loss": 1.3033,
|
|
"step": 1031
|
|
},
|
|
{
|
|
"epoch": 0.5427294241388377,
|
|
"grad_norm": 2.892317771911621,
|
|
"learning_rate": 9.906727175793487e-06,
|
|
"loss": 1.2689,
|
|
"step": 1032
|
|
},
|
|
{
|
|
"epoch": 0.5432553247436235,
|
|
"grad_norm": 2.8183653354644775,
|
|
"learning_rate": 9.906392412082988e-06,
|
|
"loss": 1.1558,
|
|
"step": 1033
|
|
},
|
|
{
|
|
"epoch": 0.5437812253484091,
|
|
"grad_norm": 2.8783605098724365,
|
|
"learning_rate": 9.906057054377591e-06,
|
|
"loss": 1.2381,
|
|
"step": 1034
|
|
},
|
|
{
|
|
"epoch": 0.5443071259531949,
|
|
"grad_norm": 3.015033721923828,
|
|
"learning_rate": 9.905721102717898e-06,
|
|
"loss": 1.2358,
|
|
"step": 1035
|
|
},
|
|
{
|
|
"epoch": 0.5448330265579805,
|
|
"grad_norm": 2.7200489044189453,
|
|
"learning_rate": 9.905384557144577e-06,
|
|
"loss": 1.2232,
|
|
"step": 1036
|
|
},
|
|
{
|
|
"epoch": 0.5453589271627662,
|
|
"grad_norm": 2.8338305950164795,
|
|
"learning_rate": 9.905047417698375e-06,
|
|
"loss": 1.2025,
|
|
"step": 1037
|
|
},
|
|
{
|
|
"epoch": 0.545884827767552,
|
|
"grad_norm": 3.029881000518799,
|
|
"learning_rate": 9.904709684420106e-06,
|
|
"loss": 1.1892,
|
|
"step": 1038
|
|
},
|
|
{
|
|
"epoch": 0.5464107283723376,
|
|
"grad_norm": 3.1685080528259277,
|
|
"learning_rate": 9.90437135735066e-06,
|
|
"loss": 1.3857,
|
|
"step": 1039
|
|
},
|
|
{
|
|
"epoch": 0.5469366289771234,
|
|
"grad_norm": 2.9704091548919678,
|
|
"learning_rate": 9.904032436530996e-06,
|
|
"loss": 1.146,
|
|
"step": 1040
|
|
},
|
|
{
|
|
"epoch": 0.547462529581909,
|
|
"grad_norm": 2.946349859237671,
|
|
"learning_rate": 9.903692922002144e-06,
|
|
"loss": 1.3018,
|
|
"step": 1041
|
|
},
|
|
{
|
|
"epoch": 0.5479884301866947,
|
|
"grad_norm": 2.9967007637023926,
|
|
"learning_rate": 9.90335281380521e-06,
|
|
"loss": 1.1316,
|
|
"step": 1042
|
|
},
|
|
{
|
|
"epoch": 0.5485143307914804,
|
|
"grad_norm": 2.9388668537139893,
|
|
"learning_rate": 9.903012111981367e-06,
|
|
"loss": 1.2947,
|
|
"step": 1043
|
|
},
|
|
{
|
|
"epoch": 0.5490402313962661,
|
|
"grad_norm": 2.8456990718841553,
|
|
"learning_rate": 9.902670816571863e-06,
|
|
"loss": 1.1633,
|
|
"step": 1044
|
|
},
|
|
{
|
|
"epoch": 0.5495661320010518,
|
|
"grad_norm": 2.7647342681884766,
|
|
"learning_rate": 9.902328927618019e-06,
|
|
"loss": 1.0863,
|
|
"step": 1045
|
|
},
|
|
{
|
|
"epoch": 0.5500920326058375,
|
|
"grad_norm": 2.6623027324676514,
|
|
"learning_rate": 9.901986445161224e-06,
|
|
"loss": 1.1034,
|
|
"step": 1046
|
|
},
|
|
{
|
|
"epoch": 0.5506179332106232,
|
|
"grad_norm": 2.963656425476074,
|
|
"learning_rate": 9.90164336924294e-06,
|
|
"loss": 1.189,
|
|
"step": 1047
|
|
},
|
|
{
|
|
"epoch": 0.5511438338154089,
|
|
"grad_norm": 2.962766408920288,
|
|
"learning_rate": 9.901299699904705e-06,
|
|
"loss": 1.2195,
|
|
"step": 1048
|
|
},
|
|
{
|
|
"epoch": 0.5516697344201946,
|
|
"grad_norm": 2.8589580059051514,
|
|
"learning_rate": 9.900955437188122e-06,
|
|
"loss": 1.2123,
|
|
"step": 1049
|
|
},
|
|
{
|
|
"epoch": 0.5521956350249803,
|
|
"grad_norm": 3.109720468521118,
|
|
"learning_rate": 9.90061058113487e-06,
|
|
"loss": 1.2765,
|
|
"step": 1050
|
|
},
|
|
{
|
|
"epoch": 0.552721535629766,
|
|
"grad_norm": 2.673027992248535,
|
|
"learning_rate": 9.900265131786702e-06,
|
|
"loss": 1.2779,
|
|
"step": 1051
|
|
},
|
|
{
|
|
"epoch": 0.5532474362345516,
|
|
"grad_norm": 2.991948366165161,
|
|
"learning_rate": 9.899919089185438e-06,
|
|
"loss": 1.1698,
|
|
"step": 1052
|
|
},
|
|
{
|
|
"epoch": 0.5537733368393374,
|
|
"grad_norm": 3.046604871749878,
|
|
"learning_rate": 9.89957245337297e-06,
|
|
"loss": 1.2609,
|
|
"step": 1053
|
|
},
|
|
{
|
|
"epoch": 0.554299237444123,
|
|
"grad_norm": 3.3288161754608154,
|
|
"learning_rate": 9.899225224391266e-06,
|
|
"loss": 1.3253,
|
|
"step": 1054
|
|
},
|
|
{
|
|
"epoch": 0.5548251380489088,
|
|
"grad_norm": 3.1719019412994385,
|
|
"learning_rate": 9.898877402282364e-06,
|
|
"loss": 1.2539,
|
|
"step": 1055
|
|
},
|
|
{
|
|
"epoch": 0.5553510386536945,
|
|
"grad_norm": 2.8405885696411133,
|
|
"learning_rate": 9.89852898708837e-06,
|
|
"loss": 1.1473,
|
|
"step": 1056
|
|
},
|
|
{
|
|
"epoch": 0.5558769392584801,
|
|
"grad_norm": 2.9865121841430664,
|
|
"learning_rate": 9.898179978851468e-06,
|
|
"loss": 1.1963,
|
|
"step": 1057
|
|
},
|
|
{
|
|
"epoch": 0.5564028398632659,
|
|
"grad_norm": 3.112133741378784,
|
|
"learning_rate": 9.89783037761391e-06,
|
|
"loss": 1.2921,
|
|
"step": 1058
|
|
},
|
|
{
|
|
"epoch": 0.5569287404680515,
|
|
"grad_norm": 3.19907283782959,
|
|
"learning_rate": 9.89748018341802e-06,
|
|
"loss": 1.1806,
|
|
"step": 1059
|
|
},
|
|
{
|
|
"epoch": 0.5574546410728373,
|
|
"grad_norm": 3.0074989795684814,
|
|
"learning_rate": 9.897129396306197e-06,
|
|
"loss": 1.2363,
|
|
"step": 1060
|
|
},
|
|
{
|
|
"epoch": 0.5579805416776229,
|
|
"grad_norm": 3.166236400604248,
|
|
"learning_rate": 9.896778016320907e-06,
|
|
"loss": 1.3203,
|
|
"step": 1061
|
|
},
|
|
{
|
|
"epoch": 0.5585064422824086,
|
|
"grad_norm": 2.9215617179870605,
|
|
"learning_rate": 9.89642604350469e-06,
|
|
"loss": 1.1946,
|
|
"step": 1062
|
|
},
|
|
{
|
|
"epoch": 0.5590323428871943,
|
|
"grad_norm": 2.781327724456787,
|
|
"learning_rate": 9.896073477900157e-06,
|
|
"loss": 1.2111,
|
|
"step": 1063
|
|
},
|
|
{
|
|
"epoch": 0.55955824349198,
|
|
"grad_norm": 2.8297834396362305,
|
|
"learning_rate": 9.895720319549993e-06,
|
|
"loss": 1.1649,
|
|
"step": 1064
|
|
},
|
|
{
|
|
"epoch": 0.5600841440967657,
|
|
"grad_norm": 2.9083189964294434,
|
|
"learning_rate": 9.895366568496954e-06,
|
|
"loss": 1.3788,
|
|
"step": 1065
|
|
},
|
|
{
|
|
"epoch": 0.5606100447015514,
|
|
"grad_norm": 2.9475326538085938,
|
|
"learning_rate": 9.895012224783866e-06,
|
|
"loss": 1.2175,
|
|
"step": 1066
|
|
},
|
|
{
|
|
"epoch": 0.5611359453063371,
|
|
"grad_norm": 2.8727071285247803,
|
|
"learning_rate": 9.894657288453627e-06,
|
|
"loss": 1.2519,
|
|
"step": 1067
|
|
},
|
|
{
|
|
"epoch": 0.5616618459111228,
|
|
"grad_norm": 3.0370585918426514,
|
|
"learning_rate": 9.894301759549208e-06,
|
|
"loss": 1.3419,
|
|
"step": 1068
|
|
},
|
|
{
|
|
"epoch": 0.5621877465159085,
|
|
"grad_norm": 2.861717939376831,
|
|
"learning_rate": 9.893945638113653e-06,
|
|
"loss": 1.2622,
|
|
"step": 1069
|
|
},
|
|
{
|
|
"epoch": 0.5627136471206942,
|
|
"grad_norm": 2.7925939559936523,
|
|
"learning_rate": 9.893588924190075e-06,
|
|
"loss": 1.1722,
|
|
"step": 1070
|
|
},
|
|
{
|
|
"epoch": 0.5632395477254799,
|
|
"grad_norm": 2.862372636795044,
|
|
"learning_rate": 9.893231617821658e-06,
|
|
"loss": 1.2507,
|
|
"step": 1071
|
|
},
|
|
{
|
|
"epoch": 0.5637654483302655,
|
|
"grad_norm": 3.047666072845459,
|
|
"learning_rate": 9.89287371905166e-06,
|
|
"loss": 1.2818,
|
|
"step": 1072
|
|
},
|
|
{
|
|
"epoch": 0.5642913489350513,
|
|
"grad_norm": 2.9702773094177246,
|
|
"learning_rate": 9.892515227923413e-06,
|
|
"loss": 1.3005,
|
|
"step": 1073
|
|
},
|
|
{
|
|
"epoch": 0.5648172495398369,
|
|
"grad_norm": 2.702691078186035,
|
|
"learning_rate": 9.892156144480315e-06,
|
|
"loss": 1.1473,
|
|
"step": 1074
|
|
},
|
|
{
|
|
"epoch": 0.5653431501446227,
|
|
"grad_norm": 2.9238154888153076,
|
|
"learning_rate": 9.891796468765839e-06,
|
|
"loss": 1.263,
|
|
"step": 1075
|
|
},
|
|
{
|
|
"epoch": 0.5658690507494084,
|
|
"grad_norm": 2.927729368209839,
|
|
"learning_rate": 9.891436200823533e-06,
|
|
"loss": 1.209,
|
|
"step": 1076
|
|
},
|
|
{
|
|
"epoch": 0.566394951354194,
|
|
"grad_norm": 3.007420063018799,
|
|
"learning_rate": 9.891075340697006e-06,
|
|
"loss": 1.2349,
|
|
"step": 1077
|
|
},
|
|
{
|
|
"epoch": 0.5669208519589798,
|
|
"grad_norm": 2.8338146209716797,
|
|
"learning_rate": 9.890713888429952e-06,
|
|
"loss": 1.2254,
|
|
"step": 1078
|
|
},
|
|
{
|
|
"epoch": 0.5674467525637654,
|
|
"grad_norm": 2.92585825920105,
|
|
"learning_rate": 9.890351844066128e-06,
|
|
"loss": 1.2831,
|
|
"step": 1079
|
|
},
|
|
{
|
|
"epoch": 0.5679726531685512,
|
|
"grad_norm": 2.7623183727264404,
|
|
"learning_rate": 9.889989207649365e-06,
|
|
"loss": 1.2199,
|
|
"step": 1080
|
|
},
|
|
{
|
|
"epoch": 0.5684985537733368,
|
|
"grad_norm": 2.9026641845703125,
|
|
"learning_rate": 9.889625979223565e-06,
|
|
"loss": 1.2962,
|
|
"step": 1081
|
|
},
|
|
{
|
|
"epoch": 0.5690244543781225,
|
|
"grad_norm": 3.1744046211242676,
|
|
"learning_rate": 9.889262158832704e-06,
|
|
"loss": 1.3463,
|
|
"step": 1082
|
|
},
|
|
{
|
|
"epoch": 0.5695503549829082,
|
|
"grad_norm": 3.002772569656372,
|
|
"learning_rate": 9.888897746520827e-06,
|
|
"loss": 1.2861,
|
|
"step": 1083
|
|
},
|
|
{
|
|
"epoch": 0.5700762555876939,
|
|
"grad_norm": 2.7843949794769287,
|
|
"learning_rate": 9.888532742332053e-06,
|
|
"loss": 1.2173,
|
|
"step": 1084
|
|
},
|
|
{
|
|
"epoch": 0.5706021561924797,
|
|
"grad_norm": 2.844496965408325,
|
|
"learning_rate": 9.888167146310572e-06,
|
|
"loss": 1.1576,
|
|
"step": 1085
|
|
},
|
|
{
|
|
"epoch": 0.5711280567972653,
|
|
"grad_norm": 2.913543462753296,
|
|
"learning_rate": 9.887800958500642e-06,
|
|
"loss": 1.2556,
|
|
"step": 1086
|
|
},
|
|
{
|
|
"epoch": 0.571653957402051,
|
|
"grad_norm": 2.9683773517608643,
|
|
"learning_rate": 9.887434178946597e-06,
|
|
"loss": 1.1537,
|
|
"step": 1087
|
|
},
|
|
{
|
|
"epoch": 0.5721798580068367,
|
|
"grad_norm": 2.9056577682495117,
|
|
"learning_rate": 9.887066807692842e-06,
|
|
"loss": 1.2671,
|
|
"step": 1088
|
|
},
|
|
{
|
|
"epoch": 0.5727057586116224,
|
|
"grad_norm": 2.9720230102539062,
|
|
"learning_rate": 9.886698844783854e-06,
|
|
"loss": 1.147,
|
|
"step": 1089
|
|
},
|
|
{
|
|
"epoch": 0.5732316592164081,
|
|
"grad_norm": 3.0075857639312744,
|
|
"learning_rate": 9.886330290264177e-06,
|
|
"loss": 1.2153,
|
|
"step": 1090
|
|
},
|
|
{
|
|
"epoch": 0.5737575598211938,
|
|
"grad_norm": 2.823330879211426,
|
|
"learning_rate": 9.885961144178435e-06,
|
|
"loss": 1.2002,
|
|
"step": 1091
|
|
},
|
|
{
|
|
"epoch": 0.5742834604259794,
|
|
"grad_norm": 2.742311477661133,
|
|
"learning_rate": 9.885591406571315e-06,
|
|
"loss": 1.1559,
|
|
"step": 1092
|
|
},
|
|
{
|
|
"epoch": 0.5748093610307652,
|
|
"grad_norm": 2.890669107437134,
|
|
"learning_rate": 9.885221077487583e-06,
|
|
"loss": 1.317,
|
|
"step": 1093
|
|
},
|
|
{
|
|
"epoch": 0.5753352616355509,
|
|
"grad_norm": 2.81160831451416,
|
|
"learning_rate": 9.884850156972069e-06,
|
|
"loss": 1.2341,
|
|
"step": 1094
|
|
},
|
|
{
|
|
"epoch": 0.5758611622403366,
|
|
"grad_norm": 2.895878553390503,
|
|
"learning_rate": 9.884478645069682e-06,
|
|
"loss": 1.0997,
|
|
"step": 1095
|
|
},
|
|
{
|
|
"epoch": 0.5763870628451223,
|
|
"grad_norm": 2.983081817626953,
|
|
"learning_rate": 9.884106541825396e-06,
|
|
"loss": 1.2206,
|
|
"step": 1096
|
|
},
|
|
{
|
|
"epoch": 0.576912963449908,
|
|
"grad_norm": 2.928705930709839,
|
|
"learning_rate": 9.883733847284264e-06,
|
|
"loss": 1.3028,
|
|
"step": 1097
|
|
},
|
|
{
|
|
"epoch": 0.5774388640546937,
|
|
"grad_norm": 3.2309935092926025,
|
|
"learning_rate": 9.883360561491403e-06,
|
|
"loss": 1.2243,
|
|
"step": 1098
|
|
},
|
|
{
|
|
"epoch": 0.5779647646594793,
|
|
"grad_norm": 2.942866325378418,
|
|
"learning_rate": 9.882986684492007e-06,
|
|
"loss": 1.1526,
|
|
"step": 1099
|
|
},
|
|
{
|
|
"epoch": 0.5784906652642651,
|
|
"grad_norm": 2.817138910293579,
|
|
"learning_rate": 9.88261221633134e-06,
|
|
"loss": 1.2336,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"epoch": 0.5790165658690507,
|
|
"grad_norm": 2.9747719764709473,
|
|
"learning_rate": 9.882237157054734e-06,
|
|
"loss": 1.2975,
|
|
"step": 1101
|
|
},
|
|
{
|
|
"epoch": 0.5795424664738364,
|
|
"grad_norm": 2.9870190620422363,
|
|
"learning_rate": 9.881861506707599e-06,
|
|
"loss": 1.2749,
|
|
"step": 1102
|
|
},
|
|
{
|
|
"epoch": 0.5800683670786222,
|
|
"grad_norm": 3.102616548538208,
|
|
"learning_rate": 9.881485265335411e-06,
|
|
"loss": 1.3019,
|
|
"step": 1103
|
|
},
|
|
{
|
|
"epoch": 0.5805942676834078,
|
|
"grad_norm": 2.946011781692505,
|
|
"learning_rate": 9.881108432983723e-06,
|
|
"loss": 1.2275,
|
|
"step": 1104
|
|
},
|
|
{
|
|
"epoch": 0.5811201682881936,
|
|
"grad_norm": 2.977484703063965,
|
|
"learning_rate": 9.880731009698153e-06,
|
|
"loss": 1.1793,
|
|
"step": 1105
|
|
},
|
|
{
|
|
"epoch": 0.5816460688929792,
|
|
"grad_norm": 2.9232490062713623,
|
|
"learning_rate": 9.880352995524396e-06,
|
|
"loss": 1.2871,
|
|
"step": 1106
|
|
},
|
|
{
|
|
"epoch": 0.582171969497765,
|
|
"grad_norm": 2.815870761871338,
|
|
"learning_rate": 9.879974390508215e-06,
|
|
"loss": 1.2531,
|
|
"step": 1107
|
|
},
|
|
{
|
|
"epoch": 0.5826978701025506,
|
|
"grad_norm": 2.8903660774230957,
|
|
"learning_rate": 9.879595194695448e-06,
|
|
"loss": 1.2228,
|
|
"step": 1108
|
|
},
|
|
{
|
|
"epoch": 0.5832237707073363,
|
|
"grad_norm": 2.855076313018799,
|
|
"learning_rate": 9.879215408132002e-06,
|
|
"loss": 1.1694,
|
|
"step": 1109
|
|
},
|
|
{
|
|
"epoch": 0.583749671312122,
|
|
"grad_norm": 3.058760166168213,
|
|
"learning_rate": 9.878835030863853e-06,
|
|
"loss": 1.1741,
|
|
"step": 1110
|
|
},
|
|
{
|
|
"epoch": 0.5842755719169077,
|
|
"grad_norm": 3.22556209564209,
|
|
"learning_rate": 9.878454062937056e-06,
|
|
"loss": 1.2717,
|
|
"step": 1111
|
|
},
|
|
{
|
|
"epoch": 0.5848014725216935,
|
|
"grad_norm": 2.8012781143188477,
|
|
"learning_rate": 9.878072504397732e-06,
|
|
"loss": 1.1686,
|
|
"step": 1112
|
|
},
|
|
{
|
|
"epoch": 0.5853273731264791,
|
|
"grad_norm": 2.9688565731048584,
|
|
"learning_rate": 9.87769035529207e-06,
|
|
"loss": 1.3618,
|
|
"step": 1113
|
|
},
|
|
{
|
|
"epoch": 0.5858532737312648,
|
|
"grad_norm": 2.9511983394622803,
|
|
"learning_rate": 9.877307615666344e-06,
|
|
"loss": 1.3365,
|
|
"step": 1114
|
|
},
|
|
{
|
|
"epoch": 0.5863791743360505,
|
|
"grad_norm": 2.7541942596435547,
|
|
"learning_rate": 9.87692428556688e-06,
|
|
"loss": 1.1818,
|
|
"step": 1115
|
|
},
|
|
{
|
|
"epoch": 0.5869050749408362,
|
|
"grad_norm": 2.952887535095215,
|
|
"learning_rate": 9.876540365040094e-06,
|
|
"loss": 1.2087,
|
|
"step": 1116
|
|
},
|
|
{
|
|
"epoch": 0.5874309755456218,
|
|
"grad_norm": 2.7914369106292725,
|
|
"learning_rate": 9.876155854132463e-06,
|
|
"loss": 1.2555,
|
|
"step": 1117
|
|
},
|
|
{
|
|
"epoch": 0.5879568761504076,
|
|
"grad_norm": 2.8920202255249023,
|
|
"learning_rate": 9.875770752890538e-06,
|
|
"loss": 1.2627,
|
|
"step": 1118
|
|
},
|
|
{
|
|
"epoch": 0.5884827767551932,
|
|
"grad_norm": 3.0850670337677,
|
|
"learning_rate": 9.87538506136094e-06,
|
|
"loss": 1.2801,
|
|
"step": 1119
|
|
},
|
|
{
|
|
"epoch": 0.589008677359979,
|
|
"grad_norm": 2.989230155944824,
|
|
"learning_rate": 9.874998779590365e-06,
|
|
"loss": 1.2169,
|
|
"step": 1120
|
|
},
|
|
{
|
|
"epoch": 0.5895345779647646,
|
|
"grad_norm": 2.738572835922241,
|
|
"learning_rate": 9.87461190762558e-06,
|
|
"loss": 1.1585,
|
|
"step": 1121
|
|
},
|
|
{
|
|
"epoch": 0.5900604785695504,
|
|
"grad_norm": 2.6718099117279053,
|
|
"learning_rate": 9.874224445513416e-06,
|
|
"loss": 1.1963,
|
|
"step": 1122
|
|
},
|
|
{
|
|
"epoch": 0.5905863791743361,
|
|
"grad_norm": 2.8818306922912598,
|
|
"learning_rate": 9.873836393300788e-06,
|
|
"loss": 1.1997,
|
|
"step": 1123
|
|
},
|
|
{
|
|
"epoch": 0.5911122797791217,
|
|
"grad_norm": 2.908029317855835,
|
|
"learning_rate": 9.873447751034671e-06,
|
|
"loss": 1.1953,
|
|
"step": 1124
|
|
},
|
|
{
|
|
"epoch": 0.5916381803839075,
|
|
"grad_norm": 2.8938114643096924,
|
|
"learning_rate": 9.873058518762117e-06,
|
|
"loss": 1.2513,
|
|
"step": 1125
|
|
},
|
|
{
|
|
"epoch": 0.5921640809886931,
|
|
"grad_norm": 2.8417742252349854,
|
|
"learning_rate": 9.872668696530251e-06,
|
|
"loss": 1.1644,
|
|
"step": 1126
|
|
},
|
|
{
|
|
"epoch": 0.5926899815934789,
|
|
"grad_norm": 2.837904691696167,
|
|
"learning_rate": 9.872278284386266e-06,
|
|
"loss": 1.2429,
|
|
"step": 1127
|
|
},
|
|
{
|
|
"epoch": 0.5932158821982645,
|
|
"grad_norm": 3.20475435256958,
|
|
"learning_rate": 9.871887282377424e-06,
|
|
"loss": 1.2443,
|
|
"step": 1128
|
|
},
|
|
{
|
|
"epoch": 0.5937417828030502,
|
|
"grad_norm": 2.779391288757324,
|
|
"learning_rate": 9.871495690551067e-06,
|
|
"loss": 1.2201,
|
|
"step": 1129
|
|
},
|
|
{
|
|
"epoch": 0.5942676834078359,
|
|
"grad_norm": 3.0963938236236572,
|
|
"learning_rate": 9.871103508954601e-06,
|
|
"loss": 1.2457,
|
|
"step": 1130
|
|
},
|
|
{
|
|
"epoch": 0.5947935840126216,
|
|
"grad_norm": 3.3356997966766357,
|
|
"learning_rate": 9.870710737635506e-06,
|
|
"loss": 1.2576,
|
|
"step": 1131
|
|
},
|
|
{
|
|
"epoch": 0.5953194846174074,
|
|
"grad_norm": 3.1054699420928955,
|
|
"learning_rate": 9.87031737664133e-06,
|
|
"loss": 1.328,
|
|
"step": 1132
|
|
},
|
|
{
|
|
"epoch": 0.595845385222193,
|
|
"grad_norm": 2.8053927421569824,
|
|
"learning_rate": 9.869923426019701e-06,
|
|
"loss": 1.2379,
|
|
"step": 1133
|
|
},
|
|
{
|
|
"epoch": 0.5963712858269787,
|
|
"grad_norm": 2.8277676105499268,
|
|
"learning_rate": 9.869528885818309e-06,
|
|
"loss": 1.2597,
|
|
"step": 1134
|
|
},
|
|
{
|
|
"epoch": 0.5968971864317644,
|
|
"grad_norm": 2.9099721908569336,
|
|
"learning_rate": 9.869133756084921e-06,
|
|
"loss": 1.2457,
|
|
"step": 1135
|
|
},
|
|
{
|
|
"epoch": 0.5974230870365501,
|
|
"grad_norm": 2.8902053833007812,
|
|
"learning_rate": 9.868738036867371e-06,
|
|
"loss": 1.2076,
|
|
"step": 1136
|
|
},
|
|
{
|
|
"epoch": 0.5979489876413357,
|
|
"grad_norm": 2.7374625205993652,
|
|
"learning_rate": 9.868341728213572e-06,
|
|
"loss": 1.1983,
|
|
"step": 1137
|
|
},
|
|
{
|
|
"epoch": 0.5984748882461215,
|
|
"grad_norm": 2.8444416522979736,
|
|
"learning_rate": 9.867944830171496e-06,
|
|
"loss": 1.2001,
|
|
"step": 1138
|
|
},
|
|
{
|
|
"epoch": 0.5990007888509071,
|
|
"grad_norm": 3.0819814205169678,
|
|
"learning_rate": 9.8675473427892e-06,
|
|
"loss": 1.2628,
|
|
"step": 1139
|
|
},
|
|
{
|
|
"epoch": 0.5995266894556929,
|
|
"grad_norm": 2.9534473419189453,
|
|
"learning_rate": 9.867149266114804e-06,
|
|
"loss": 1.2968,
|
|
"step": 1140
|
|
},
|
|
{
|
|
"epoch": 0.6000525900604786,
|
|
"grad_norm": 3.171416997909546,
|
|
"learning_rate": 9.8667506001965e-06,
|
|
"loss": 1.2631,
|
|
"step": 1141
|
|
},
|
|
{
|
|
"epoch": 0.6005784906652643,
|
|
"grad_norm": 2.8839311599731445,
|
|
"learning_rate": 9.866351345082554e-06,
|
|
"loss": 1.2534,
|
|
"step": 1142
|
|
},
|
|
{
|
|
"epoch": 0.60110439127005,
|
|
"grad_norm": 2.842177391052246,
|
|
"learning_rate": 9.865951500821303e-06,
|
|
"loss": 1.2521,
|
|
"step": 1143
|
|
},
|
|
{
|
|
"epoch": 0.6016302918748356,
|
|
"grad_norm": 3.3243181705474854,
|
|
"learning_rate": 9.865551067461152e-06,
|
|
"loss": 1.129,
|
|
"step": 1144
|
|
},
|
|
{
|
|
"epoch": 0.6021561924796214,
|
|
"grad_norm": 3.1493728160858154,
|
|
"learning_rate": 9.86515004505058e-06,
|
|
"loss": 1.1211,
|
|
"step": 1145
|
|
},
|
|
{
|
|
"epoch": 0.602682093084407,
|
|
"grad_norm": 2.7987709045410156,
|
|
"learning_rate": 9.864748433638138e-06,
|
|
"loss": 1.1291,
|
|
"step": 1146
|
|
},
|
|
{
|
|
"epoch": 0.6032079936891928,
|
|
"grad_norm": 2.9978768825531006,
|
|
"learning_rate": 9.864346233272449e-06,
|
|
"loss": 1.2629,
|
|
"step": 1147
|
|
},
|
|
{
|
|
"epoch": 0.6037338942939784,
|
|
"grad_norm": 2.908360719680786,
|
|
"learning_rate": 9.863943444002202e-06,
|
|
"loss": 1.165,
|
|
"step": 1148
|
|
},
|
|
{
|
|
"epoch": 0.6042597948987641,
|
|
"grad_norm": 2.9886698722839355,
|
|
"learning_rate": 9.863540065876164e-06,
|
|
"loss": 1.247,
|
|
"step": 1149
|
|
},
|
|
{
|
|
"epoch": 0.6047856955035499,
|
|
"grad_norm": 2.7450129985809326,
|
|
"learning_rate": 9.863136098943168e-06,
|
|
"loss": 1.1252,
|
|
"step": 1150
|
|
},
|
|
{
|
|
"epoch": 0.6053115961083355,
|
|
"grad_norm": 2.9063286781311035,
|
|
"learning_rate": 9.86273154325212e-06,
|
|
"loss": 1.4199,
|
|
"step": 1151
|
|
},
|
|
{
|
|
"epoch": 0.6058374967131213,
|
|
"grad_norm": 2.8193304538726807,
|
|
"learning_rate": 9.862326398852e-06,
|
|
"loss": 1.1992,
|
|
"step": 1152
|
|
},
|
|
{
|
|
"epoch": 0.6063633973179069,
|
|
"grad_norm": 2.908806562423706,
|
|
"learning_rate": 9.861920665791858e-06,
|
|
"loss": 1.2212,
|
|
"step": 1153
|
|
},
|
|
{
|
|
"epoch": 0.6068892979226926,
|
|
"grad_norm": 2.949800968170166,
|
|
"learning_rate": 9.861514344120811e-06,
|
|
"loss": 1.3546,
|
|
"step": 1154
|
|
},
|
|
{
|
|
"epoch": 0.6074151985274783,
|
|
"grad_norm": 2.7376151084899902,
|
|
"learning_rate": 9.86110743388805e-06,
|
|
"loss": 1.1216,
|
|
"step": 1155
|
|
},
|
|
{
|
|
"epoch": 0.607941099132264,
|
|
"grad_norm": 3.061697483062744,
|
|
"learning_rate": 9.860699935142842e-06,
|
|
"loss": 1.2528,
|
|
"step": 1156
|
|
},
|
|
{
|
|
"epoch": 0.6084669997370497,
|
|
"grad_norm": 3.096349000930786,
|
|
"learning_rate": 9.860291847934515e-06,
|
|
"loss": 1.2462,
|
|
"step": 1157
|
|
},
|
|
{
|
|
"epoch": 0.6089929003418354,
|
|
"grad_norm": 2.7088189125061035,
|
|
"learning_rate": 9.85988317231248e-06,
|
|
"loss": 1.1799,
|
|
"step": 1158
|
|
},
|
|
{
|
|
"epoch": 0.6095188009466211,
|
|
"grad_norm": 2.8645360469818115,
|
|
"learning_rate": 9.859473908326214e-06,
|
|
"loss": 1.1607,
|
|
"step": 1159
|
|
},
|
|
{
|
|
"epoch": 0.6100447015514068,
|
|
"grad_norm": 2.8342061042785645,
|
|
"learning_rate": 9.85906405602526e-06,
|
|
"loss": 1.1878,
|
|
"step": 1160
|
|
},
|
|
{
|
|
"epoch": 0.6105706021561925,
|
|
"grad_norm": 3.057142734527588,
|
|
"learning_rate": 9.85865361545924e-06,
|
|
"loss": 1.2427,
|
|
"step": 1161
|
|
},
|
|
{
|
|
"epoch": 0.6110965027609782,
|
|
"grad_norm": 3.0184073448181152,
|
|
"learning_rate": 9.858242586677842e-06,
|
|
"loss": 1.1566,
|
|
"step": 1162
|
|
},
|
|
{
|
|
"epoch": 0.6116224033657639,
|
|
"grad_norm": 2.936267614364624,
|
|
"learning_rate": 9.857830969730828e-06,
|
|
"loss": 1.1398,
|
|
"step": 1163
|
|
},
|
|
{
|
|
"epoch": 0.6121483039705495,
|
|
"grad_norm": 3.052213430404663,
|
|
"learning_rate": 9.857418764668033e-06,
|
|
"loss": 1.1912,
|
|
"step": 1164
|
|
},
|
|
{
|
|
"epoch": 0.6126742045753353,
|
|
"grad_norm": 3.0020461082458496,
|
|
"learning_rate": 9.85700597153936e-06,
|
|
"loss": 1.3158,
|
|
"step": 1165
|
|
},
|
|
{
|
|
"epoch": 0.6132001051801209,
|
|
"grad_norm": 2.701803684234619,
|
|
"learning_rate": 9.856592590394781e-06,
|
|
"loss": 1.0968,
|
|
"step": 1166
|
|
},
|
|
{
|
|
"epoch": 0.6137260057849067,
|
|
"grad_norm": 3.0734827518463135,
|
|
"learning_rate": 9.856178621284346e-06,
|
|
"loss": 1.3177,
|
|
"step": 1167
|
|
},
|
|
{
|
|
"epoch": 0.6142519063896924,
|
|
"grad_norm": 2.695302963256836,
|
|
"learning_rate": 9.85576406425817e-06,
|
|
"loss": 1.2224,
|
|
"step": 1168
|
|
},
|
|
{
|
|
"epoch": 0.614777806994478,
|
|
"grad_norm": 2.7761330604553223,
|
|
"learning_rate": 9.855348919366444e-06,
|
|
"loss": 1.2177,
|
|
"step": 1169
|
|
},
|
|
{
|
|
"epoch": 0.6153037075992638,
|
|
"grad_norm": 2.905829906463623,
|
|
"learning_rate": 9.854933186659424e-06,
|
|
"loss": 1.1947,
|
|
"step": 1170
|
|
},
|
|
{
|
|
"epoch": 0.6158296082040494,
|
|
"grad_norm": 3.0176827907562256,
|
|
"learning_rate": 9.854516866187443e-06,
|
|
"loss": 1.1763,
|
|
"step": 1171
|
|
},
|
|
{
|
|
"epoch": 0.6163555088088352,
|
|
"grad_norm": 2.899966239929199,
|
|
"learning_rate": 9.854099958000904e-06,
|
|
"loss": 1.2443,
|
|
"step": 1172
|
|
},
|
|
{
|
|
"epoch": 0.6168814094136208,
|
|
"grad_norm": 3.0032522678375244,
|
|
"learning_rate": 9.853682462150281e-06,
|
|
"loss": 1.1348,
|
|
"step": 1173
|
|
},
|
|
{
|
|
"epoch": 0.6174073100184065,
|
|
"grad_norm": 2.8970019817352295,
|
|
"learning_rate": 9.853264378686114e-06,
|
|
"loss": 1.2285,
|
|
"step": 1174
|
|
},
|
|
{
|
|
"epoch": 0.6179332106231922,
|
|
"grad_norm": 2.7990944385528564,
|
|
"learning_rate": 9.852845707659022e-06,
|
|
"loss": 1.2566,
|
|
"step": 1175
|
|
},
|
|
{
|
|
"epoch": 0.6184591112279779,
|
|
"grad_norm": 2.727907419204712,
|
|
"learning_rate": 9.85242644911969e-06,
|
|
"loss": 1.1558,
|
|
"step": 1176
|
|
},
|
|
{
|
|
"epoch": 0.6189850118327636,
|
|
"grad_norm": 2.920515298843384,
|
|
"learning_rate": 9.852006603118876e-06,
|
|
"loss": 1.3521,
|
|
"step": 1177
|
|
},
|
|
{
|
|
"epoch": 0.6195109124375493,
|
|
"grad_norm": 3.1616733074188232,
|
|
"learning_rate": 9.85158616970741e-06,
|
|
"loss": 1.2357,
|
|
"step": 1178
|
|
},
|
|
{
|
|
"epoch": 0.620036813042335,
|
|
"grad_norm": 2.8488614559173584,
|
|
"learning_rate": 9.85116514893619e-06,
|
|
"loss": 1.2732,
|
|
"step": 1179
|
|
},
|
|
{
|
|
"epoch": 0.6205627136471207,
|
|
"grad_norm": 2.872487783432007,
|
|
"learning_rate": 9.850743540856191e-06,
|
|
"loss": 1.2596,
|
|
"step": 1180
|
|
},
|
|
{
|
|
"epoch": 0.6210886142519064,
|
|
"grad_norm": 2.7959396839141846,
|
|
"learning_rate": 9.850321345518451e-06,
|
|
"loss": 1.2529,
|
|
"step": 1181
|
|
},
|
|
{
|
|
"epoch": 0.6216145148566921,
|
|
"grad_norm": 3.124046564102173,
|
|
"learning_rate": 9.849898562974087e-06,
|
|
"loss": 1.3045,
|
|
"step": 1182
|
|
},
|
|
{
|
|
"epoch": 0.6221404154614778,
|
|
"grad_norm": 2.9152283668518066,
|
|
"learning_rate": 9.849475193274278e-06,
|
|
"loss": 1.2158,
|
|
"step": 1183
|
|
},
|
|
{
|
|
"epoch": 0.6226663160662634,
|
|
"grad_norm": 2.8270018100738525,
|
|
"learning_rate": 9.849051236470285e-06,
|
|
"loss": 1.1475,
|
|
"step": 1184
|
|
},
|
|
{
|
|
"epoch": 0.6231922166710492,
|
|
"grad_norm": 2.934312343597412,
|
|
"learning_rate": 9.848626692613432e-06,
|
|
"loss": 1.2236,
|
|
"step": 1185
|
|
},
|
|
{
|
|
"epoch": 0.6237181172758348,
|
|
"grad_norm": 2.8617770671844482,
|
|
"learning_rate": 9.848201561755116e-06,
|
|
"loss": 1.2421,
|
|
"step": 1186
|
|
},
|
|
{
|
|
"epoch": 0.6242440178806206,
|
|
"grad_norm": 2.822355031967163,
|
|
"learning_rate": 9.847775843946807e-06,
|
|
"loss": 1.1606,
|
|
"step": 1187
|
|
},
|
|
{
|
|
"epoch": 0.6247699184854063,
|
|
"grad_norm": 2.829218626022339,
|
|
"learning_rate": 9.847349539240045e-06,
|
|
"loss": 1.1846,
|
|
"step": 1188
|
|
},
|
|
{
|
|
"epoch": 0.6252958190901919,
|
|
"grad_norm": 2.81400990486145,
|
|
"learning_rate": 9.846922647686439e-06,
|
|
"loss": 1.2544,
|
|
"step": 1189
|
|
},
|
|
{
|
|
"epoch": 0.6258217196949777,
|
|
"grad_norm": 2.747570276260376,
|
|
"learning_rate": 9.846495169337672e-06,
|
|
"loss": 1.224,
|
|
"step": 1190
|
|
},
|
|
{
|
|
"epoch": 0.6263476202997633,
|
|
"grad_norm": 2.7922067642211914,
|
|
"learning_rate": 9.846067104245499e-06,
|
|
"loss": 1.1157,
|
|
"step": 1191
|
|
},
|
|
{
|
|
"epoch": 0.6268735209045491,
|
|
"grad_norm": 2.948368549346924,
|
|
"learning_rate": 9.84563845246174e-06,
|
|
"loss": 1.3504,
|
|
"step": 1192
|
|
},
|
|
{
|
|
"epoch": 0.6273994215093347,
|
|
"grad_norm": 2.7187671661376953,
|
|
"learning_rate": 9.84520921403829e-06,
|
|
"loss": 1.0773,
|
|
"step": 1193
|
|
},
|
|
{
|
|
"epoch": 0.6279253221141204,
|
|
"grad_norm": 2.882554054260254,
|
|
"learning_rate": 9.844779389027118e-06,
|
|
"loss": 1.2262,
|
|
"step": 1194
|
|
},
|
|
{
|
|
"epoch": 0.6284512227189061,
|
|
"grad_norm": 2.9084200859069824,
|
|
"learning_rate": 9.844348977480261e-06,
|
|
"loss": 1.3024,
|
|
"step": 1195
|
|
},
|
|
{
|
|
"epoch": 0.6289771233236918,
|
|
"grad_norm": 2.8025171756744385,
|
|
"learning_rate": 9.843917979449825e-06,
|
|
"loss": 1.2576,
|
|
"step": 1196
|
|
},
|
|
{
|
|
"epoch": 0.6295030239284776,
|
|
"grad_norm": 2.887089729309082,
|
|
"learning_rate": 9.843486394987989e-06,
|
|
"loss": 1.2942,
|
|
"step": 1197
|
|
},
|
|
{
|
|
"epoch": 0.6300289245332632,
|
|
"grad_norm": 2.900055170059204,
|
|
"learning_rate": 9.843054224147003e-06,
|
|
"loss": 1.1832,
|
|
"step": 1198
|
|
},
|
|
{
|
|
"epoch": 0.630554825138049,
|
|
"grad_norm": 2.9115540981292725,
|
|
"learning_rate": 9.842621466979191e-06,
|
|
"loss": 1.2126,
|
|
"step": 1199
|
|
},
|
|
{
|
|
"epoch": 0.6310807257428346,
|
|
"grad_norm": 3.0154106616973877,
|
|
"learning_rate": 9.842188123536942e-06,
|
|
"loss": 1.2598,
|
|
"step": 1200
|
|
},
|
|
{
|
|
"epoch": 0.6316066263476203,
|
|
"grad_norm": 2.8471665382385254,
|
|
"learning_rate": 9.84175419387272e-06,
|
|
"loss": 1.1353,
|
|
"step": 1201
|
|
},
|
|
{
|
|
"epoch": 0.632132526952406,
|
|
"grad_norm": 3.0407161712646484,
|
|
"learning_rate": 9.841319678039056e-06,
|
|
"loss": 1.1914,
|
|
"step": 1202
|
|
},
|
|
{
|
|
"epoch": 0.6326584275571917,
|
|
"grad_norm": 2.578490972518921,
|
|
"learning_rate": 9.840884576088558e-06,
|
|
"loss": 1.1705,
|
|
"step": 1203
|
|
},
|
|
{
|
|
"epoch": 0.6331843281619773,
|
|
"grad_norm": 3.2552783489227295,
|
|
"learning_rate": 9.840448888073902e-06,
|
|
"loss": 1.3417,
|
|
"step": 1204
|
|
},
|
|
{
|
|
"epoch": 0.6337102287667631,
|
|
"grad_norm": 3.0077826976776123,
|
|
"learning_rate": 9.840012614047833e-06,
|
|
"loss": 1.2057,
|
|
"step": 1205
|
|
},
|
|
{
|
|
"epoch": 0.6342361293715488,
|
|
"grad_norm": 2.7373194694519043,
|
|
"learning_rate": 9.839575754063171e-06,
|
|
"loss": 1.3792,
|
|
"step": 1206
|
|
},
|
|
{
|
|
"epoch": 0.6347620299763345,
|
|
"grad_norm": 3.0783498287200928,
|
|
"learning_rate": 9.839138308172803e-06,
|
|
"loss": 1.3181,
|
|
"step": 1207
|
|
},
|
|
{
|
|
"epoch": 0.6352879305811202,
|
|
"grad_norm": 3.001420021057129,
|
|
"learning_rate": 9.83870027642969e-06,
|
|
"loss": 1.2544,
|
|
"step": 1208
|
|
},
|
|
{
|
|
"epoch": 0.6358138311859058,
|
|
"grad_norm": 2.9726805686950684,
|
|
"learning_rate": 9.83826165888686e-06,
|
|
"loss": 1.2803,
|
|
"step": 1209
|
|
},
|
|
{
|
|
"epoch": 0.6363397317906916,
|
|
"grad_norm": 2.8482422828674316,
|
|
"learning_rate": 9.837822455597417e-06,
|
|
"loss": 1.13,
|
|
"step": 1210
|
|
},
|
|
{
|
|
"epoch": 0.6368656323954772,
|
|
"grad_norm": 2.7807843685150146,
|
|
"learning_rate": 9.83738266661453e-06,
|
|
"loss": 1.1279,
|
|
"step": 1211
|
|
},
|
|
{
|
|
"epoch": 0.637391533000263,
|
|
"grad_norm": 2.919663906097412,
|
|
"learning_rate": 9.836942291991448e-06,
|
|
"loss": 1.187,
|
|
"step": 1212
|
|
},
|
|
{
|
|
"epoch": 0.6379174336050486,
|
|
"grad_norm": 2.8245253562927246,
|
|
"learning_rate": 9.83650133178148e-06,
|
|
"loss": 1.2324,
|
|
"step": 1213
|
|
},
|
|
{
|
|
"epoch": 0.6384433342098343,
|
|
"grad_norm": 2.828397512435913,
|
|
"learning_rate": 9.836059786038013e-06,
|
|
"loss": 1.2649,
|
|
"step": 1214
|
|
},
|
|
{
|
|
"epoch": 0.6389692348146201,
|
|
"grad_norm": 2.9688873291015625,
|
|
"learning_rate": 9.835617654814503e-06,
|
|
"loss": 1.1529,
|
|
"step": 1215
|
|
},
|
|
{
|
|
"epoch": 0.6394951354194057,
|
|
"grad_norm": 2.8672544956207275,
|
|
"learning_rate": 9.835174938164476e-06,
|
|
"loss": 1.2503,
|
|
"step": 1216
|
|
},
|
|
{
|
|
"epoch": 0.6400210360241915,
|
|
"grad_norm": 3.2327804565429688,
|
|
"learning_rate": 9.83473163614153e-06,
|
|
"loss": 1.2998,
|
|
"step": 1217
|
|
},
|
|
{
|
|
"epoch": 0.6405469366289771,
|
|
"grad_norm": 3.0182299613952637,
|
|
"learning_rate": 9.834287748799336e-06,
|
|
"loss": 1.0715,
|
|
"step": 1218
|
|
},
|
|
{
|
|
"epoch": 0.6410728372337628,
|
|
"grad_norm": 2.929248094558716,
|
|
"learning_rate": 9.83384327619163e-06,
|
|
"loss": 1.2087,
|
|
"step": 1219
|
|
},
|
|
{
|
|
"epoch": 0.6415987378385485,
|
|
"grad_norm": 3.1292219161987305,
|
|
"learning_rate": 9.833398218372225e-06,
|
|
"loss": 1.243,
|
|
"step": 1220
|
|
},
|
|
{
|
|
"epoch": 0.6421246384433342,
|
|
"grad_norm": 2.7341246604919434,
|
|
"learning_rate": 9.832952575395001e-06,
|
|
"loss": 1.2228,
|
|
"step": 1221
|
|
},
|
|
{
|
|
"epoch": 0.6426505390481199,
|
|
"grad_norm": 2.839085102081299,
|
|
"learning_rate": 9.832506347313908e-06,
|
|
"loss": 1.1816,
|
|
"step": 1222
|
|
},
|
|
{
|
|
"epoch": 0.6431764396529056,
|
|
"grad_norm": 2.7558884620666504,
|
|
"learning_rate": 9.832059534182971e-06,
|
|
"loss": 1.2117,
|
|
"step": 1223
|
|
},
|
|
{
|
|
"epoch": 0.6437023402576912,
|
|
"grad_norm": 2.795069932937622,
|
|
"learning_rate": 9.831612136056285e-06,
|
|
"loss": 1.1914,
|
|
"step": 1224
|
|
},
|
|
{
|
|
"epoch": 0.644228240862477,
|
|
"grad_norm": 2.7708098888397217,
|
|
"learning_rate": 9.831164152988013e-06,
|
|
"loss": 1.2077,
|
|
"step": 1225
|
|
},
|
|
{
|
|
"epoch": 0.6447541414672627,
|
|
"grad_norm": 2.9189963340759277,
|
|
"learning_rate": 9.830715585032387e-06,
|
|
"loss": 1.2246,
|
|
"step": 1226
|
|
},
|
|
{
|
|
"epoch": 0.6452800420720484,
|
|
"grad_norm": 2.8668136596679688,
|
|
"learning_rate": 9.830266432243718e-06,
|
|
"loss": 1.2088,
|
|
"step": 1227
|
|
},
|
|
{
|
|
"epoch": 0.6458059426768341,
|
|
"grad_norm": 3.1409153938293457,
|
|
"learning_rate": 9.829816694676383e-06,
|
|
"loss": 1.3142,
|
|
"step": 1228
|
|
},
|
|
{
|
|
"epoch": 0.6463318432816197,
|
|
"grad_norm": 2.898503065109253,
|
|
"learning_rate": 9.829366372384826e-06,
|
|
"loss": 1.1916,
|
|
"step": 1229
|
|
},
|
|
{
|
|
"epoch": 0.6468577438864055,
|
|
"grad_norm": 2.9369659423828125,
|
|
"learning_rate": 9.828915465423567e-06,
|
|
"loss": 1.3092,
|
|
"step": 1230
|
|
},
|
|
{
|
|
"epoch": 0.6473836444911911,
|
|
"grad_norm": 3.1951394081115723,
|
|
"learning_rate": 9.828463973847198e-06,
|
|
"loss": 1.1887,
|
|
"step": 1231
|
|
},
|
|
{
|
|
"epoch": 0.6479095450959769,
|
|
"grad_norm": 3.117074489593506,
|
|
"learning_rate": 9.828011897710374e-06,
|
|
"loss": 1.2152,
|
|
"step": 1232
|
|
},
|
|
{
|
|
"epoch": 0.6484354457007625,
|
|
"grad_norm": 2.9105660915374756,
|
|
"learning_rate": 9.82755923706783e-06,
|
|
"loss": 1.2562,
|
|
"step": 1233
|
|
},
|
|
{
|
|
"epoch": 0.6489613463055482,
|
|
"grad_norm": 2.9219205379486084,
|
|
"learning_rate": 9.827105991974365e-06,
|
|
"loss": 1.2865,
|
|
"step": 1234
|
|
},
|
|
{
|
|
"epoch": 0.649487246910334,
|
|
"grad_norm": 2.790290117263794,
|
|
"learning_rate": 9.826652162484855e-06,
|
|
"loss": 1.3198,
|
|
"step": 1235
|
|
},
|
|
{
|
|
"epoch": 0.6500131475151196,
|
|
"grad_norm": 2.998270273208618,
|
|
"learning_rate": 9.826197748654239e-06,
|
|
"loss": 1.326,
|
|
"step": 1236
|
|
},
|
|
{
|
|
"epoch": 0.6505390481199054,
|
|
"grad_norm": 2.7972640991210938,
|
|
"learning_rate": 9.825742750537532e-06,
|
|
"loss": 1.1789,
|
|
"step": 1237
|
|
},
|
|
{
|
|
"epoch": 0.651064948724691,
|
|
"grad_norm": 2.8276233673095703,
|
|
"learning_rate": 9.825287168189818e-06,
|
|
"loss": 1.166,
|
|
"step": 1238
|
|
},
|
|
{
|
|
"epoch": 0.6515908493294768,
|
|
"grad_norm": 2.7419650554656982,
|
|
"learning_rate": 9.824831001666255e-06,
|
|
"loss": 1.1821,
|
|
"step": 1239
|
|
},
|
|
{
|
|
"epoch": 0.6521167499342624,
|
|
"grad_norm": 3.2307729721069336,
|
|
"learning_rate": 9.824374251022069e-06,
|
|
"loss": 1.2171,
|
|
"step": 1240
|
|
},
|
|
{
|
|
"epoch": 0.6526426505390481,
|
|
"grad_norm": 3.0134873390197754,
|
|
"learning_rate": 9.823916916312553e-06,
|
|
"loss": 1.1688,
|
|
"step": 1241
|
|
},
|
|
{
|
|
"epoch": 0.6531685511438338,
|
|
"grad_norm": 2.692688465118408,
|
|
"learning_rate": 9.823458997593076e-06,
|
|
"loss": 1.1905,
|
|
"step": 1242
|
|
},
|
|
{
|
|
"epoch": 0.6536944517486195,
|
|
"grad_norm": 2.909564971923828,
|
|
"learning_rate": 9.823000494919077e-06,
|
|
"loss": 1.286,
|
|
"step": 1243
|
|
},
|
|
{
|
|
"epoch": 0.6542203523534053,
|
|
"grad_norm": 2.6030991077423096,
|
|
"learning_rate": 9.822541408346066e-06,
|
|
"loss": 1.0474,
|
|
"step": 1244
|
|
},
|
|
{
|
|
"epoch": 0.6547462529581909,
|
|
"grad_norm": 2.8486313819885254,
|
|
"learning_rate": 9.82208173792962e-06,
|
|
"loss": 1.2002,
|
|
"step": 1245
|
|
},
|
|
{
|
|
"epoch": 0.6552721535629766,
|
|
"grad_norm": 2.8443713188171387,
|
|
"learning_rate": 9.82162148372539e-06,
|
|
"loss": 1.2134,
|
|
"step": 1246
|
|
},
|
|
{
|
|
"epoch": 0.6557980541677623,
|
|
"grad_norm": 2.943019390106201,
|
|
"learning_rate": 9.821160645789099e-06,
|
|
"loss": 1.2753,
|
|
"step": 1247
|
|
},
|
|
{
|
|
"epoch": 0.656323954772548,
|
|
"grad_norm": 3.293339252471924,
|
|
"learning_rate": 9.820699224176536e-06,
|
|
"loss": 1.3005,
|
|
"step": 1248
|
|
},
|
|
{
|
|
"epoch": 0.6568498553773336,
|
|
"grad_norm": 2.992213249206543,
|
|
"learning_rate": 9.820237218943562e-06,
|
|
"loss": 1.2528,
|
|
"step": 1249
|
|
},
|
|
{
|
|
"epoch": 0.6573757559821194,
|
|
"grad_norm": 2.7967517375946045,
|
|
"learning_rate": 9.819774630146115e-06,
|
|
"loss": 1.2428,
|
|
"step": 1250
|
|
},
|
|
{
|
|
"epoch": 0.657901656586905,
|
|
"grad_norm": 2.9807705879211426,
|
|
"learning_rate": 9.819311457840195e-06,
|
|
"loss": 1.3461,
|
|
"step": 1251
|
|
},
|
|
{
|
|
"epoch": 0.6584275571916908,
|
|
"grad_norm": 2.9345173835754395,
|
|
"learning_rate": 9.818847702081875e-06,
|
|
"loss": 1.233,
|
|
"step": 1252
|
|
},
|
|
{
|
|
"epoch": 0.6589534577964765,
|
|
"grad_norm": 2.913520336151123,
|
|
"learning_rate": 9.818383362927301e-06,
|
|
"loss": 1.1322,
|
|
"step": 1253
|
|
},
|
|
{
|
|
"epoch": 0.6594793584012622,
|
|
"grad_norm": 2.926238536834717,
|
|
"learning_rate": 9.817918440432693e-06,
|
|
"loss": 1.2674,
|
|
"step": 1254
|
|
},
|
|
{
|
|
"epoch": 0.6600052590060479,
|
|
"grad_norm": 2.8500962257385254,
|
|
"learning_rate": 9.81745293465433e-06,
|
|
"loss": 1.2204,
|
|
"step": 1255
|
|
},
|
|
{
|
|
"epoch": 0.6605311596108335,
|
|
"grad_norm": 2.6990396976470947,
|
|
"learning_rate": 9.816986845648573e-06,
|
|
"loss": 1.1902,
|
|
"step": 1256
|
|
},
|
|
{
|
|
"epoch": 0.6610570602156193,
|
|
"grad_norm": 2.7895913124084473,
|
|
"learning_rate": 9.816520173471846e-06,
|
|
"loss": 1.0772,
|
|
"step": 1257
|
|
},
|
|
{
|
|
"epoch": 0.6615829608204049,
|
|
"grad_norm": 2.7517292499542236,
|
|
"learning_rate": 9.816052918180651e-06,
|
|
"loss": 1.132,
|
|
"step": 1258
|
|
},
|
|
{
|
|
"epoch": 0.6621088614251907,
|
|
"grad_norm": 2.650796413421631,
|
|
"learning_rate": 9.815585079831556e-06,
|
|
"loss": 1.2648,
|
|
"step": 1259
|
|
},
|
|
{
|
|
"epoch": 0.6626347620299763,
|
|
"grad_norm": 2.7993175983428955,
|
|
"learning_rate": 9.815116658481195e-06,
|
|
"loss": 1.2211,
|
|
"step": 1260
|
|
},
|
|
{
|
|
"epoch": 0.663160662634762,
|
|
"grad_norm": 2.7765727043151855,
|
|
"learning_rate": 9.814647654186284e-06,
|
|
"loss": 1.2076,
|
|
"step": 1261
|
|
},
|
|
{
|
|
"epoch": 0.6636865632395478,
|
|
"grad_norm": 2.8589956760406494,
|
|
"learning_rate": 9.8141780670036e-06,
|
|
"loss": 1.1855,
|
|
"step": 1262
|
|
},
|
|
{
|
|
"epoch": 0.6642124638443334,
|
|
"grad_norm": 2.747509241104126,
|
|
"learning_rate": 9.813707896989993e-06,
|
|
"loss": 1.1891,
|
|
"step": 1263
|
|
},
|
|
{
|
|
"epoch": 0.6647383644491192,
|
|
"grad_norm": 2.9731147289276123,
|
|
"learning_rate": 9.813237144202386e-06,
|
|
"loss": 1.2268,
|
|
"step": 1264
|
|
},
|
|
{
|
|
"epoch": 0.6652642650539048,
|
|
"grad_norm": 2.821700096130371,
|
|
"learning_rate": 9.81276580869777e-06,
|
|
"loss": 1.1627,
|
|
"step": 1265
|
|
},
|
|
{
|
|
"epoch": 0.6657901656586905,
|
|
"grad_norm": 2.6697123050689697,
|
|
"learning_rate": 9.81229389053321e-06,
|
|
"loss": 1.1357,
|
|
"step": 1266
|
|
},
|
|
{
|
|
"epoch": 0.6663160662634762,
|
|
"grad_norm": 3.072340488433838,
|
|
"learning_rate": 9.811821389765837e-06,
|
|
"loss": 1.2575,
|
|
"step": 1267
|
|
},
|
|
{
|
|
"epoch": 0.6668419668682619,
|
|
"grad_norm": 3.4784977436065674,
|
|
"learning_rate": 9.811348306452854e-06,
|
|
"loss": 1.2855,
|
|
"step": 1268
|
|
},
|
|
{
|
|
"epoch": 0.6673678674730475,
|
|
"grad_norm": 2.715639591217041,
|
|
"learning_rate": 9.810874640651535e-06,
|
|
"loss": 1.2201,
|
|
"step": 1269
|
|
},
|
|
{
|
|
"epoch": 0.6678937680778333,
|
|
"grad_norm": 3.0968172550201416,
|
|
"learning_rate": 9.810400392419226e-06,
|
|
"loss": 1.1911,
|
|
"step": 1270
|
|
},
|
|
{
|
|
"epoch": 0.668419668682619,
|
|
"grad_norm": 2.838432550430298,
|
|
"learning_rate": 9.809925561813341e-06,
|
|
"loss": 1.2458,
|
|
"step": 1271
|
|
},
|
|
{
|
|
"epoch": 0.6689455692874047,
|
|
"grad_norm": 3.4290661811828613,
|
|
"learning_rate": 9.809450148891367e-06,
|
|
"loss": 1.2278,
|
|
"step": 1272
|
|
},
|
|
{
|
|
"epoch": 0.6694714698921904,
|
|
"grad_norm": 2.8974251747131348,
|
|
"learning_rate": 9.808974153710858e-06,
|
|
"loss": 1.2707,
|
|
"step": 1273
|
|
},
|
|
{
|
|
"epoch": 0.669997370496976,
|
|
"grad_norm": 2.876126527786255,
|
|
"learning_rate": 9.808497576329443e-06,
|
|
"loss": 1.295,
|
|
"step": 1274
|
|
},
|
|
{
|
|
"epoch": 0.6705232711017618,
|
|
"grad_norm": 2.706268787384033,
|
|
"learning_rate": 9.808020416804818e-06,
|
|
"loss": 1.2147,
|
|
"step": 1275
|
|
},
|
|
{
|
|
"epoch": 0.6710491717065474,
|
|
"grad_norm": 2.9583985805511475,
|
|
"learning_rate": 9.807542675194749e-06,
|
|
"loss": 1.2328,
|
|
"step": 1276
|
|
},
|
|
{
|
|
"epoch": 0.6715750723113332,
|
|
"grad_norm": 2.669457197189331,
|
|
"learning_rate": 9.807064351557075e-06,
|
|
"loss": 1.1344,
|
|
"step": 1277
|
|
},
|
|
{
|
|
"epoch": 0.6721009729161188,
|
|
"grad_norm": 2.7239530086517334,
|
|
"learning_rate": 9.806585445949705e-06,
|
|
"loss": 1.1877,
|
|
"step": 1278
|
|
},
|
|
{
|
|
"epoch": 0.6726268735209046,
|
|
"grad_norm": 2.735130786895752,
|
|
"learning_rate": 9.80610595843062e-06,
|
|
"loss": 1.2284,
|
|
"step": 1279
|
|
},
|
|
{
|
|
"epoch": 0.6731527741256902,
|
|
"grad_norm": 2.8726375102996826,
|
|
"learning_rate": 9.805625889057867e-06,
|
|
"loss": 1.2294,
|
|
"step": 1280
|
|
},
|
|
{
|
|
"epoch": 0.6736786747304759,
|
|
"grad_norm": 2.8797404766082764,
|
|
"learning_rate": 9.805145237889564e-06,
|
|
"loss": 1.2197,
|
|
"step": 1281
|
|
},
|
|
{
|
|
"epoch": 0.6742045753352617,
|
|
"grad_norm": 2.775703191757202,
|
|
"learning_rate": 9.804664004983904e-06,
|
|
"loss": 1.2112,
|
|
"step": 1282
|
|
},
|
|
{
|
|
"epoch": 0.6747304759400473,
|
|
"grad_norm": 2.8595194816589355,
|
|
"learning_rate": 9.804182190399148e-06,
|
|
"loss": 1.0974,
|
|
"step": 1283
|
|
},
|
|
{
|
|
"epoch": 0.6752563765448331,
|
|
"grad_norm": 2.7802953720092773,
|
|
"learning_rate": 9.803699794193624e-06,
|
|
"loss": 1.1061,
|
|
"step": 1284
|
|
},
|
|
{
|
|
"epoch": 0.6757822771496187,
|
|
"grad_norm": 2.9632604122161865,
|
|
"learning_rate": 9.803216816425739e-06,
|
|
"loss": 1.2302,
|
|
"step": 1285
|
|
},
|
|
{
|
|
"epoch": 0.6763081777544044,
|
|
"grad_norm": 2.938826560974121,
|
|
"learning_rate": 9.80273325715396e-06,
|
|
"loss": 1.2992,
|
|
"step": 1286
|
|
},
|
|
{
|
|
"epoch": 0.6768340783591901,
|
|
"grad_norm": 2.979973077774048,
|
|
"learning_rate": 9.80224911643683e-06,
|
|
"loss": 1.2349,
|
|
"step": 1287
|
|
},
|
|
{
|
|
"epoch": 0.6773599789639758,
|
|
"grad_norm": 2.8141942024230957,
|
|
"learning_rate": 9.801764394332965e-06,
|
|
"loss": 1.1172,
|
|
"step": 1288
|
|
},
|
|
{
|
|
"epoch": 0.6778858795687615,
|
|
"grad_norm": 2.7535274028778076,
|
|
"learning_rate": 9.801279090901043e-06,
|
|
"loss": 1.2199,
|
|
"step": 1289
|
|
},
|
|
{
|
|
"epoch": 0.6784117801735472,
|
|
"grad_norm": 2.8061017990112305,
|
|
"learning_rate": 9.800793206199824e-06,
|
|
"loss": 1.2858,
|
|
"step": 1290
|
|
},
|
|
{
|
|
"epoch": 0.6789376807783329,
|
|
"grad_norm": 2.6226682662963867,
|
|
"learning_rate": 9.800306740288125e-06,
|
|
"loss": 1.1104,
|
|
"step": 1291
|
|
},
|
|
{
|
|
"epoch": 0.6794635813831186,
|
|
"grad_norm": 2.8841917514801025,
|
|
"learning_rate": 9.799819693224847e-06,
|
|
"loss": 1.3023,
|
|
"step": 1292
|
|
},
|
|
{
|
|
"epoch": 0.6799894819879043,
|
|
"grad_norm": 2.8167245388031006,
|
|
"learning_rate": 9.79933206506895e-06,
|
|
"loss": 1.2868,
|
|
"step": 1293
|
|
},
|
|
{
|
|
"epoch": 0.68051538259269,
|
|
"grad_norm": 3.0541136264801025,
|
|
"learning_rate": 9.798843855879467e-06,
|
|
"loss": 1.2844,
|
|
"step": 1294
|
|
},
|
|
{
|
|
"epoch": 0.6810412831974757,
|
|
"grad_norm": 2.958061456680298,
|
|
"learning_rate": 9.79835506571551e-06,
|
|
"loss": 1.2545,
|
|
"step": 1295
|
|
},
|
|
{
|
|
"epoch": 0.6815671838022613,
|
|
"grad_norm": 2.716444730758667,
|
|
"learning_rate": 9.79786569463625e-06,
|
|
"loss": 1.1819,
|
|
"step": 1296
|
|
},
|
|
{
|
|
"epoch": 0.6820930844070471,
|
|
"grad_norm": 2.840081214904785,
|
|
"learning_rate": 9.797375742700935e-06,
|
|
"loss": 1.3657,
|
|
"step": 1297
|
|
},
|
|
{
|
|
"epoch": 0.6826189850118327,
|
|
"grad_norm": 2.6418232917785645,
|
|
"learning_rate": 9.796885209968878e-06,
|
|
"loss": 1.2174,
|
|
"step": 1298
|
|
},
|
|
{
|
|
"epoch": 0.6831448856166185,
|
|
"grad_norm": 2.76863956451416,
|
|
"learning_rate": 9.796394096499469e-06,
|
|
"loss": 1.2722,
|
|
"step": 1299
|
|
},
|
|
{
|
|
"epoch": 0.6836707862214042,
|
|
"grad_norm": 2.7945618629455566,
|
|
"learning_rate": 9.795902402352164e-06,
|
|
"loss": 1.1382,
|
|
"step": 1300
|
|
},
|
|
{
|
|
"epoch": 0.6841966868261898,
|
|
"grad_norm": 2.7365007400512695,
|
|
"learning_rate": 9.79541012758649e-06,
|
|
"loss": 1.1769,
|
|
"step": 1301
|
|
},
|
|
{
|
|
"epoch": 0.6847225874309756,
|
|
"grad_norm": 2.8116114139556885,
|
|
"learning_rate": 9.794917272262044e-06,
|
|
"loss": 1.2609,
|
|
"step": 1302
|
|
},
|
|
{
|
|
"epoch": 0.6852484880357612,
|
|
"grad_norm": 2.843782901763916,
|
|
"learning_rate": 9.794423836438493e-06,
|
|
"loss": 1.2173,
|
|
"step": 1303
|
|
},
|
|
{
|
|
"epoch": 0.685774388640547,
|
|
"grad_norm": 2.8320963382720947,
|
|
"learning_rate": 9.793929820175578e-06,
|
|
"loss": 1.3348,
|
|
"step": 1304
|
|
},
|
|
{
|
|
"epoch": 0.6863002892453326,
|
|
"grad_norm": 2.6727852821350098,
|
|
"learning_rate": 9.793435223533107e-06,
|
|
"loss": 1.1298,
|
|
"step": 1305
|
|
},
|
|
{
|
|
"epoch": 0.6868261898501183,
|
|
"grad_norm": 2.9314067363739014,
|
|
"learning_rate": 9.792940046570956e-06,
|
|
"loss": 1.2818,
|
|
"step": 1306
|
|
},
|
|
{
|
|
"epoch": 0.687352090454904,
|
|
"grad_norm": 3.0059540271759033,
|
|
"learning_rate": 9.792444289349074e-06,
|
|
"loss": 1.294,
|
|
"step": 1307
|
|
},
|
|
{
|
|
"epoch": 0.6878779910596897,
|
|
"grad_norm": 2.6765685081481934,
|
|
"learning_rate": 9.791947951927482e-06,
|
|
"loss": 1.1883,
|
|
"step": 1308
|
|
},
|
|
{
|
|
"epoch": 0.6884038916644755,
|
|
"grad_norm": 2.6198089122772217,
|
|
"learning_rate": 9.791451034366268e-06,
|
|
"loss": 1.2013,
|
|
"step": 1309
|
|
},
|
|
{
|
|
"epoch": 0.6889297922692611,
|
|
"grad_norm": 2.633068323135376,
|
|
"learning_rate": 9.790953536725593e-06,
|
|
"loss": 1.3408,
|
|
"step": 1310
|
|
},
|
|
{
|
|
"epoch": 0.6894556928740468,
|
|
"grad_norm": 2.853569507598877,
|
|
"learning_rate": 9.790455459065685e-06,
|
|
"loss": 1.3197,
|
|
"step": 1311
|
|
},
|
|
{
|
|
"epoch": 0.6899815934788325,
|
|
"grad_norm": 2.6801280975341797,
|
|
"learning_rate": 9.789956801446844e-06,
|
|
"loss": 1.1305,
|
|
"step": 1312
|
|
},
|
|
{
|
|
"epoch": 0.6905074940836182,
|
|
"grad_norm": 2.7111008167266846,
|
|
"learning_rate": 9.789457563929445e-06,
|
|
"loss": 1.1495,
|
|
"step": 1313
|
|
},
|
|
{
|
|
"epoch": 0.6910333946884039,
|
|
"grad_norm": 2.8028833866119385,
|
|
"learning_rate": 9.788957746573921e-06,
|
|
"loss": 1.206,
|
|
"step": 1314
|
|
},
|
|
{
|
|
"epoch": 0.6915592952931896,
|
|
"grad_norm": 2.8011131286621094,
|
|
"learning_rate": 9.788457349440788e-06,
|
|
"loss": 1.2375,
|
|
"step": 1315
|
|
},
|
|
{
|
|
"epoch": 0.6920851958979752,
|
|
"grad_norm": 2.7260706424713135,
|
|
"learning_rate": 9.787956372590625e-06,
|
|
"loss": 1.2261,
|
|
"step": 1316
|
|
},
|
|
{
|
|
"epoch": 0.692611096502761,
|
|
"grad_norm": 2.7182161808013916,
|
|
"learning_rate": 9.787454816084083e-06,
|
|
"loss": 1.1954,
|
|
"step": 1317
|
|
},
|
|
{
|
|
"epoch": 0.6931369971075467,
|
|
"grad_norm": 2.7074167728424072,
|
|
"learning_rate": 9.786952679981882e-06,
|
|
"loss": 1.24,
|
|
"step": 1318
|
|
},
|
|
{
|
|
"epoch": 0.6936628977123324,
|
|
"grad_norm": 2.8799378871917725,
|
|
"learning_rate": 9.786449964344816e-06,
|
|
"loss": 1.214,
|
|
"step": 1319
|
|
},
|
|
{
|
|
"epoch": 0.6941887983171181,
|
|
"grad_norm": 2.7780277729034424,
|
|
"learning_rate": 9.785946669233745e-06,
|
|
"loss": 1.1407,
|
|
"step": 1320
|
|
},
|
|
{
|
|
"epoch": 0.6947146989219037,
|
|
"grad_norm": 2.805628776550293,
|
|
"learning_rate": 9.7854427947096e-06,
|
|
"loss": 1.1857,
|
|
"step": 1321
|
|
},
|
|
{
|
|
"epoch": 0.6952405995266895,
|
|
"grad_norm": 2.9444875717163086,
|
|
"learning_rate": 9.784938340833385e-06,
|
|
"loss": 1.2822,
|
|
"step": 1322
|
|
},
|
|
{
|
|
"epoch": 0.6957665001314751,
|
|
"grad_norm": 2.7424352169036865,
|
|
"learning_rate": 9.78443330766617e-06,
|
|
"loss": 1.2053,
|
|
"step": 1323
|
|
},
|
|
{
|
|
"epoch": 0.6962924007362609,
|
|
"grad_norm": 3.000232458114624,
|
|
"learning_rate": 9.783927695269097e-06,
|
|
"loss": 1.3083,
|
|
"step": 1324
|
|
},
|
|
{
|
|
"epoch": 0.6968183013410465,
|
|
"grad_norm": 2.8363633155822754,
|
|
"learning_rate": 9.783421503703381e-06,
|
|
"loss": 1.1913,
|
|
"step": 1325
|
|
},
|
|
{
|
|
"epoch": 0.6973442019458322,
|
|
"grad_norm": 2.597050905227661,
|
|
"learning_rate": 9.782914733030301e-06,
|
|
"loss": 1.1102,
|
|
"step": 1326
|
|
},
|
|
{
|
|
"epoch": 0.697870102550618,
|
|
"grad_norm": 2.917841911315918,
|
|
"learning_rate": 9.78240738331121e-06,
|
|
"loss": 1.2282,
|
|
"step": 1327
|
|
},
|
|
{
|
|
"epoch": 0.6983960031554036,
|
|
"grad_norm": 2.661543369293213,
|
|
"learning_rate": 9.781899454607534e-06,
|
|
"loss": 1.2251,
|
|
"step": 1328
|
|
},
|
|
{
|
|
"epoch": 0.6989219037601894,
|
|
"grad_norm": 2.6003634929656982,
|
|
"learning_rate": 9.78139094698076e-06,
|
|
"loss": 1.2062,
|
|
"step": 1329
|
|
},
|
|
{
|
|
"epoch": 0.699447804364975,
|
|
"grad_norm": 2.948180675506592,
|
|
"learning_rate": 9.780881860492454e-06,
|
|
"loss": 1.3112,
|
|
"step": 1330
|
|
},
|
|
{
|
|
"epoch": 0.6999737049697607,
|
|
"grad_norm": 2.9805593490600586,
|
|
"learning_rate": 9.78037219520425e-06,
|
|
"loss": 1.266,
|
|
"step": 1331
|
|
},
|
|
{
|
|
"epoch": 0.7004996055745464,
|
|
"grad_norm": 2.769473075866699,
|
|
"learning_rate": 9.77986195117785e-06,
|
|
"loss": 1.088,
|
|
"step": 1332
|
|
},
|
|
{
|
|
"epoch": 0.7010255061793321,
|
|
"grad_norm": 2.8353352546691895,
|
|
"learning_rate": 9.779351128475026e-06,
|
|
"loss": 1.3729,
|
|
"step": 1333
|
|
},
|
|
{
|
|
"epoch": 0.7015514067841178,
|
|
"grad_norm": 2.7563560009002686,
|
|
"learning_rate": 9.77883972715762e-06,
|
|
"loss": 1.1516,
|
|
"step": 1334
|
|
},
|
|
{
|
|
"epoch": 0.7020773073889035,
|
|
"grad_norm": 2.752509355545044,
|
|
"learning_rate": 9.778327747287547e-06,
|
|
"loss": 1.2241,
|
|
"step": 1335
|
|
},
|
|
{
|
|
"epoch": 0.7026032079936891,
|
|
"grad_norm": 2.656015634536743,
|
|
"learning_rate": 9.77781518892679e-06,
|
|
"loss": 1.1736,
|
|
"step": 1336
|
|
},
|
|
{
|
|
"epoch": 0.7031291085984749,
|
|
"grad_norm": 2.8986456394195557,
|
|
"learning_rate": 9.777302052137401e-06,
|
|
"loss": 1.2098,
|
|
"step": 1337
|
|
},
|
|
{
|
|
"epoch": 0.7036550092032606,
|
|
"grad_norm": 3.0297398567199707,
|
|
"learning_rate": 9.776788336981504e-06,
|
|
"loss": 1.1921,
|
|
"step": 1338
|
|
},
|
|
{
|
|
"epoch": 0.7041809098080463,
|
|
"grad_norm": 2.814699411392212,
|
|
"learning_rate": 9.776274043521292e-06,
|
|
"loss": 1.1457,
|
|
"step": 1339
|
|
},
|
|
{
|
|
"epoch": 0.704706810412832,
|
|
"grad_norm": 2.7386231422424316,
|
|
"learning_rate": 9.775759171819029e-06,
|
|
"loss": 1.2069,
|
|
"step": 1340
|
|
},
|
|
{
|
|
"epoch": 0.7052327110176176,
|
|
"grad_norm": 2.724783182144165,
|
|
"learning_rate": 9.775243721937047e-06,
|
|
"loss": 1.2887,
|
|
"step": 1341
|
|
},
|
|
{
|
|
"epoch": 0.7057586116224034,
|
|
"grad_norm": 2.7317190170288086,
|
|
"learning_rate": 9.77472769393775e-06,
|
|
"loss": 1.1309,
|
|
"step": 1342
|
|
},
|
|
{
|
|
"epoch": 0.706284512227189,
|
|
"grad_norm": 2.848374605178833,
|
|
"learning_rate": 9.774211087883609e-06,
|
|
"loss": 1.1525,
|
|
"step": 1343
|
|
},
|
|
{
|
|
"epoch": 0.7068104128319748,
|
|
"grad_norm": 2.824650526046753,
|
|
"learning_rate": 9.77369390383717e-06,
|
|
"loss": 1.1945,
|
|
"step": 1344
|
|
},
|
|
{
|
|
"epoch": 0.7073363134367604,
|
|
"grad_norm": 2.8293139934539795,
|
|
"learning_rate": 9.773176141861047e-06,
|
|
"loss": 1.1254,
|
|
"step": 1345
|
|
},
|
|
{
|
|
"epoch": 0.7078622140415461,
|
|
"grad_norm": 2.7478904724121094,
|
|
"learning_rate": 9.77265780201792e-06,
|
|
"loss": 1.1709,
|
|
"step": 1346
|
|
},
|
|
{
|
|
"epoch": 0.7083881146463319,
|
|
"grad_norm": 2.9537835121154785,
|
|
"learning_rate": 9.772138884370544e-06,
|
|
"loss": 1.3395,
|
|
"step": 1347
|
|
},
|
|
{
|
|
"epoch": 0.7089140152511175,
|
|
"grad_norm": 2.722642421722412,
|
|
"learning_rate": 9.77161938898174e-06,
|
|
"loss": 1.1703,
|
|
"step": 1348
|
|
},
|
|
{
|
|
"epoch": 0.7094399158559033,
|
|
"grad_norm": 2.910426616668701,
|
|
"learning_rate": 9.771099315914403e-06,
|
|
"loss": 1.3255,
|
|
"step": 1349
|
|
},
|
|
{
|
|
"epoch": 0.7099658164606889,
|
|
"grad_norm": 2.807332754135132,
|
|
"learning_rate": 9.770578665231495e-06,
|
|
"loss": 1.2563,
|
|
"step": 1350
|
|
},
|
|
{
|
|
"epoch": 0.7104917170654746,
|
|
"grad_norm": 3.104891300201416,
|
|
"learning_rate": 9.77005743699605e-06,
|
|
"loss": 1.167,
|
|
"step": 1351
|
|
},
|
|
{
|
|
"epoch": 0.7110176176702603,
|
|
"grad_norm": 2.781662702560425,
|
|
"learning_rate": 9.769535631271168e-06,
|
|
"loss": 1.2503,
|
|
"step": 1352
|
|
},
|
|
{
|
|
"epoch": 0.711543518275046,
|
|
"grad_norm": 3.080871105194092,
|
|
"learning_rate": 9.769013248120024e-06,
|
|
"loss": 1.3664,
|
|
"step": 1353
|
|
},
|
|
{
|
|
"epoch": 0.7120694188798317,
|
|
"grad_norm": 2.8637149333953857,
|
|
"learning_rate": 9.768490287605862e-06,
|
|
"loss": 1.167,
|
|
"step": 1354
|
|
},
|
|
{
|
|
"epoch": 0.7125953194846174,
|
|
"grad_norm": 2.672787666320801,
|
|
"learning_rate": 9.767966749791993e-06,
|
|
"loss": 1.1882,
|
|
"step": 1355
|
|
},
|
|
{
|
|
"epoch": 0.7131212200894032,
|
|
"grad_norm": 2.779557228088379,
|
|
"learning_rate": 9.767442634741797e-06,
|
|
"loss": 1.2158,
|
|
"step": 1356
|
|
},
|
|
{
|
|
"epoch": 0.7136471206941888,
|
|
"grad_norm": 2.921419382095337,
|
|
"learning_rate": 9.766917942518729e-06,
|
|
"loss": 1.1433,
|
|
"step": 1357
|
|
},
|
|
{
|
|
"epoch": 0.7141730212989745,
|
|
"grad_norm": 2.7900619506835938,
|
|
"learning_rate": 9.76639267318631e-06,
|
|
"loss": 1.2327,
|
|
"step": 1358
|
|
},
|
|
{
|
|
"epoch": 0.7146989219037602,
|
|
"grad_norm": 2.931241750717163,
|
|
"learning_rate": 9.765866826808133e-06,
|
|
"loss": 1.2261,
|
|
"step": 1359
|
|
},
|
|
{
|
|
"epoch": 0.7152248225085459,
|
|
"grad_norm": 2.788248062133789,
|
|
"learning_rate": 9.765340403447858e-06,
|
|
"loss": 1.1508,
|
|
"step": 1360
|
|
},
|
|
{
|
|
"epoch": 0.7157507231133315,
|
|
"grad_norm": 2.95015025138855,
|
|
"learning_rate": 9.76481340316922e-06,
|
|
"loss": 1.1687,
|
|
"step": 1361
|
|
},
|
|
{
|
|
"epoch": 0.7162766237181173,
|
|
"grad_norm": 2.6803245544433594,
|
|
"learning_rate": 9.764285826036017e-06,
|
|
"loss": 1.142,
|
|
"step": 1362
|
|
},
|
|
{
|
|
"epoch": 0.7168025243229029,
|
|
"grad_norm": 2.8650946617126465,
|
|
"learning_rate": 9.763757672112121e-06,
|
|
"loss": 1.2523,
|
|
"step": 1363
|
|
},
|
|
{
|
|
"epoch": 0.7173284249276887,
|
|
"grad_norm": 2.6279947757720947,
|
|
"learning_rate": 9.763228941461476e-06,
|
|
"loss": 1.1729,
|
|
"step": 1364
|
|
},
|
|
{
|
|
"epoch": 0.7178543255324744,
|
|
"grad_norm": 2.9464049339294434,
|
|
"learning_rate": 9.76269963414809e-06,
|
|
"loss": 1.2874,
|
|
"step": 1365
|
|
},
|
|
{
|
|
"epoch": 0.71838022613726,
|
|
"grad_norm": 2.7368006706237793,
|
|
"learning_rate": 9.762169750236046e-06,
|
|
"loss": 1.1764,
|
|
"step": 1366
|
|
},
|
|
{
|
|
"epoch": 0.7189061267420458,
|
|
"grad_norm": 2.951662540435791,
|
|
"learning_rate": 9.761639289789494e-06,
|
|
"loss": 1.2407,
|
|
"step": 1367
|
|
},
|
|
{
|
|
"epoch": 0.7194320273468314,
|
|
"grad_norm": 2.683938980102539,
|
|
"learning_rate": 9.761108252872653e-06,
|
|
"loss": 1.198,
|
|
"step": 1368
|
|
},
|
|
{
|
|
"epoch": 0.7199579279516172,
|
|
"grad_norm": 2.727053642272949,
|
|
"learning_rate": 9.760576639549815e-06,
|
|
"loss": 1.2477,
|
|
"step": 1369
|
|
},
|
|
{
|
|
"epoch": 0.7204838285564028,
|
|
"grad_norm": 2.6488916873931885,
|
|
"learning_rate": 9.760044449885338e-06,
|
|
"loss": 1.13,
|
|
"step": 1370
|
|
},
|
|
{
|
|
"epoch": 0.7210097291611886,
|
|
"grad_norm": 2.870844841003418,
|
|
"learning_rate": 9.759511683943654e-06,
|
|
"loss": 1.1655,
|
|
"step": 1371
|
|
},
|
|
{
|
|
"epoch": 0.7215356297659742,
|
|
"grad_norm": 2.7343673706054688,
|
|
"learning_rate": 9.758978341789262e-06,
|
|
"loss": 1.214,
|
|
"step": 1372
|
|
},
|
|
{
|
|
"epoch": 0.7220615303707599,
|
|
"grad_norm": 2.781615734100342,
|
|
"learning_rate": 9.758444423486733e-06,
|
|
"loss": 1.1527,
|
|
"step": 1373
|
|
},
|
|
{
|
|
"epoch": 0.7225874309755457,
|
|
"grad_norm": 2.974266529083252,
|
|
"learning_rate": 9.757909929100702e-06,
|
|
"loss": 1.2244,
|
|
"step": 1374
|
|
},
|
|
{
|
|
"epoch": 0.7231133315803313,
|
|
"grad_norm": 2.9403786659240723,
|
|
"learning_rate": 9.757374858695882e-06,
|
|
"loss": 1.277,
|
|
"step": 1375
|
|
},
|
|
{
|
|
"epoch": 0.723639232185117,
|
|
"grad_norm": 2.8673176765441895,
|
|
"learning_rate": 9.756839212337048e-06,
|
|
"loss": 1.1648,
|
|
"step": 1376
|
|
},
|
|
{
|
|
"epoch": 0.7241651327899027,
|
|
"grad_norm": 2.870924949645996,
|
|
"learning_rate": 9.75630299008905e-06,
|
|
"loss": 1.2885,
|
|
"step": 1377
|
|
},
|
|
{
|
|
"epoch": 0.7246910333946884,
|
|
"grad_norm": 2.923553705215454,
|
|
"learning_rate": 9.755766192016807e-06,
|
|
"loss": 1.2213,
|
|
"step": 1378
|
|
},
|
|
{
|
|
"epoch": 0.7252169339994741,
|
|
"grad_norm": 2.7853357791900635,
|
|
"learning_rate": 9.755228818185305e-06,
|
|
"loss": 1.1803,
|
|
"step": 1379
|
|
},
|
|
{
|
|
"epoch": 0.7257428346042598,
|
|
"grad_norm": 2.8394417762756348,
|
|
"learning_rate": 9.754690868659603e-06,
|
|
"loss": 1.2191,
|
|
"step": 1380
|
|
},
|
|
{
|
|
"epoch": 0.7262687352090454,
|
|
"grad_norm": 3.001978635787964,
|
|
"learning_rate": 9.754152343504828e-06,
|
|
"loss": 1.2483,
|
|
"step": 1381
|
|
},
|
|
{
|
|
"epoch": 0.7267946358138312,
|
|
"grad_norm": 3.209522247314453,
|
|
"learning_rate": 9.753613242786173e-06,
|
|
"loss": 1.2995,
|
|
"step": 1382
|
|
},
|
|
{
|
|
"epoch": 0.7273205364186169,
|
|
"grad_norm": 2.718196153640747,
|
|
"learning_rate": 9.75307356656891e-06,
|
|
"loss": 1.1526,
|
|
"step": 1383
|
|
},
|
|
{
|
|
"epoch": 0.7278464370234026,
|
|
"grad_norm": 2.802798271179199,
|
|
"learning_rate": 9.752533314918373e-06,
|
|
"loss": 1.2404,
|
|
"step": 1384
|
|
},
|
|
{
|
|
"epoch": 0.7283723376281883,
|
|
"grad_norm": 2.9258315563201904,
|
|
"learning_rate": 9.751992487899967e-06,
|
|
"loss": 1.2361,
|
|
"step": 1385
|
|
},
|
|
{
|
|
"epoch": 0.728898238232974,
|
|
"grad_norm": 2.7048518657684326,
|
|
"learning_rate": 9.751451085579168e-06,
|
|
"loss": 1.1471,
|
|
"step": 1386
|
|
},
|
|
{
|
|
"epoch": 0.7294241388377597,
|
|
"grad_norm": 2.7629144191741943,
|
|
"learning_rate": 9.750909108021522e-06,
|
|
"loss": 1.1752,
|
|
"step": 1387
|
|
},
|
|
{
|
|
"epoch": 0.7299500394425453,
|
|
"grad_norm": 2.7664830684661865,
|
|
"learning_rate": 9.750366555292641e-06,
|
|
"loss": 1.088,
|
|
"step": 1388
|
|
},
|
|
{
|
|
"epoch": 0.7304759400473311,
|
|
"grad_norm": 3.139861583709717,
|
|
"learning_rate": 9.749823427458213e-06,
|
|
"loss": 1.2183,
|
|
"step": 1389
|
|
},
|
|
{
|
|
"epoch": 0.7310018406521167,
|
|
"grad_norm": 2.8266355991363525,
|
|
"learning_rate": 9.749279724583992e-06,
|
|
"loss": 1.2911,
|
|
"step": 1390
|
|
},
|
|
{
|
|
"epoch": 0.7315277412569025,
|
|
"grad_norm": 2.8797991275787354,
|
|
"learning_rate": 9.748735446735798e-06,
|
|
"loss": 1.2236,
|
|
"step": 1391
|
|
},
|
|
{
|
|
"epoch": 0.7320536418616881,
|
|
"grad_norm": 2.8501455783843994,
|
|
"learning_rate": 9.748190593979527e-06,
|
|
"loss": 1.2196,
|
|
"step": 1392
|
|
},
|
|
{
|
|
"epoch": 0.7325795424664738,
|
|
"grad_norm": 2.8247628211975098,
|
|
"learning_rate": 9.747645166381143e-06,
|
|
"loss": 1.2005,
|
|
"step": 1393
|
|
},
|
|
{
|
|
"epoch": 0.7331054430712596,
|
|
"grad_norm": 2.8893861770629883,
|
|
"learning_rate": 9.747099164006674e-06,
|
|
"loss": 1.2212,
|
|
"step": 1394
|
|
},
|
|
{
|
|
"epoch": 0.7336313436760452,
|
|
"grad_norm": 2.80829119682312,
|
|
"learning_rate": 9.746552586922227e-06,
|
|
"loss": 1.2218,
|
|
"step": 1395
|
|
},
|
|
{
|
|
"epoch": 0.734157244280831,
|
|
"grad_norm": 3.064228057861328,
|
|
"learning_rate": 9.746005435193971e-06,
|
|
"loss": 1.2473,
|
|
"step": 1396
|
|
},
|
|
{
|
|
"epoch": 0.7346831448856166,
|
|
"grad_norm": 2.7381856441497803,
|
|
"learning_rate": 9.74545770888815e-06,
|
|
"loss": 1.2446,
|
|
"step": 1397
|
|
},
|
|
{
|
|
"epoch": 0.7352090454904023,
|
|
"grad_norm": 3.0347237586975098,
|
|
"learning_rate": 9.744909408071069e-06,
|
|
"loss": 1.2639,
|
|
"step": 1398
|
|
},
|
|
{
|
|
"epoch": 0.735734946095188,
|
|
"grad_norm": 2.8705241680145264,
|
|
"learning_rate": 9.744360532809111e-06,
|
|
"loss": 1.2286,
|
|
"step": 1399
|
|
},
|
|
{
|
|
"epoch": 0.7362608466999737,
|
|
"grad_norm": 2.863595962524414,
|
|
"learning_rate": 9.74381108316873e-06,
|
|
"loss": 1.1809,
|
|
"step": 1400
|
|
},
|
|
{
|
|
"epoch": 0.7367867473047593,
|
|
"grad_norm": 2.6183600425720215,
|
|
"learning_rate": 9.74326105921644e-06,
|
|
"loss": 1.1218,
|
|
"step": 1401
|
|
},
|
|
{
|
|
"epoch": 0.7373126479095451,
|
|
"grad_norm": 2.905881881713867,
|
|
"learning_rate": 9.742710461018833e-06,
|
|
"loss": 1.2069,
|
|
"step": 1402
|
|
},
|
|
{
|
|
"epoch": 0.7378385485143308,
|
|
"grad_norm": 2.8619956970214844,
|
|
"learning_rate": 9.742159288642568e-06,
|
|
"loss": 1.2052,
|
|
"step": 1403
|
|
},
|
|
{
|
|
"epoch": 0.7383644491191165,
|
|
"grad_norm": 2.9845991134643555,
|
|
"learning_rate": 9.741607542154369e-06,
|
|
"loss": 1.2573,
|
|
"step": 1404
|
|
},
|
|
{
|
|
"epoch": 0.7388903497239022,
|
|
"grad_norm": 2.82490873336792,
|
|
"learning_rate": 9.741055221621037e-06,
|
|
"loss": 1.1591,
|
|
"step": 1405
|
|
},
|
|
{
|
|
"epoch": 0.7394162503286879,
|
|
"grad_norm": 2.853015899658203,
|
|
"learning_rate": 9.740502327109436e-06,
|
|
"loss": 1.225,
|
|
"step": 1406
|
|
},
|
|
{
|
|
"epoch": 0.7399421509334736,
|
|
"grad_norm": 2.931688070297241,
|
|
"learning_rate": 9.739948858686507e-06,
|
|
"loss": 1.2393,
|
|
"step": 1407
|
|
},
|
|
{
|
|
"epoch": 0.7404680515382592,
|
|
"grad_norm": 2.880453586578369,
|
|
"learning_rate": 9.739394816419253e-06,
|
|
"loss": 1.2787,
|
|
"step": 1408
|
|
},
|
|
{
|
|
"epoch": 0.740993952143045,
|
|
"grad_norm": 3.198512554168701,
|
|
"learning_rate": 9.738840200374748e-06,
|
|
"loss": 1.2672,
|
|
"step": 1409
|
|
},
|
|
{
|
|
"epoch": 0.7415198527478306,
|
|
"grad_norm": 2.791623115539551,
|
|
"learning_rate": 9.738285010620138e-06,
|
|
"loss": 1.227,
|
|
"step": 1410
|
|
},
|
|
{
|
|
"epoch": 0.7420457533526164,
|
|
"grad_norm": 2.815396308898926,
|
|
"learning_rate": 9.73772924722264e-06,
|
|
"loss": 1.2609,
|
|
"step": 1411
|
|
},
|
|
{
|
|
"epoch": 0.7425716539574021,
|
|
"grad_norm": 3.077667236328125,
|
|
"learning_rate": 9.737172910249534e-06,
|
|
"loss": 1.3137,
|
|
"step": 1412
|
|
},
|
|
{
|
|
"epoch": 0.7430975545621877,
|
|
"grad_norm": 2.688175678253174,
|
|
"learning_rate": 9.736615999768177e-06,
|
|
"loss": 1.1644,
|
|
"step": 1413
|
|
},
|
|
{
|
|
"epoch": 0.7436234551669735,
|
|
"grad_norm": 2.9544355869293213,
|
|
"learning_rate": 9.736058515845987e-06,
|
|
"loss": 1.182,
|
|
"step": 1414
|
|
},
|
|
{
|
|
"epoch": 0.7441493557717591,
|
|
"grad_norm": 2.9537580013275146,
|
|
"learning_rate": 9.73550045855046e-06,
|
|
"loss": 1.2033,
|
|
"step": 1415
|
|
},
|
|
{
|
|
"epoch": 0.7446752563765449,
|
|
"grad_norm": 2.711751937866211,
|
|
"learning_rate": 9.734941827949155e-06,
|
|
"loss": 1.2172,
|
|
"step": 1416
|
|
},
|
|
{
|
|
"epoch": 0.7452011569813305,
|
|
"grad_norm": 2.7822883129119873,
|
|
"learning_rate": 9.734382624109707e-06,
|
|
"loss": 1.1815,
|
|
"step": 1417
|
|
},
|
|
{
|
|
"epoch": 0.7457270575861162,
|
|
"grad_norm": 2.6219594478607178,
|
|
"learning_rate": 9.73382284709981e-06,
|
|
"loss": 1.1533,
|
|
"step": 1418
|
|
},
|
|
{
|
|
"epoch": 0.7462529581909019,
|
|
"grad_norm": 2.690303325653076,
|
|
"learning_rate": 9.733262496987237e-06,
|
|
"loss": 1.1117,
|
|
"step": 1419
|
|
},
|
|
{
|
|
"epoch": 0.7467788587956876,
|
|
"grad_norm": 2.8884165287017822,
|
|
"learning_rate": 9.73270157383983e-06,
|
|
"loss": 1.2597,
|
|
"step": 1420
|
|
},
|
|
{
|
|
"epoch": 0.7473047594004734,
|
|
"grad_norm": 2.6800577640533447,
|
|
"learning_rate": 9.732140077725492e-06,
|
|
"loss": 1.1863,
|
|
"step": 1421
|
|
},
|
|
{
|
|
"epoch": 0.747830660005259,
|
|
"grad_norm": 2.7506210803985596,
|
|
"learning_rate": 9.731578008712203e-06,
|
|
"loss": 1.2551,
|
|
"step": 1422
|
|
},
|
|
{
|
|
"epoch": 0.7483565606100447,
|
|
"grad_norm": 2.542346239089966,
|
|
"learning_rate": 9.731015366868013e-06,
|
|
"loss": 1.237,
|
|
"step": 1423
|
|
},
|
|
{
|
|
"epoch": 0.7488824612148304,
|
|
"grad_norm": 2.747251272201538,
|
|
"learning_rate": 9.730452152261035e-06,
|
|
"loss": 1.3099,
|
|
"step": 1424
|
|
},
|
|
{
|
|
"epoch": 0.7494083618196161,
|
|
"grad_norm": 3.0582525730133057,
|
|
"learning_rate": 9.729888364959455e-06,
|
|
"loss": 1.2324,
|
|
"step": 1425
|
|
},
|
|
{
|
|
"epoch": 0.7499342624244018,
|
|
"grad_norm": 2.754293918609619,
|
|
"learning_rate": 9.729324005031529e-06,
|
|
"loss": 1.2196,
|
|
"step": 1426
|
|
},
|
|
{
|
|
"epoch": 0.7504601630291875,
|
|
"grad_norm": 2.7300713062286377,
|
|
"learning_rate": 9.728759072545582e-06,
|
|
"loss": 1.2502,
|
|
"step": 1427
|
|
},
|
|
{
|
|
"epoch": 0.7509860636339731,
|
|
"grad_norm": 2.6909639835357666,
|
|
"learning_rate": 9.728193567570007e-06,
|
|
"loss": 1.2159,
|
|
"step": 1428
|
|
},
|
|
{
|
|
"epoch": 0.7515119642387589,
|
|
"grad_norm": 2.814767837524414,
|
|
"learning_rate": 9.727627490173268e-06,
|
|
"loss": 1.1637,
|
|
"step": 1429
|
|
},
|
|
{
|
|
"epoch": 0.7520378648435446,
|
|
"grad_norm": 2.699333906173706,
|
|
"learning_rate": 9.727060840423896e-06,
|
|
"loss": 1.1854,
|
|
"step": 1430
|
|
},
|
|
{
|
|
"epoch": 0.7525637654483303,
|
|
"grad_norm": 2.8222222328186035,
|
|
"learning_rate": 9.726493618390496e-06,
|
|
"loss": 1.1504,
|
|
"step": 1431
|
|
},
|
|
{
|
|
"epoch": 0.753089666053116,
|
|
"grad_norm": 2.8420610427856445,
|
|
"learning_rate": 9.725925824141736e-06,
|
|
"loss": 1.1877,
|
|
"step": 1432
|
|
},
|
|
{
|
|
"epoch": 0.7536155666579016,
|
|
"grad_norm": 2.959024429321289,
|
|
"learning_rate": 9.725357457746357e-06,
|
|
"loss": 1.1795,
|
|
"step": 1433
|
|
},
|
|
{
|
|
"epoch": 0.7541414672626874,
|
|
"grad_norm": 2.6873838901519775,
|
|
"learning_rate": 9.724788519273166e-06,
|
|
"loss": 1.2036,
|
|
"step": 1434
|
|
},
|
|
{
|
|
"epoch": 0.754667367867473,
|
|
"grad_norm": 2.7125673294067383,
|
|
"learning_rate": 9.724219008791048e-06,
|
|
"loss": 1.3133,
|
|
"step": 1435
|
|
},
|
|
{
|
|
"epoch": 0.7551932684722588,
|
|
"grad_norm": 3.0049500465393066,
|
|
"learning_rate": 9.723648926368944e-06,
|
|
"loss": 1.3039,
|
|
"step": 1436
|
|
},
|
|
{
|
|
"epoch": 0.7557191690770444,
|
|
"grad_norm": 2.754687547683716,
|
|
"learning_rate": 9.723078272075878e-06,
|
|
"loss": 1.2635,
|
|
"step": 1437
|
|
},
|
|
{
|
|
"epoch": 0.7562450696818301,
|
|
"grad_norm": 2.6424946784973145,
|
|
"learning_rate": 9.722507045980934e-06,
|
|
"loss": 1.1939,
|
|
"step": 1438
|
|
},
|
|
{
|
|
"epoch": 0.7567709702866159,
|
|
"grad_norm": 2.7751216888427734,
|
|
"learning_rate": 9.721935248153266e-06,
|
|
"loss": 1.1956,
|
|
"step": 1439
|
|
},
|
|
{
|
|
"epoch": 0.7572968708914015,
|
|
"grad_norm": 2.7871320247650146,
|
|
"learning_rate": 9.7213628786621e-06,
|
|
"loss": 1.1799,
|
|
"step": 1440
|
|
},
|
|
{
|
|
"epoch": 0.7578227714961873,
|
|
"grad_norm": 2.936032295227051,
|
|
"learning_rate": 9.720789937576729e-06,
|
|
"loss": 1.2394,
|
|
"step": 1441
|
|
},
|
|
{
|
|
"epoch": 0.7583486721009729,
|
|
"grad_norm": 2.86445951461792,
|
|
"learning_rate": 9.72021642496652e-06,
|
|
"loss": 1.1907,
|
|
"step": 1442
|
|
},
|
|
{
|
|
"epoch": 0.7588745727057586,
|
|
"grad_norm": 2.8147592544555664,
|
|
"learning_rate": 9.719642340900902e-06,
|
|
"loss": 1.2008,
|
|
"step": 1443
|
|
},
|
|
{
|
|
"epoch": 0.7594004733105443,
|
|
"grad_norm": 2.983302593231201,
|
|
"learning_rate": 9.71906768544938e-06,
|
|
"loss": 1.2931,
|
|
"step": 1444
|
|
},
|
|
{
|
|
"epoch": 0.75992637391533,
|
|
"grad_norm": 2.762317419052124,
|
|
"learning_rate": 9.71849245868152e-06,
|
|
"loss": 1.2494,
|
|
"step": 1445
|
|
},
|
|
{
|
|
"epoch": 0.7604522745201157,
|
|
"grad_norm": 2.7992143630981445,
|
|
"learning_rate": 9.717916660666969e-06,
|
|
"loss": 1.228,
|
|
"step": 1446
|
|
},
|
|
{
|
|
"epoch": 0.7609781751249014,
|
|
"grad_norm": 2.9557297229766846,
|
|
"learning_rate": 9.71734029147543e-06,
|
|
"loss": 1.2652,
|
|
"step": 1447
|
|
},
|
|
{
|
|
"epoch": 0.761504075729687,
|
|
"grad_norm": 2.9317028522491455,
|
|
"learning_rate": 9.716763351176684e-06,
|
|
"loss": 1.3598,
|
|
"step": 1448
|
|
},
|
|
{
|
|
"epoch": 0.7620299763344728,
|
|
"grad_norm": 2.808631181716919,
|
|
"learning_rate": 9.71618583984058e-06,
|
|
"loss": 1.2328,
|
|
"step": 1449
|
|
},
|
|
{
|
|
"epoch": 0.7625558769392585,
|
|
"grad_norm": 2.7970492839813232,
|
|
"learning_rate": 9.715607757537031e-06,
|
|
"loss": 1.2694,
|
|
"step": 1450
|
|
},
|
|
{
|
|
"epoch": 0.7630817775440442,
|
|
"grad_norm": 2.7287187576293945,
|
|
"learning_rate": 9.715029104336028e-06,
|
|
"loss": 1.2151,
|
|
"step": 1451
|
|
},
|
|
{
|
|
"epoch": 0.7636076781488299,
|
|
"grad_norm": 2.716937303543091,
|
|
"learning_rate": 9.714449880307622e-06,
|
|
"loss": 1.2343,
|
|
"step": 1452
|
|
},
|
|
{
|
|
"epoch": 0.7641335787536155,
|
|
"grad_norm": 2.7608489990234375,
|
|
"learning_rate": 9.713870085521937e-06,
|
|
"loss": 1.2579,
|
|
"step": 1453
|
|
},
|
|
{
|
|
"epoch": 0.7646594793584013,
|
|
"grad_norm": 2.9985344409942627,
|
|
"learning_rate": 9.713289720049168e-06,
|
|
"loss": 1.3053,
|
|
"step": 1454
|
|
},
|
|
{
|
|
"epoch": 0.7651853799631869,
|
|
"grad_norm": 2.9629762172698975,
|
|
"learning_rate": 9.712708783959577e-06,
|
|
"loss": 1.2267,
|
|
"step": 1455
|
|
},
|
|
{
|
|
"epoch": 0.7657112805679727,
|
|
"grad_norm": 2.689018487930298,
|
|
"learning_rate": 9.712127277323493e-06,
|
|
"loss": 1.2154,
|
|
"step": 1456
|
|
},
|
|
{
|
|
"epoch": 0.7662371811727583,
|
|
"grad_norm": 2.7860639095306396,
|
|
"learning_rate": 9.71154520021132e-06,
|
|
"loss": 1.2777,
|
|
"step": 1457
|
|
},
|
|
{
|
|
"epoch": 0.766763081777544,
|
|
"grad_norm": 2.7623190879821777,
|
|
"learning_rate": 9.710962552693525e-06,
|
|
"loss": 1.1838,
|
|
"step": 1458
|
|
},
|
|
{
|
|
"epoch": 0.7672889823823298,
|
|
"grad_norm": 2.714494228363037,
|
|
"learning_rate": 9.710379334840647e-06,
|
|
"loss": 1.144,
|
|
"step": 1459
|
|
},
|
|
{
|
|
"epoch": 0.7678148829871154,
|
|
"grad_norm": 2.842541217803955,
|
|
"learning_rate": 9.709795546723294e-06,
|
|
"loss": 1.2005,
|
|
"step": 1460
|
|
},
|
|
{
|
|
"epoch": 0.7683407835919012,
|
|
"grad_norm": 2.7892494201660156,
|
|
"learning_rate": 9.709211188412142e-06,
|
|
"loss": 1.1651,
|
|
"step": 1461
|
|
},
|
|
{
|
|
"epoch": 0.7688666841966868,
|
|
"grad_norm": 2.9375109672546387,
|
|
"learning_rate": 9.708626259977937e-06,
|
|
"loss": 1.3047,
|
|
"step": 1462
|
|
},
|
|
{
|
|
"epoch": 0.7693925848014725,
|
|
"grad_norm": 3.0857961177825928,
|
|
"learning_rate": 9.708040761491493e-06,
|
|
"loss": 1.1145,
|
|
"step": 1463
|
|
},
|
|
{
|
|
"epoch": 0.7699184854062582,
|
|
"grad_norm": 2.8779168128967285,
|
|
"learning_rate": 9.707454693023696e-06,
|
|
"loss": 1.221,
|
|
"step": 1464
|
|
},
|
|
{
|
|
"epoch": 0.7704443860110439,
|
|
"grad_norm": 2.6664581298828125,
|
|
"learning_rate": 9.706868054645496e-06,
|
|
"loss": 1.0693,
|
|
"step": 1465
|
|
},
|
|
{
|
|
"epoch": 0.7709702866158296,
|
|
"grad_norm": 2.7917699813842773,
|
|
"learning_rate": 9.706280846427916e-06,
|
|
"loss": 1.1572,
|
|
"step": 1466
|
|
},
|
|
{
|
|
"epoch": 0.7714961872206153,
|
|
"grad_norm": 2.7519021034240723,
|
|
"learning_rate": 9.705693068442046e-06,
|
|
"loss": 1.1943,
|
|
"step": 1467
|
|
},
|
|
{
|
|
"epoch": 0.772022087825401,
|
|
"grad_norm": 2.8214452266693115,
|
|
"learning_rate": 9.705104720759046e-06,
|
|
"loss": 1.2514,
|
|
"step": 1468
|
|
},
|
|
{
|
|
"epoch": 0.7725479884301867,
|
|
"grad_norm": 2.6991899013519287,
|
|
"learning_rate": 9.704515803450145e-06,
|
|
"loss": 1.2737,
|
|
"step": 1469
|
|
},
|
|
{
|
|
"epoch": 0.7730738890349724,
|
|
"grad_norm": 2.701305866241455,
|
|
"learning_rate": 9.703926316586641e-06,
|
|
"loss": 1.3114,
|
|
"step": 1470
|
|
},
|
|
{
|
|
"epoch": 0.7735997896397581,
|
|
"grad_norm": 2.5843467712402344,
|
|
"learning_rate": 9.703336260239898e-06,
|
|
"loss": 1.222,
|
|
"step": 1471
|
|
},
|
|
{
|
|
"epoch": 0.7741256902445438,
|
|
"grad_norm": 2.744755744934082,
|
|
"learning_rate": 9.702745634481351e-06,
|
|
"loss": 1.2264,
|
|
"step": 1472
|
|
},
|
|
{
|
|
"epoch": 0.7746515908493294,
|
|
"grad_norm": 2.678034543991089,
|
|
"learning_rate": 9.702154439382511e-06,
|
|
"loss": 1.0706,
|
|
"step": 1473
|
|
},
|
|
{
|
|
"epoch": 0.7751774914541152,
|
|
"grad_norm": 2.779024124145508,
|
|
"learning_rate": 9.701562675014945e-06,
|
|
"loss": 1.3297,
|
|
"step": 1474
|
|
},
|
|
{
|
|
"epoch": 0.7757033920589008,
|
|
"grad_norm": 2.9749972820281982,
|
|
"learning_rate": 9.700970341450298e-06,
|
|
"loss": 1.207,
|
|
"step": 1475
|
|
},
|
|
{
|
|
"epoch": 0.7762292926636866,
|
|
"grad_norm": 2.751960039138794,
|
|
"learning_rate": 9.700377438760278e-06,
|
|
"loss": 1.225,
|
|
"step": 1476
|
|
},
|
|
{
|
|
"epoch": 0.7767551932684723,
|
|
"grad_norm": 2.8523972034454346,
|
|
"learning_rate": 9.699783967016669e-06,
|
|
"loss": 1.2835,
|
|
"step": 1477
|
|
},
|
|
{
|
|
"epoch": 0.7772810938732579,
|
|
"grad_norm": 3.0514094829559326,
|
|
"learning_rate": 9.69918992629132e-06,
|
|
"loss": 1.258,
|
|
"step": 1478
|
|
},
|
|
{
|
|
"epoch": 0.7778069944780437,
|
|
"grad_norm": 2.8080227375030518,
|
|
"learning_rate": 9.698595316656146e-06,
|
|
"loss": 1.1644,
|
|
"step": 1479
|
|
},
|
|
{
|
|
"epoch": 0.7783328950828293,
|
|
"grad_norm": 2.838542938232422,
|
|
"learning_rate": 9.698000138183132e-06,
|
|
"loss": 1.2538,
|
|
"step": 1480
|
|
},
|
|
{
|
|
"epoch": 0.7788587956876151,
|
|
"grad_norm": 2.6626014709472656,
|
|
"learning_rate": 9.69740439094434e-06,
|
|
"loss": 1.2139,
|
|
"step": 1481
|
|
},
|
|
{
|
|
"epoch": 0.7793846962924007,
|
|
"grad_norm": 2.8611414432525635,
|
|
"learning_rate": 9.69680807501189e-06,
|
|
"loss": 1.2703,
|
|
"step": 1482
|
|
},
|
|
{
|
|
"epoch": 0.7799105968971864,
|
|
"grad_norm": 2.6769509315490723,
|
|
"learning_rate": 9.696211190457976e-06,
|
|
"loss": 1.245,
|
|
"step": 1483
|
|
},
|
|
{
|
|
"epoch": 0.7804364975019721,
|
|
"grad_norm": 2.861598014831543,
|
|
"learning_rate": 9.69561373735486e-06,
|
|
"loss": 1.1994,
|
|
"step": 1484
|
|
},
|
|
{
|
|
"epoch": 0.7809623981067578,
|
|
"grad_norm": 2.820357084274292,
|
|
"learning_rate": 9.695015715774874e-06,
|
|
"loss": 1.1305,
|
|
"step": 1485
|
|
},
|
|
{
|
|
"epoch": 0.7814882987115436,
|
|
"grad_norm": 2.6703267097473145,
|
|
"learning_rate": 9.694417125790416e-06,
|
|
"loss": 1.1189,
|
|
"step": 1486
|
|
},
|
|
{
|
|
"epoch": 0.7820141993163292,
|
|
"grad_norm": 2.812563896179199,
|
|
"learning_rate": 9.693817967473958e-06,
|
|
"loss": 1.17,
|
|
"step": 1487
|
|
},
|
|
{
|
|
"epoch": 0.782540099921115,
|
|
"grad_norm": 2.8793160915374756,
|
|
"learning_rate": 9.693218240898032e-06,
|
|
"loss": 1.3207,
|
|
"step": 1488
|
|
},
|
|
{
|
|
"epoch": 0.7830660005259006,
|
|
"grad_norm": 2.8847970962524414,
|
|
"learning_rate": 9.69261794613525e-06,
|
|
"loss": 1.2568,
|
|
"step": 1489
|
|
},
|
|
{
|
|
"epoch": 0.7835919011306863,
|
|
"grad_norm": 2.8020126819610596,
|
|
"learning_rate": 9.692017083258283e-06,
|
|
"loss": 1.2622,
|
|
"step": 1490
|
|
},
|
|
{
|
|
"epoch": 0.784117801735472,
|
|
"grad_norm": 2.8324670791625977,
|
|
"learning_rate": 9.691415652339878e-06,
|
|
"loss": 1.2896,
|
|
"step": 1491
|
|
},
|
|
{
|
|
"epoch": 0.7846437023402577,
|
|
"grad_norm": 3.024817705154419,
|
|
"learning_rate": 9.690813653452842e-06,
|
|
"loss": 1.228,
|
|
"step": 1492
|
|
},
|
|
{
|
|
"epoch": 0.7851696029450433,
|
|
"grad_norm": 2.6857359409332275,
|
|
"learning_rate": 9.690211086670062e-06,
|
|
"loss": 1.1288,
|
|
"step": 1493
|
|
},
|
|
{
|
|
"epoch": 0.7856955035498291,
|
|
"grad_norm": 3.032203435897827,
|
|
"learning_rate": 9.689607952064486e-06,
|
|
"loss": 1.2762,
|
|
"step": 1494
|
|
},
|
|
{
|
|
"epoch": 0.7862214041546147,
|
|
"grad_norm": 2.770153522491455,
|
|
"learning_rate": 9.689004249709134e-06,
|
|
"loss": 1.1849,
|
|
"step": 1495
|
|
},
|
|
{
|
|
"epoch": 0.7867473047594005,
|
|
"grad_norm": 2.8623955249786377,
|
|
"learning_rate": 9.68839997967709e-06,
|
|
"loss": 1.2389,
|
|
"step": 1496
|
|
},
|
|
{
|
|
"epoch": 0.7872732053641862,
|
|
"grad_norm": 2.812040328979492,
|
|
"learning_rate": 9.687795142041513e-06,
|
|
"loss": 1.1651,
|
|
"step": 1497
|
|
},
|
|
{
|
|
"epoch": 0.7877991059689718,
|
|
"grad_norm": 2.7999038696289062,
|
|
"learning_rate": 9.687189736875629e-06,
|
|
"loss": 1.2087,
|
|
"step": 1498
|
|
},
|
|
{
|
|
"epoch": 0.7883250065737576,
|
|
"grad_norm": 2.9131109714508057,
|
|
"learning_rate": 9.686583764252729e-06,
|
|
"loss": 1.2183,
|
|
"step": 1499
|
|
},
|
|
{
|
|
"epoch": 0.7888509071785432,
|
|
"grad_norm": 2.697418689727783,
|
|
"learning_rate": 9.685977224246176e-06,
|
|
"loss": 1.2039,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"epoch": 0.7888509071785432,
|
|
"eval_loss": 1.2435553073883057,
|
|
"eval_runtime": 8.8613,
|
|
"eval_samples_per_second": 45.14,
|
|
"eval_steps_per_second": 2.821,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"epoch": 0.789376807783329,
|
|
"grad_norm": 2.8642144203186035,
|
|
"learning_rate": 9.685370116929404e-06,
|
|
"loss": 1.1528,
|
|
"step": 1501
|
|
},
|
|
{
|
|
"epoch": 0.7899027083881146,
|
|
"grad_norm": 2.710876703262329,
|
|
"learning_rate": 9.68476244237591e-06,
|
|
"loss": 1.1756,
|
|
"step": 1502
|
|
},
|
|
{
|
|
"epoch": 0.7904286089929004,
|
|
"grad_norm": 3.0552873611450195,
|
|
"learning_rate": 9.684154200659261e-06,
|
|
"loss": 1.2883,
|
|
"step": 1503
|
|
},
|
|
{
|
|
"epoch": 0.790954509597686,
|
|
"grad_norm": 2.713456153869629,
|
|
"learning_rate": 9.683545391853097e-06,
|
|
"loss": 1.2375,
|
|
"step": 1504
|
|
},
|
|
{
|
|
"epoch": 0.7914804102024717,
|
|
"grad_norm": 2.8772003650665283,
|
|
"learning_rate": 9.682936016031122e-06,
|
|
"loss": 1.2263,
|
|
"step": 1505
|
|
},
|
|
{
|
|
"epoch": 0.7920063108072575,
|
|
"grad_norm": 2.7407686710357666,
|
|
"learning_rate": 9.682326073267111e-06,
|
|
"loss": 1.2547,
|
|
"step": 1506
|
|
},
|
|
{
|
|
"epoch": 0.7925322114120431,
|
|
"grad_norm": 2.675150156021118,
|
|
"learning_rate": 9.681715563634909e-06,
|
|
"loss": 1.1739,
|
|
"step": 1507
|
|
},
|
|
{
|
|
"epoch": 0.7930581120168289,
|
|
"grad_norm": 2.896423816680908,
|
|
"learning_rate": 9.681104487208425e-06,
|
|
"loss": 1.2873,
|
|
"step": 1508
|
|
},
|
|
{
|
|
"epoch": 0.7935840126216145,
|
|
"grad_norm": 2.7748866081237793,
|
|
"learning_rate": 9.68049284406164e-06,
|
|
"loss": 1.2119,
|
|
"step": 1509
|
|
},
|
|
{
|
|
"epoch": 0.7941099132264002,
|
|
"grad_norm": 3.022756338119507,
|
|
"learning_rate": 9.679880634268602e-06,
|
|
"loss": 1.1149,
|
|
"step": 1510
|
|
},
|
|
{
|
|
"epoch": 0.7946358138311859,
|
|
"grad_norm": 2.7284884452819824,
|
|
"learning_rate": 9.679267857903431e-06,
|
|
"loss": 1.2136,
|
|
"step": 1511
|
|
},
|
|
{
|
|
"epoch": 0.7951617144359716,
|
|
"grad_norm": 2.794715642929077,
|
|
"learning_rate": 9.678654515040312e-06,
|
|
"loss": 1.191,
|
|
"step": 1512
|
|
},
|
|
{
|
|
"epoch": 0.7956876150407572,
|
|
"grad_norm": 2.8923885822296143,
|
|
"learning_rate": 9.678040605753497e-06,
|
|
"loss": 1.2394,
|
|
"step": 1513
|
|
},
|
|
{
|
|
"epoch": 0.796213515645543,
|
|
"grad_norm": 2.664931535720825,
|
|
"learning_rate": 9.677426130117313e-06,
|
|
"loss": 1.1686,
|
|
"step": 1514
|
|
},
|
|
{
|
|
"epoch": 0.7967394162503287,
|
|
"grad_norm": 2.777085065841675,
|
|
"learning_rate": 9.676811088206149e-06,
|
|
"loss": 1.1981,
|
|
"step": 1515
|
|
},
|
|
{
|
|
"epoch": 0.7972653168551144,
|
|
"grad_norm": 2.884260654449463,
|
|
"learning_rate": 9.676195480094468e-06,
|
|
"loss": 1.2575,
|
|
"step": 1516
|
|
},
|
|
{
|
|
"epoch": 0.7977912174599001,
|
|
"grad_norm": 2.7648863792419434,
|
|
"learning_rate": 9.675579305856797e-06,
|
|
"loss": 1.2155,
|
|
"step": 1517
|
|
},
|
|
{
|
|
"epoch": 0.7983171180646857,
|
|
"grad_norm": 2.772732973098755,
|
|
"learning_rate": 9.674962565567733e-06,
|
|
"loss": 1.1764,
|
|
"step": 1518
|
|
},
|
|
{
|
|
"epoch": 0.7988430186694715,
|
|
"grad_norm": 2.6146974563598633,
|
|
"learning_rate": 9.674345259301945e-06,
|
|
"loss": 1.2336,
|
|
"step": 1519
|
|
},
|
|
{
|
|
"epoch": 0.7993689192742571,
|
|
"grad_norm": 2.628629446029663,
|
|
"learning_rate": 9.673727387134165e-06,
|
|
"loss": 1.2812,
|
|
"step": 1520
|
|
},
|
|
{
|
|
"epoch": 0.7998948198790429,
|
|
"grad_norm": 2.738891839981079,
|
|
"learning_rate": 9.673108949139194e-06,
|
|
"loss": 1.1998,
|
|
"step": 1521
|
|
},
|
|
{
|
|
"epoch": 0.8004207204838285,
|
|
"grad_norm": 3.0572421550750732,
|
|
"learning_rate": 9.672489945391908e-06,
|
|
"loss": 1.1963,
|
|
"step": 1522
|
|
},
|
|
{
|
|
"epoch": 0.8009466210886143,
|
|
"grad_norm": 2.6757426261901855,
|
|
"learning_rate": 9.671870375967245e-06,
|
|
"loss": 1.227,
|
|
"step": 1523
|
|
},
|
|
{
|
|
"epoch": 0.8014725216934,
|
|
"grad_norm": 2.7098464965820312,
|
|
"learning_rate": 9.671250240940212e-06,
|
|
"loss": 1.1649,
|
|
"step": 1524
|
|
},
|
|
{
|
|
"epoch": 0.8019984222981856,
|
|
"grad_norm": 2.760552167892456,
|
|
"learning_rate": 9.67062954038589e-06,
|
|
"loss": 1.2074,
|
|
"step": 1525
|
|
},
|
|
{
|
|
"epoch": 0.8025243229029714,
|
|
"grad_norm": 2.601506471633911,
|
|
"learning_rate": 9.67000827437942e-06,
|
|
"loss": 1.1217,
|
|
"step": 1526
|
|
},
|
|
{
|
|
"epoch": 0.803050223507757,
|
|
"grad_norm": 2.871858596801758,
|
|
"learning_rate": 9.669386442996018e-06,
|
|
"loss": 1.2198,
|
|
"step": 1527
|
|
},
|
|
{
|
|
"epoch": 0.8035761241125428,
|
|
"grad_norm": 2.9557905197143555,
|
|
"learning_rate": 9.668764046310969e-06,
|
|
"loss": 1.1655,
|
|
"step": 1528
|
|
},
|
|
{
|
|
"epoch": 0.8041020247173284,
|
|
"grad_norm": 2.5137264728546143,
|
|
"learning_rate": 9.668141084399617e-06,
|
|
"loss": 1.1481,
|
|
"step": 1529
|
|
},
|
|
{
|
|
"epoch": 0.8046279253221141,
|
|
"grad_norm": 2.7946362495422363,
|
|
"learning_rate": 9.667517557337386e-06,
|
|
"loss": 1.2432,
|
|
"step": 1530
|
|
},
|
|
{
|
|
"epoch": 0.8051538259268998,
|
|
"grad_norm": 2.6417903900146484,
|
|
"learning_rate": 9.666893465199763e-06,
|
|
"loss": 1.1808,
|
|
"step": 1531
|
|
},
|
|
{
|
|
"epoch": 0.8056797265316855,
|
|
"grad_norm": 2.709798574447632,
|
|
"learning_rate": 9.666268808062303e-06,
|
|
"loss": 1.2042,
|
|
"step": 1532
|
|
},
|
|
{
|
|
"epoch": 0.8062056271364713,
|
|
"grad_norm": 2.9464313983917236,
|
|
"learning_rate": 9.665643586000632e-06,
|
|
"loss": 1.1403,
|
|
"step": 1533
|
|
},
|
|
{
|
|
"epoch": 0.8067315277412569,
|
|
"grad_norm": 2.5108072757720947,
|
|
"learning_rate": 9.665017799090442e-06,
|
|
"loss": 1.1089,
|
|
"step": 1534
|
|
},
|
|
{
|
|
"epoch": 0.8072574283460426,
|
|
"grad_norm": 2.8049614429473877,
|
|
"learning_rate": 9.664391447407494e-06,
|
|
"loss": 1.2523,
|
|
"step": 1535
|
|
},
|
|
{
|
|
"epoch": 0.8077833289508283,
|
|
"grad_norm": 2.8692851066589355,
|
|
"learning_rate": 9.663764531027617e-06,
|
|
"loss": 1.2332,
|
|
"step": 1536
|
|
},
|
|
{
|
|
"epoch": 0.808309229555614,
|
|
"grad_norm": 2.7594306468963623,
|
|
"learning_rate": 9.663137050026711e-06,
|
|
"loss": 1.341,
|
|
"step": 1537
|
|
},
|
|
{
|
|
"epoch": 0.8088351301603997,
|
|
"grad_norm": 2.9078269004821777,
|
|
"learning_rate": 9.66250900448074e-06,
|
|
"loss": 1.2631,
|
|
"step": 1538
|
|
},
|
|
{
|
|
"epoch": 0.8093610307651854,
|
|
"grad_norm": 2.6946253776550293,
|
|
"learning_rate": 9.66188039446574e-06,
|
|
"loss": 1.0892,
|
|
"step": 1539
|
|
},
|
|
{
|
|
"epoch": 0.809886931369971,
|
|
"grad_norm": 2.969816207885742,
|
|
"learning_rate": 9.661251220057814e-06,
|
|
"loss": 1.2373,
|
|
"step": 1540
|
|
},
|
|
{
|
|
"epoch": 0.8104128319747568,
|
|
"grad_norm": 3.0155832767486572,
|
|
"learning_rate": 9.660621481333133e-06,
|
|
"loss": 1.2545,
|
|
"step": 1541
|
|
},
|
|
{
|
|
"epoch": 0.8109387325795425,
|
|
"grad_norm": 2.671207904815674,
|
|
"learning_rate": 9.659991178367937e-06,
|
|
"loss": 1.0747,
|
|
"step": 1542
|
|
},
|
|
{
|
|
"epoch": 0.8114646331843282,
|
|
"grad_norm": 2.7716901302337646,
|
|
"learning_rate": 9.659360311238533e-06,
|
|
"loss": 1.1775,
|
|
"step": 1543
|
|
},
|
|
{
|
|
"epoch": 0.8119905337891139,
|
|
"grad_norm": 2.6992552280426025,
|
|
"learning_rate": 9.658728880021297e-06,
|
|
"loss": 1.2053,
|
|
"step": 1544
|
|
},
|
|
{
|
|
"epoch": 0.8125164343938995,
|
|
"grad_norm": 2.6758289337158203,
|
|
"learning_rate": 9.658096884792675e-06,
|
|
"loss": 1.2054,
|
|
"step": 1545
|
|
},
|
|
{
|
|
"epoch": 0.8130423349986853,
|
|
"grad_norm": 2.6225664615631104,
|
|
"learning_rate": 9.65746432562918e-06,
|
|
"loss": 1.1216,
|
|
"step": 1546
|
|
},
|
|
{
|
|
"epoch": 0.8135682356034709,
|
|
"grad_norm": 2.705693244934082,
|
|
"learning_rate": 9.656831202607393e-06,
|
|
"loss": 1.1195,
|
|
"step": 1547
|
|
},
|
|
{
|
|
"epoch": 0.8140941362082567,
|
|
"grad_norm": 2.83599591255188,
|
|
"learning_rate": 9.65619751580396e-06,
|
|
"loss": 1.1785,
|
|
"step": 1548
|
|
},
|
|
{
|
|
"epoch": 0.8146200368130423,
|
|
"grad_norm": 2.6572041511535645,
|
|
"learning_rate": 9.655563265295605e-06,
|
|
"loss": 1.1493,
|
|
"step": 1549
|
|
},
|
|
{
|
|
"epoch": 0.815145937417828,
|
|
"grad_norm": 2.8744571208953857,
|
|
"learning_rate": 9.654928451159107e-06,
|
|
"loss": 1.2495,
|
|
"step": 1550
|
|
},
|
|
{
|
|
"epoch": 0.8156718380226137,
|
|
"grad_norm": 2.761871337890625,
|
|
"learning_rate": 9.654293073471324e-06,
|
|
"loss": 1.0754,
|
|
"step": 1551
|
|
},
|
|
{
|
|
"epoch": 0.8161977386273994,
|
|
"grad_norm": 2.6714866161346436,
|
|
"learning_rate": 9.653657132309178e-06,
|
|
"loss": 1.1499,
|
|
"step": 1552
|
|
},
|
|
{
|
|
"epoch": 0.8167236392321852,
|
|
"grad_norm": 2.821441650390625,
|
|
"learning_rate": 9.653020627749658e-06,
|
|
"loss": 1.2267,
|
|
"step": 1553
|
|
},
|
|
{
|
|
"epoch": 0.8172495398369708,
|
|
"grad_norm": 2.8228530883789062,
|
|
"learning_rate": 9.652383559869824e-06,
|
|
"loss": 1.2957,
|
|
"step": 1554
|
|
},
|
|
{
|
|
"epoch": 0.8177754404417565,
|
|
"grad_norm": 3.02482271194458,
|
|
"learning_rate": 9.651745928746804e-06,
|
|
"loss": 1.2251,
|
|
"step": 1555
|
|
},
|
|
{
|
|
"epoch": 0.8183013410465422,
|
|
"grad_norm": 3.066932439804077,
|
|
"learning_rate": 9.651107734457792e-06,
|
|
"loss": 1.2282,
|
|
"step": 1556
|
|
},
|
|
{
|
|
"epoch": 0.8188272416513279,
|
|
"grad_norm": 3.0092151165008545,
|
|
"learning_rate": 9.65046897708005e-06,
|
|
"loss": 1.1859,
|
|
"step": 1557
|
|
},
|
|
{
|
|
"epoch": 0.8193531422561136,
|
|
"grad_norm": 3.231405258178711,
|
|
"learning_rate": 9.649829656690912e-06,
|
|
"loss": 1.1641,
|
|
"step": 1558
|
|
},
|
|
{
|
|
"epoch": 0.8198790428608993,
|
|
"grad_norm": 2.867708206176758,
|
|
"learning_rate": 9.649189773367775e-06,
|
|
"loss": 1.2825,
|
|
"step": 1559
|
|
},
|
|
{
|
|
"epoch": 0.8204049434656849,
|
|
"grad_norm": 3.029829978942871,
|
|
"learning_rate": 9.648549327188107e-06,
|
|
"loss": 1.278,
|
|
"step": 1560
|
|
},
|
|
{
|
|
"epoch": 0.8209308440704707,
|
|
"grad_norm": 2.7639687061309814,
|
|
"learning_rate": 9.647908318229448e-06,
|
|
"loss": 1.224,
|
|
"step": 1561
|
|
},
|
|
{
|
|
"epoch": 0.8214567446752564,
|
|
"grad_norm": 2.9943764209747314,
|
|
"learning_rate": 9.647266746569399e-06,
|
|
"loss": 1.2738,
|
|
"step": 1562
|
|
},
|
|
{
|
|
"epoch": 0.8219826452800421,
|
|
"grad_norm": 2.9306750297546387,
|
|
"learning_rate": 9.64662461228563e-06,
|
|
"loss": 1.3553,
|
|
"step": 1563
|
|
},
|
|
{
|
|
"epoch": 0.8225085458848278,
|
|
"grad_norm": 2.955819606781006,
|
|
"learning_rate": 9.645981915455886e-06,
|
|
"loss": 1.2884,
|
|
"step": 1564
|
|
},
|
|
{
|
|
"epoch": 0.8230344464896134,
|
|
"grad_norm": 2.813762664794922,
|
|
"learning_rate": 9.645338656157974e-06,
|
|
"loss": 1.2712,
|
|
"step": 1565
|
|
},
|
|
{
|
|
"epoch": 0.8235603470943992,
|
|
"grad_norm": 2.7259156703948975,
|
|
"learning_rate": 9.644694834469768e-06,
|
|
"loss": 1.2022,
|
|
"step": 1566
|
|
},
|
|
{
|
|
"epoch": 0.8240862476991848,
|
|
"grad_norm": 3.0204336643218994,
|
|
"learning_rate": 9.644050450469214e-06,
|
|
"loss": 1.2034,
|
|
"step": 1567
|
|
},
|
|
{
|
|
"epoch": 0.8246121483039706,
|
|
"grad_norm": 2.636988401412964,
|
|
"learning_rate": 9.643405504234324e-06,
|
|
"loss": 1.0881,
|
|
"step": 1568
|
|
},
|
|
{
|
|
"epoch": 0.8251380489087562,
|
|
"grad_norm": 2.6215572357177734,
|
|
"learning_rate": 9.642759995843183e-06,
|
|
"loss": 1.029,
|
|
"step": 1569
|
|
},
|
|
{
|
|
"epoch": 0.8256639495135419,
|
|
"grad_norm": 2.6341898441314697,
|
|
"learning_rate": 9.642113925373935e-06,
|
|
"loss": 1.2634,
|
|
"step": 1570
|
|
},
|
|
{
|
|
"epoch": 0.8261898501183277,
|
|
"grad_norm": 2.9175565242767334,
|
|
"learning_rate": 9.641467292904797e-06,
|
|
"loss": 1.2804,
|
|
"step": 1571
|
|
},
|
|
{
|
|
"epoch": 0.8267157507231133,
|
|
"grad_norm": 2.9280312061309814,
|
|
"learning_rate": 9.640820098514056e-06,
|
|
"loss": 1.2794,
|
|
"step": 1572
|
|
},
|
|
{
|
|
"epoch": 0.8272416513278991,
|
|
"grad_norm": 3.088383436203003,
|
|
"learning_rate": 9.640172342280062e-06,
|
|
"loss": 1.1755,
|
|
"step": 1573
|
|
},
|
|
{
|
|
"epoch": 0.8277675519326847,
|
|
"grad_norm": 3.0267891883850098,
|
|
"learning_rate": 9.639524024281242e-06,
|
|
"loss": 1.2888,
|
|
"step": 1574
|
|
},
|
|
{
|
|
"epoch": 0.8282934525374704,
|
|
"grad_norm": 2.783097743988037,
|
|
"learning_rate": 9.63887514459608e-06,
|
|
"loss": 1.156,
|
|
"step": 1575
|
|
},
|
|
{
|
|
"epoch": 0.8288193531422561,
|
|
"grad_norm": 2.840513229370117,
|
|
"learning_rate": 9.638225703303132e-06,
|
|
"loss": 1.2147,
|
|
"step": 1576
|
|
},
|
|
{
|
|
"epoch": 0.8293452537470418,
|
|
"grad_norm": 2.728584051132202,
|
|
"learning_rate": 9.637575700481027e-06,
|
|
"loss": 1.1953,
|
|
"step": 1577
|
|
},
|
|
{
|
|
"epoch": 0.8298711543518275,
|
|
"grad_norm": 2.929257869720459,
|
|
"learning_rate": 9.636925136208455e-06,
|
|
"loss": 1.2551,
|
|
"step": 1578
|
|
},
|
|
{
|
|
"epoch": 0.8303970549566132,
|
|
"grad_norm": 2.8386595249176025,
|
|
"learning_rate": 9.636274010564176e-06,
|
|
"loss": 1.2281,
|
|
"step": 1579
|
|
},
|
|
{
|
|
"epoch": 0.830922955561399,
|
|
"grad_norm": 2.7139179706573486,
|
|
"learning_rate": 9.635622323627023e-06,
|
|
"loss": 1.2726,
|
|
"step": 1580
|
|
},
|
|
{
|
|
"epoch": 0.8314488561661846,
|
|
"grad_norm": 2.7066378593444824,
|
|
"learning_rate": 9.634970075475892e-06,
|
|
"loss": 1.307,
|
|
"step": 1581
|
|
},
|
|
{
|
|
"epoch": 0.8319747567709703,
|
|
"grad_norm": 2.9181277751922607,
|
|
"learning_rate": 9.634317266189743e-06,
|
|
"loss": 1.2538,
|
|
"step": 1582
|
|
},
|
|
{
|
|
"epoch": 0.832500657375756,
|
|
"grad_norm": 2.7798993587493896,
|
|
"learning_rate": 9.633663895847611e-06,
|
|
"loss": 1.2407,
|
|
"step": 1583
|
|
},
|
|
{
|
|
"epoch": 0.8330265579805417,
|
|
"grad_norm": 2.651545524597168,
|
|
"learning_rate": 9.6330099645286e-06,
|
|
"loss": 1.1648,
|
|
"step": 1584
|
|
},
|
|
{
|
|
"epoch": 0.8335524585853273,
|
|
"grad_norm": 2.7882111072540283,
|
|
"learning_rate": 9.632355472311876e-06,
|
|
"loss": 1.1942,
|
|
"step": 1585
|
|
},
|
|
{
|
|
"epoch": 0.8340783591901131,
|
|
"grad_norm": 2.7278146743774414,
|
|
"learning_rate": 9.631700419276675e-06,
|
|
"loss": 1.2767,
|
|
"step": 1586
|
|
},
|
|
{
|
|
"epoch": 0.8346042597948987,
|
|
"grad_norm": 2.6416289806365967,
|
|
"learning_rate": 9.6310448055023e-06,
|
|
"loss": 1.197,
|
|
"step": 1587
|
|
},
|
|
{
|
|
"epoch": 0.8351301603996845,
|
|
"grad_norm": 2.7964277267456055,
|
|
"learning_rate": 9.630388631068127e-06,
|
|
"loss": 1.2493,
|
|
"step": 1588
|
|
},
|
|
{
|
|
"epoch": 0.8356560610044702,
|
|
"grad_norm": 2.770033836364746,
|
|
"learning_rate": 9.629731896053593e-06,
|
|
"loss": 1.1773,
|
|
"step": 1589
|
|
},
|
|
{
|
|
"epoch": 0.8361819616092558,
|
|
"grad_norm": 2.763554096221924,
|
|
"learning_rate": 9.629074600538209e-06,
|
|
"loss": 1.1839,
|
|
"step": 1590
|
|
},
|
|
{
|
|
"epoch": 0.8367078622140416,
|
|
"grad_norm": 2.7049455642700195,
|
|
"learning_rate": 9.628416744601546e-06,
|
|
"loss": 1.2921,
|
|
"step": 1591
|
|
},
|
|
{
|
|
"epoch": 0.8372337628188272,
|
|
"grad_norm": 2.660151720046997,
|
|
"learning_rate": 9.627758328323251e-06,
|
|
"loss": 1.1908,
|
|
"step": 1592
|
|
},
|
|
{
|
|
"epoch": 0.837759663423613,
|
|
"grad_norm": 2.721008777618408,
|
|
"learning_rate": 9.627099351783037e-06,
|
|
"loss": 1.2355,
|
|
"step": 1593
|
|
},
|
|
{
|
|
"epoch": 0.8382855640283986,
|
|
"grad_norm": 2.943514108657837,
|
|
"learning_rate": 9.626439815060679e-06,
|
|
"loss": 1.1832,
|
|
"step": 1594
|
|
},
|
|
{
|
|
"epoch": 0.8388114646331843,
|
|
"grad_norm": 2.6027379035949707,
|
|
"learning_rate": 9.625779718236027e-06,
|
|
"loss": 1.266,
|
|
"step": 1595
|
|
},
|
|
{
|
|
"epoch": 0.83933736523797,
|
|
"grad_norm": 2.5319266319274902,
|
|
"learning_rate": 9.625119061388996e-06,
|
|
"loss": 1.2421,
|
|
"step": 1596
|
|
},
|
|
{
|
|
"epoch": 0.8398632658427557,
|
|
"grad_norm": 2.7466495037078857,
|
|
"learning_rate": 9.624457844599569e-06,
|
|
"loss": 1.3561,
|
|
"step": 1597
|
|
},
|
|
{
|
|
"epoch": 0.8403891664475415,
|
|
"grad_norm": 2.9346201419830322,
|
|
"learning_rate": 9.623796067947795e-06,
|
|
"loss": 1.2873,
|
|
"step": 1598
|
|
},
|
|
{
|
|
"epoch": 0.8409150670523271,
|
|
"grad_norm": 2.708796262741089,
|
|
"learning_rate": 9.623133731513794e-06,
|
|
"loss": 1.1697,
|
|
"step": 1599
|
|
},
|
|
{
|
|
"epoch": 0.8414409676571128,
|
|
"grad_norm": 3.1593384742736816,
|
|
"learning_rate": 9.622470835377751e-06,
|
|
"loss": 1.2573,
|
|
"step": 1600
|
|
},
|
|
{
|
|
"epoch": 0.8419668682618985,
|
|
"grad_norm": 2.7191717624664307,
|
|
"learning_rate": 9.621807379619919e-06,
|
|
"loss": 1.1563,
|
|
"step": 1601
|
|
},
|
|
{
|
|
"epoch": 0.8424927688666842,
|
|
"grad_norm": 2.8388102054595947,
|
|
"learning_rate": 9.62114336432062e-06,
|
|
"loss": 1.3111,
|
|
"step": 1602
|
|
},
|
|
{
|
|
"epoch": 0.8430186694714699,
|
|
"grad_norm": 2.9051408767700195,
|
|
"learning_rate": 9.620478789560246e-06,
|
|
"loss": 1.254,
|
|
"step": 1603
|
|
},
|
|
{
|
|
"epoch": 0.8435445700762556,
|
|
"grad_norm": 2.6088008880615234,
|
|
"learning_rate": 9.61981365541925e-06,
|
|
"loss": 1.2482,
|
|
"step": 1604
|
|
},
|
|
{
|
|
"epoch": 0.8440704706810412,
|
|
"grad_norm": 2.7293787002563477,
|
|
"learning_rate": 9.61914796197816e-06,
|
|
"loss": 1.2244,
|
|
"step": 1605
|
|
},
|
|
{
|
|
"epoch": 0.844596371285827,
|
|
"grad_norm": 2.8841519355773926,
|
|
"learning_rate": 9.618481709317568e-06,
|
|
"loss": 1.1791,
|
|
"step": 1606
|
|
},
|
|
{
|
|
"epoch": 0.8451222718906126,
|
|
"grad_norm": 2.9555821418762207,
|
|
"learning_rate": 9.617814897518132e-06,
|
|
"loss": 1.3493,
|
|
"step": 1607
|
|
},
|
|
{
|
|
"epoch": 0.8456481724953984,
|
|
"grad_norm": 2.7201898097991943,
|
|
"learning_rate": 9.617147526660584e-06,
|
|
"loss": 1.1364,
|
|
"step": 1608
|
|
},
|
|
{
|
|
"epoch": 0.8461740731001841,
|
|
"grad_norm": 2.704317808151245,
|
|
"learning_rate": 9.616479596825714e-06,
|
|
"loss": 1.1275,
|
|
"step": 1609
|
|
},
|
|
{
|
|
"epoch": 0.8466999737049697,
|
|
"grad_norm": 2.974184036254883,
|
|
"learning_rate": 9.615811108094391e-06,
|
|
"loss": 1.3097,
|
|
"step": 1610
|
|
},
|
|
{
|
|
"epoch": 0.8472258743097555,
|
|
"grad_norm": 2.597834348678589,
|
|
"learning_rate": 9.615142060547542e-06,
|
|
"loss": 1.1942,
|
|
"step": 1611
|
|
},
|
|
{
|
|
"epoch": 0.8477517749145411,
|
|
"grad_norm": 2.8080523014068604,
|
|
"learning_rate": 9.614472454266166e-06,
|
|
"loss": 1.1843,
|
|
"step": 1612
|
|
},
|
|
{
|
|
"epoch": 0.8482776755193269,
|
|
"grad_norm": 2.7124593257904053,
|
|
"learning_rate": 9.613802289331332e-06,
|
|
"loss": 1.1048,
|
|
"step": 1613
|
|
},
|
|
{
|
|
"epoch": 0.8488035761241125,
|
|
"grad_norm": 2.571730375289917,
|
|
"learning_rate": 9.613131565824171e-06,
|
|
"loss": 1.1632,
|
|
"step": 1614
|
|
},
|
|
{
|
|
"epoch": 0.8493294767288982,
|
|
"grad_norm": 2.5843396186828613,
|
|
"learning_rate": 9.612460283825886e-06,
|
|
"loss": 1.1164,
|
|
"step": 1615
|
|
},
|
|
{
|
|
"epoch": 0.8498553773336839,
|
|
"grad_norm": 2.759028673171997,
|
|
"learning_rate": 9.611788443417744e-06,
|
|
"loss": 1.211,
|
|
"step": 1616
|
|
},
|
|
{
|
|
"epoch": 0.8503812779384696,
|
|
"grad_norm": 2.680767059326172,
|
|
"learning_rate": 9.611116044681085e-06,
|
|
"loss": 1.2094,
|
|
"step": 1617
|
|
},
|
|
{
|
|
"epoch": 0.8509071785432554,
|
|
"grad_norm": 2.775465488433838,
|
|
"learning_rate": 9.610443087697312e-06,
|
|
"loss": 1.1855,
|
|
"step": 1618
|
|
},
|
|
{
|
|
"epoch": 0.851433079148041,
|
|
"grad_norm": 2.877070903778076,
|
|
"learning_rate": 9.609769572547895e-06,
|
|
"loss": 1.3096,
|
|
"step": 1619
|
|
},
|
|
{
|
|
"epoch": 0.8519589797528268,
|
|
"grad_norm": 2.8171451091766357,
|
|
"learning_rate": 9.609095499314376e-06,
|
|
"loss": 1.1947,
|
|
"step": 1620
|
|
},
|
|
{
|
|
"epoch": 0.8524848803576124,
|
|
"grad_norm": 2.6643967628479004,
|
|
"learning_rate": 9.608420868078359e-06,
|
|
"loss": 1.1345,
|
|
"step": 1621
|
|
},
|
|
{
|
|
"epoch": 0.8530107809623981,
|
|
"grad_norm": 2.875979423522949,
|
|
"learning_rate": 9.607745678921523e-06,
|
|
"loss": 1.2495,
|
|
"step": 1622
|
|
},
|
|
{
|
|
"epoch": 0.8535366815671838,
|
|
"grad_norm": 2.613020658493042,
|
|
"learning_rate": 9.607069931925605e-06,
|
|
"loss": 1.2295,
|
|
"step": 1623
|
|
},
|
|
{
|
|
"epoch": 0.8540625821719695,
|
|
"grad_norm": 2.676098346710205,
|
|
"learning_rate": 9.60639362717242e-06,
|
|
"loss": 1.2937,
|
|
"step": 1624
|
|
},
|
|
{
|
|
"epoch": 0.8545884827767551,
|
|
"grad_norm": 2.7270593643188477,
|
|
"learning_rate": 9.605716764743841e-06,
|
|
"loss": 1.1854,
|
|
"step": 1625
|
|
},
|
|
{
|
|
"epoch": 0.8551143833815409,
|
|
"grad_norm": 2.966747283935547,
|
|
"learning_rate": 9.605039344721813e-06,
|
|
"loss": 1.2555,
|
|
"step": 1626
|
|
},
|
|
{
|
|
"epoch": 0.8556402839863266,
|
|
"grad_norm": 2.951371669769287,
|
|
"learning_rate": 9.604361367188351e-06,
|
|
"loss": 1.1998,
|
|
"step": 1627
|
|
},
|
|
{
|
|
"epoch": 0.8561661845911123,
|
|
"grad_norm": 2.8641161918640137,
|
|
"learning_rate": 9.603682832225534e-06,
|
|
"loss": 1.1748,
|
|
"step": 1628
|
|
},
|
|
{
|
|
"epoch": 0.856692085195898,
|
|
"grad_norm": 2.683035135269165,
|
|
"learning_rate": 9.603003739915507e-06,
|
|
"loss": 1.2257,
|
|
"step": 1629
|
|
},
|
|
{
|
|
"epoch": 0.8572179858006836,
|
|
"grad_norm": 2.8766727447509766,
|
|
"learning_rate": 9.602324090340484e-06,
|
|
"loss": 1.1615,
|
|
"step": 1630
|
|
},
|
|
{
|
|
"epoch": 0.8577438864054694,
|
|
"grad_norm": 2.948503017425537,
|
|
"learning_rate": 9.601643883582752e-06,
|
|
"loss": 1.3048,
|
|
"step": 1631
|
|
},
|
|
{
|
|
"epoch": 0.858269787010255,
|
|
"grad_norm": 2.6622724533081055,
|
|
"learning_rate": 9.600963119724655e-06,
|
|
"loss": 1.1786,
|
|
"step": 1632
|
|
},
|
|
{
|
|
"epoch": 0.8587956876150408,
|
|
"grad_norm": 2.6820144653320312,
|
|
"learning_rate": 9.600281798848614e-06,
|
|
"loss": 1.125,
|
|
"step": 1633
|
|
},
|
|
{
|
|
"epoch": 0.8593215882198264,
|
|
"grad_norm": 2.8401622772216797,
|
|
"learning_rate": 9.599599921037112e-06,
|
|
"loss": 1.265,
|
|
"step": 1634
|
|
},
|
|
{
|
|
"epoch": 0.8598474888246121,
|
|
"grad_norm": 2.92401385307312,
|
|
"learning_rate": 9.5989174863727e-06,
|
|
"loss": 1.205,
|
|
"step": 1635
|
|
},
|
|
{
|
|
"epoch": 0.8603733894293979,
|
|
"grad_norm": 2.9983785152435303,
|
|
"learning_rate": 9.598234494938e-06,
|
|
"loss": 1.2931,
|
|
"step": 1636
|
|
},
|
|
{
|
|
"epoch": 0.8608992900341835,
|
|
"grad_norm": 2.72944712638855,
|
|
"learning_rate": 9.597550946815696e-06,
|
|
"loss": 1.2224,
|
|
"step": 1637
|
|
},
|
|
{
|
|
"epoch": 0.8614251906389693,
|
|
"grad_norm": 2.869067430496216,
|
|
"learning_rate": 9.596866842088543e-06,
|
|
"loss": 1.2696,
|
|
"step": 1638
|
|
},
|
|
{
|
|
"epoch": 0.8619510912437549,
|
|
"grad_norm": 3.075871229171753,
|
|
"learning_rate": 9.596182180839365e-06,
|
|
"loss": 1.2194,
|
|
"step": 1639
|
|
},
|
|
{
|
|
"epoch": 0.8624769918485407,
|
|
"grad_norm": 2.780397891998291,
|
|
"learning_rate": 9.595496963151045e-06,
|
|
"loss": 1.2157,
|
|
"step": 1640
|
|
},
|
|
{
|
|
"epoch": 0.8630028924533263,
|
|
"grad_norm": 2.9231302738189697,
|
|
"learning_rate": 9.594811189106543e-06,
|
|
"loss": 1.1671,
|
|
"step": 1641
|
|
},
|
|
{
|
|
"epoch": 0.863528793058112,
|
|
"grad_norm": 2.9471402168273926,
|
|
"learning_rate": 9.594124858788883e-06,
|
|
"loss": 1.2272,
|
|
"step": 1642
|
|
},
|
|
{
|
|
"epoch": 0.8640546936628977,
|
|
"grad_norm": 2.711583137512207,
|
|
"learning_rate": 9.593437972281154e-06,
|
|
"loss": 1.1012,
|
|
"step": 1643
|
|
},
|
|
{
|
|
"epoch": 0.8645805942676834,
|
|
"grad_norm": 2.864274263381958,
|
|
"learning_rate": 9.592750529666516e-06,
|
|
"loss": 1.2068,
|
|
"step": 1644
|
|
},
|
|
{
|
|
"epoch": 0.8651064948724692,
|
|
"grad_norm": 2.681593418121338,
|
|
"learning_rate": 9.592062531028196e-06,
|
|
"loss": 1.222,
|
|
"step": 1645
|
|
},
|
|
{
|
|
"epoch": 0.8656323954772548,
|
|
"grad_norm": 2.634826183319092,
|
|
"learning_rate": 9.591373976449482e-06,
|
|
"loss": 1.1253,
|
|
"step": 1646
|
|
},
|
|
{
|
|
"epoch": 0.8661582960820405,
|
|
"grad_norm": 2.9753003120422363,
|
|
"learning_rate": 9.59068486601374e-06,
|
|
"loss": 1.2985,
|
|
"step": 1647
|
|
},
|
|
{
|
|
"epoch": 0.8666841966868262,
|
|
"grad_norm": 2.6949801445007324,
|
|
"learning_rate": 9.589995199804392e-06,
|
|
"loss": 1.1964,
|
|
"step": 1648
|
|
},
|
|
{
|
|
"epoch": 0.8672100972916119,
|
|
"grad_norm": 2.7531394958496094,
|
|
"learning_rate": 9.589304977904936e-06,
|
|
"loss": 1.2457,
|
|
"step": 1649
|
|
},
|
|
{
|
|
"epoch": 0.8677359978963975,
|
|
"grad_norm": 2.827765464782715,
|
|
"learning_rate": 9.588614200398935e-06,
|
|
"loss": 1.3127,
|
|
"step": 1650
|
|
},
|
|
{
|
|
"epoch": 0.8682618985011833,
|
|
"grad_norm": 2.861640214920044,
|
|
"learning_rate": 9.587922867370016e-06,
|
|
"loss": 1.3015,
|
|
"step": 1651
|
|
},
|
|
{
|
|
"epoch": 0.8687877991059689,
|
|
"grad_norm": 2.6090991497039795,
|
|
"learning_rate": 9.587230978901876e-06,
|
|
"loss": 1.1822,
|
|
"step": 1652
|
|
},
|
|
{
|
|
"epoch": 0.8693136997107547,
|
|
"grad_norm": 2.6083171367645264,
|
|
"learning_rate": 9.586538535078278e-06,
|
|
"loss": 1.0594,
|
|
"step": 1653
|
|
},
|
|
{
|
|
"epoch": 0.8698396003155404,
|
|
"grad_norm": 2.6702592372894287,
|
|
"learning_rate": 9.585845535983056e-06,
|
|
"loss": 1.1494,
|
|
"step": 1654
|
|
},
|
|
{
|
|
"epoch": 0.870365500920326,
|
|
"grad_norm": 2.609999418258667,
|
|
"learning_rate": 9.585151981700105e-06,
|
|
"loss": 1.1322,
|
|
"step": 1655
|
|
},
|
|
{
|
|
"epoch": 0.8708914015251118,
|
|
"grad_norm": 2.881410598754883,
|
|
"learning_rate": 9.584457872313395e-06,
|
|
"loss": 1.2187,
|
|
"step": 1656
|
|
},
|
|
{
|
|
"epoch": 0.8714173021298974,
|
|
"grad_norm": 2.8677978515625,
|
|
"learning_rate": 9.583763207906953e-06,
|
|
"loss": 1.2493,
|
|
"step": 1657
|
|
},
|
|
{
|
|
"epoch": 0.8719432027346832,
|
|
"grad_norm": 2.7290611267089844,
|
|
"learning_rate": 9.583067988564883e-06,
|
|
"loss": 1.254,
|
|
"step": 1658
|
|
},
|
|
{
|
|
"epoch": 0.8724691033394688,
|
|
"grad_norm": 2.5928165912628174,
|
|
"learning_rate": 9.58237221437135e-06,
|
|
"loss": 1.0573,
|
|
"step": 1659
|
|
},
|
|
{
|
|
"epoch": 0.8729950039442546,
|
|
"grad_norm": 2.5771067142486572,
|
|
"learning_rate": 9.58167588541059e-06,
|
|
"loss": 1.1964,
|
|
"step": 1660
|
|
},
|
|
{
|
|
"epoch": 0.8735209045490402,
|
|
"grad_norm": 2.8102543354034424,
|
|
"learning_rate": 9.580979001766903e-06,
|
|
"loss": 1.1886,
|
|
"step": 1661
|
|
},
|
|
{
|
|
"epoch": 0.8740468051538259,
|
|
"grad_norm": 2.874027967453003,
|
|
"learning_rate": 9.580281563524659e-06,
|
|
"loss": 1.2827,
|
|
"step": 1662
|
|
},
|
|
{
|
|
"epoch": 0.8745727057586116,
|
|
"grad_norm": 2.948605537414551,
|
|
"learning_rate": 9.579583570768292e-06,
|
|
"loss": 1.2532,
|
|
"step": 1663
|
|
},
|
|
{
|
|
"epoch": 0.8750986063633973,
|
|
"grad_norm": 2.7634012699127197,
|
|
"learning_rate": 9.578885023582305e-06,
|
|
"loss": 1.2713,
|
|
"step": 1664
|
|
},
|
|
{
|
|
"epoch": 0.8756245069681831,
|
|
"grad_norm": 2.8409080505371094,
|
|
"learning_rate": 9.578185922051269e-06,
|
|
"loss": 1.1976,
|
|
"step": 1665
|
|
},
|
|
{
|
|
"epoch": 0.8761504075729687,
|
|
"grad_norm": 2.8217763900756836,
|
|
"learning_rate": 9.577486266259819e-06,
|
|
"loss": 1.2592,
|
|
"step": 1666
|
|
},
|
|
{
|
|
"epoch": 0.8766763081777544,
|
|
"grad_norm": 2.8466365337371826,
|
|
"learning_rate": 9.576786056292662e-06,
|
|
"loss": 1.1083,
|
|
"step": 1667
|
|
},
|
|
{
|
|
"epoch": 0.8772022087825401,
|
|
"grad_norm": 2.854689836502075,
|
|
"learning_rate": 9.57608529223457e-06,
|
|
"loss": 1.1877,
|
|
"step": 1668
|
|
},
|
|
{
|
|
"epoch": 0.8777281093873258,
|
|
"grad_norm": 2.72452712059021,
|
|
"learning_rate": 9.575383974170378e-06,
|
|
"loss": 1.2756,
|
|
"step": 1669
|
|
},
|
|
{
|
|
"epoch": 0.8782540099921115,
|
|
"grad_norm": 2.937098741531372,
|
|
"learning_rate": 9.574682102184992e-06,
|
|
"loss": 1.3667,
|
|
"step": 1670
|
|
},
|
|
{
|
|
"epoch": 0.8787799105968972,
|
|
"grad_norm": 2.7525620460510254,
|
|
"learning_rate": 9.573979676363387e-06,
|
|
"loss": 1.1786,
|
|
"step": 1671
|
|
},
|
|
{
|
|
"epoch": 0.8793058112016828,
|
|
"grad_norm": 2.6156086921691895,
|
|
"learning_rate": 9.573276696790598e-06,
|
|
"loss": 1.1714,
|
|
"step": 1672
|
|
},
|
|
{
|
|
"epoch": 0.8798317118064686,
|
|
"grad_norm": 2.6241087913513184,
|
|
"learning_rate": 9.572573163551737e-06,
|
|
"loss": 1.2705,
|
|
"step": 1673
|
|
},
|
|
{
|
|
"epoch": 0.8803576124112543,
|
|
"grad_norm": 2.7373952865600586,
|
|
"learning_rate": 9.571869076731975e-06,
|
|
"loss": 1.2307,
|
|
"step": 1674
|
|
},
|
|
{
|
|
"epoch": 0.88088351301604,
|
|
"grad_norm": 2.6821327209472656,
|
|
"learning_rate": 9.571164436416554e-06,
|
|
"loss": 1.2173,
|
|
"step": 1675
|
|
},
|
|
{
|
|
"epoch": 0.8814094136208257,
|
|
"grad_norm": 2.782353401184082,
|
|
"learning_rate": 9.570459242690778e-06,
|
|
"loss": 1.1994,
|
|
"step": 1676
|
|
},
|
|
{
|
|
"epoch": 0.8819353142256113,
|
|
"grad_norm": 2.7853691577911377,
|
|
"learning_rate": 9.569753495640024e-06,
|
|
"loss": 1.1132,
|
|
"step": 1677
|
|
},
|
|
{
|
|
"epoch": 0.8824612148303971,
|
|
"grad_norm": 2.4772088527679443,
|
|
"learning_rate": 9.569047195349734e-06,
|
|
"loss": 1.2382,
|
|
"step": 1678
|
|
},
|
|
{
|
|
"epoch": 0.8829871154351827,
|
|
"grad_norm": 2.6584970951080322,
|
|
"learning_rate": 9.568340341905417e-06,
|
|
"loss": 1.2327,
|
|
"step": 1679
|
|
},
|
|
{
|
|
"epoch": 0.8835130160399685,
|
|
"grad_norm": 2.623839855194092,
|
|
"learning_rate": 9.567632935392648e-06,
|
|
"loss": 1.1533,
|
|
"step": 1680
|
|
},
|
|
{
|
|
"epoch": 0.8840389166447541,
|
|
"grad_norm": 2.6726222038269043,
|
|
"learning_rate": 9.56692497589707e-06,
|
|
"loss": 1.1712,
|
|
"step": 1681
|
|
},
|
|
{
|
|
"epoch": 0.8845648172495398,
|
|
"grad_norm": 2.76956844329834,
|
|
"learning_rate": 9.56621646350439e-06,
|
|
"loss": 1.2597,
|
|
"step": 1682
|
|
},
|
|
{
|
|
"epoch": 0.8850907178543256,
|
|
"grad_norm": 2.8912782669067383,
|
|
"learning_rate": 9.565507398300387e-06,
|
|
"loss": 1.2906,
|
|
"step": 1683
|
|
},
|
|
{
|
|
"epoch": 0.8856166184591112,
|
|
"grad_norm": 2.771631956100464,
|
|
"learning_rate": 9.564797780370902e-06,
|
|
"loss": 1.1196,
|
|
"step": 1684
|
|
},
|
|
{
|
|
"epoch": 0.886142519063897,
|
|
"grad_norm": 2.686568260192871,
|
|
"learning_rate": 9.564087609801848e-06,
|
|
"loss": 1.2047,
|
|
"step": 1685
|
|
},
|
|
{
|
|
"epoch": 0.8866684196686826,
|
|
"grad_norm": 2.6102840900421143,
|
|
"learning_rate": 9.563376886679204e-06,
|
|
"loss": 1.1108,
|
|
"step": 1686
|
|
},
|
|
{
|
|
"epoch": 0.8871943202734683,
|
|
"grad_norm": 2.6182503700256348,
|
|
"learning_rate": 9.562665611089008e-06,
|
|
"loss": 1.1216,
|
|
"step": 1687
|
|
},
|
|
{
|
|
"epoch": 0.887720220878254,
|
|
"grad_norm": 2.8653738498687744,
|
|
"learning_rate": 9.561953783117373e-06,
|
|
"loss": 1.1662,
|
|
"step": 1688
|
|
},
|
|
{
|
|
"epoch": 0.8882461214830397,
|
|
"grad_norm": 2.9791691303253174,
|
|
"learning_rate": 9.56124140285048e-06,
|
|
"loss": 1.2497,
|
|
"step": 1689
|
|
},
|
|
{
|
|
"epoch": 0.8887720220878254,
|
|
"grad_norm": 2.918632745742798,
|
|
"learning_rate": 9.560528470374571e-06,
|
|
"loss": 1.1975,
|
|
"step": 1690
|
|
},
|
|
{
|
|
"epoch": 0.8892979226926111,
|
|
"grad_norm": 2.7990922927856445,
|
|
"learning_rate": 9.559814985775958e-06,
|
|
"loss": 1.2148,
|
|
"step": 1691
|
|
},
|
|
{
|
|
"epoch": 0.8898238232973968,
|
|
"grad_norm": 2.925909996032715,
|
|
"learning_rate": 9.559100949141019e-06,
|
|
"loss": 1.16,
|
|
"step": 1692
|
|
},
|
|
{
|
|
"epoch": 0.8903497239021825,
|
|
"grad_norm": 2.641547679901123,
|
|
"learning_rate": 9.5583863605562e-06,
|
|
"loss": 1.1669,
|
|
"step": 1693
|
|
},
|
|
{
|
|
"epoch": 0.8908756245069682,
|
|
"grad_norm": 2.6737024784088135,
|
|
"learning_rate": 9.55767122010801e-06,
|
|
"loss": 1.2981,
|
|
"step": 1694
|
|
},
|
|
{
|
|
"epoch": 0.8914015251117539,
|
|
"grad_norm": 2.8731393814086914,
|
|
"learning_rate": 9.556955527883034e-06,
|
|
"loss": 1.1604,
|
|
"step": 1695
|
|
},
|
|
{
|
|
"epoch": 0.8919274257165396,
|
|
"grad_norm": 2.709575891494751,
|
|
"learning_rate": 9.556239283967914e-06,
|
|
"loss": 1.2394,
|
|
"step": 1696
|
|
},
|
|
{
|
|
"epoch": 0.8924533263213252,
|
|
"grad_norm": 2.790377616882324,
|
|
"learning_rate": 9.55552248844936e-06,
|
|
"loss": 1.3432,
|
|
"step": 1697
|
|
},
|
|
{
|
|
"epoch": 0.892979226926111,
|
|
"grad_norm": 2.919825553894043,
|
|
"learning_rate": 9.554805141414157e-06,
|
|
"loss": 1.3895,
|
|
"step": 1698
|
|
},
|
|
{
|
|
"epoch": 0.8935051275308966,
|
|
"grad_norm": 2.8806886672973633,
|
|
"learning_rate": 9.554087242949146e-06,
|
|
"loss": 1.1383,
|
|
"step": 1699
|
|
},
|
|
{
|
|
"epoch": 0.8940310281356824,
|
|
"grad_norm": 2.811005115509033,
|
|
"learning_rate": 9.553368793141242e-06,
|
|
"loss": 1.232,
|
|
"step": 1700
|
|
},
|
|
{
|
|
"epoch": 0.8945569287404681,
|
|
"grad_norm": 2.7031471729278564,
|
|
"learning_rate": 9.552649792077422e-06,
|
|
"loss": 1.1435,
|
|
"step": 1701
|
|
},
|
|
{
|
|
"epoch": 0.8950828293452537,
|
|
"grad_norm": 2.6482608318328857,
|
|
"learning_rate": 9.551930239844736e-06,
|
|
"loss": 1.1595,
|
|
"step": 1702
|
|
},
|
|
{
|
|
"epoch": 0.8956087299500395,
|
|
"grad_norm": 2.7931220531463623,
|
|
"learning_rate": 9.551210136530295e-06,
|
|
"loss": 1.1378,
|
|
"step": 1703
|
|
},
|
|
{
|
|
"epoch": 0.8961346305548251,
|
|
"grad_norm": 2.961080551147461,
|
|
"learning_rate": 9.550489482221279e-06,
|
|
"loss": 1.2329,
|
|
"step": 1704
|
|
},
|
|
{
|
|
"epoch": 0.8966605311596109,
|
|
"grad_norm": 2.9231629371643066,
|
|
"learning_rate": 9.549768277004934e-06,
|
|
"loss": 1.2753,
|
|
"step": 1705
|
|
},
|
|
{
|
|
"epoch": 0.8971864317643965,
|
|
"grad_norm": 2.8648524284362793,
|
|
"learning_rate": 9.549046520968575e-06,
|
|
"loss": 1.1821,
|
|
"step": 1706
|
|
},
|
|
{
|
|
"epoch": 0.8977123323691822,
|
|
"grad_norm": 2.7520124912261963,
|
|
"learning_rate": 9.548324214199578e-06,
|
|
"loss": 1.2729,
|
|
"step": 1707
|
|
},
|
|
{
|
|
"epoch": 0.8982382329739679,
|
|
"grad_norm": 2.6680541038513184,
|
|
"learning_rate": 9.547601356785393e-06,
|
|
"loss": 1.1954,
|
|
"step": 1708
|
|
},
|
|
{
|
|
"epoch": 0.8987641335787536,
|
|
"grad_norm": 2.842585325241089,
|
|
"learning_rate": 9.54687794881353e-06,
|
|
"loss": 1.2484,
|
|
"step": 1709
|
|
},
|
|
{
|
|
"epoch": 0.8992900341835394,
|
|
"grad_norm": 3.1276707649230957,
|
|
"learning_rate": 9.54615399037157e-06,
|
|
"loss": 1.2308,
|
|
"step": 1710
|
|
},
|
|
{
|
|
"epoch": 0.899815934788325,
|
|
"grad_norm": 2.8347909450531006,
|
|
"learning_rate": 9.545429481547161e-06,
|
|
"loss": 1.1784,
|
|
"step": 1711
|
|
},
|
|
{
|
|
"epoch": 0.9003418353931107,
|
|
"grad_norm": 2.6112828254699707,
|
|
"learning_rate": 9.544704422428016e-06,
|
|
"loss": 1.17,
|
|
"step": 1712
|
|
},
|
|
{
|
|
"epoch": 0.9008677359978964,
|
|
"grad_norm": 2.755176067352295,
|
|
"learning_rate": 9.543978813101912e-06,
|
|
"loss": 1.1255,
|
|
"step": 1713
|
|
},
|
|
{
|
|
"epoch": 0.9013936366026821,
|
|
"grad_norm": 2.7036972045898438,
|
|
"learning_rate": 9.543252653656697e-06,
|
|
"loss": 1.2038,
|
|
"step": 1714
|
|
},
|
|
{
|
|
"epoch": 0.9019195372074678,
|
|
"grad_norm": 2.702186107635498,
|
|
"learning_rate": 9.542525944180283e-06,
|
|
"loss": 1.1542,
|
|
"step": 1715
|
|
},
|
|
{
|
|
"epoch": 0.9024454378122535,
|
|
"grad_norm": 2.849114179611206,
|
|
"learning_rate": 9.54179868476065e-06,
|
|
"loss": 1.1788,
|
|
"step": 1716
|
|
},
|
|
{
|
|
"epoch": 0.9029713384170391,
|
|
"grad_norm": 2.810856580734253,
|
|
"learning_rate": 9.541070875485845e-06,
|
|
"loss": 1.2182,
|
|
"step": 1717
|
|
},
|
|
{
|
|
"epoch": 0.9034972390218249,
|
|
"grad_norm": 2.8117074966430664,
|
|
"learning_rate": 9.54034251644398e-06,
|
|
"loss": 1.1801,
|
|
"step": 1718
|
|
},
|
|
{
|
|
"epoch": 0.9040231396266105,
|
|
"grad_norm": 2.5521607398986816,
|
|
"learning_rate": 9.539613607723232e-06,
|
|
"loss": 1.1717,
|
|
"step": 1719
|
|
},
|
|
{
|
|
"epoch": 0.9045490402313963,
|
|
"grad_norm": 2.7161865234375,
|
|
"learning_rate": 9.53888414941185e-06,
|
|
"loss": 1.2331,
|
|
"step": 1720
|
|
},
|
|
{
|
|
"epoch": 0.905074940836182,
|
|
"grad_norm": 2.623772382736206,
|
|
"learning_rate": 9.538154141598144e-06,
|
|
"loss": 1.2407,
|
|
"step": 1721
|
|
},
|
|
{
|
|
"epoch": 0.9056008414409676,
|
|
"grad_norm": 3.033604145050049,
|
|
"learning_rate": 9.537423584370494e-06,
|
|
"loss": 1.304,
|
|
"step": 1722
|
|
},
|
|
{
|
|
"epoch": 0.9061267420457534,
|
|
"grad_norm": 2.7027854919433594,
|
|
"learning_rate": 9.536692477817345e-06,
|
|
"loss": 1.1964,
|
|
"step": 1723
|
|
},
|
|
{
|
|
"epoch": 0.906652642650539,
|
|
"grad_norm": 2.9221043586730957,
|
|
"learning_rate": 9.535960822027209e-06,
|
|
"loss": 1.2959,
|
|
"step": 1724
|
|
},
|
|
{
|
|
"epoch": 0.9071785432553248,
|
|
"grad_norm": 2.5912117958068848,
|
|
"learning_rate": 9.535228617088664e-06,
|
|
"loss": 1.1879,
|
|
"step": 1725
|
|
},
|
|
{
|
|
"epoch": 0.9077044438601104,
|
|
"grad_norm": 2.8192689418792725,
|
|
"learning_rate": 9.534495863090352e-06,
|
|
"loss": 1.3172,
|
|
"step": 1726
|
|
},
|
|
{
|
|
"epoch": 0.9082303444648961,
|
|
"grad_norm": 2.6915669441223145,
|
|
"learning_rate": 9.53376256012099e-06,
|
|
"loss": 1.1717,
|
|
"step": 1727
|
|
},
|
|
{
|
|
"epoch": 0.9087562450696818,
|
|
"grad_norm": 2.712204694747925,
|
|
"learning_rate": 9.533028708269352e-06,
|
|
"loss": 1.2305,
|
|
"step": 1728
|
|
},
|
|
{
|
|
"epoch": 0.9092821456744675,
|
|
"grad_norm": 2.6095693111419678,
|
|
"learning_rate": 9.532294307624281e-06,
|
|
"loss": 1.1862,
|
|
"step": 1729
|
|
},
|
|
{
|
|
"epoch": 0.9098080462792533,
|
|
"grad_norm": 3.140153169631958,
|
|
"learning_rate": 9.53155935827469e-06,
|
|
"loss": 1.2313,
|
|
"step": 1730
|
|
},
|
|
{
|
|
"epoch": 0.9103339468840389,
|
|
"grad_norm": 2.692613363265991,
|
|
"learning_rate": 9.530823860309557e-06,
|
|
"loss": 1.1831,
|
|
"step": 1731
|
|
},
|
|
{
|
|
"epoch": 0.9108598474888246,
|
|
"grad_norm": 2.744626760482788,
|
|
"learning_rate": 9.530087813817922e-06,
|
|
"loss": 1.1589,
|
|
"step": 1732
|
|
},
|
|
{
|
|
"epoch": 0.9113857480936103,
|
|
"grad_norm": 3.036444902420044,
|
|
"learning_rate": 9.529351218888899e-06,
|
|
"loss": 1.3099,
|
|
"step": 1733
|
|
},
|
|
{
|
|
"epoch": 0.911911648698396,
|
|
"grad_norm": 2.867050886154175,
|
|
"learning_rate": 9.528614075611659e-06,
|
|
"loss": 1.2377,
|
|
"step": 1734
|
|
},
|
|
{
|
|
"epoch": 0.9124375493031817,
|
|
"grad_norm": 2.810929536819458,
|
|
"learning_rate": 9.527876384075448e-06,
|
|
"loss": 1.3111,
|
|
"step": 1735
|
|
},
|
|
{
|
|
"epoch": 0.9129634499079674,
|
|
"grad_norm": 2.8564724922180176,
|
|
"learning_rate": 9.527138144369576e-06,
|
|
"loss": 1.1734,
|
|
"step": 1736
|
|
},
|
|
{
|
|
"epoch": 0.913489350512753,
|
|
"grad_norm": 2.699228286743164,
|
|
"learning_rate": 9.526399356583415e-06,
|
|
"loss": 1.2142,
|
|
"step": 1737
|
|
},
|
|
{
|
|
"epoch": 0.9140152511175388,
|
|
"grad_norm": 2.6879518032073975,
|
|
"learning_rate": 9.52566002080641e-06,
|
|
"loss": 1.2623,
|
|
"step": 1738
|
|
},
|
|
{
|
|
"epoch": 0.9145411517223245,
|
|
"grad_norm": 2.899137496948242,
|
|
"learning_rate": 9.524920137128067e-06,
|
|
"loss": 1.2559,
|
|
"step": 1739
|
|
},
|
|
{
|
|
"epoch": 0.9150670523271102,
|
|
"grad_norm": 2.6803410053253174,
|
|
"learning_rate": 9.524179705637961e-06,
|
|
"loss": 1.2264,
|
|
"step": 1740
|
|
},
|
|
{
|
|
"epoch": 0.9155929529318959,
|
|
"grad_norm": 2.637681245803833,
|
|
"learning_rate": 9.523438726425733e-06,
|
|
"loss": 1.1736,
|
|
"step": 1741
|
|
},
|
|
{
|
|
"epoch": 0.9161188535366815,
|
|
"grad_norm": 2.7320239543914795,
|
|
"learning_rate": 9.522697199581091e-06,
|
|
"loss": 1.1491,
|
|
"step": 1742
|
|
},
|
|
{
|
|
"epoch": 0.9166447541414673,
|
|
"grad_norm": 2.8139328956604004,
|
|
"learning_rate": 9.521955125193806e-06,
|
|
"loss": 1.2464,
|
|
"step": 1743
|
|
},
|
|
{
|
|
"epoch": 0.9171706547462529,
|
|
"grad_norm": 2.745562791824341,
|
|
"learning_rate": 9.521212503353719e-06,
|
|
"loss": 1.1256,
|
|
"step": 1744
|
|
},
|
|
{
|
|
"epoch": 0.9176965553510387,
|
|
"grad_norm": 3.0273523330688477,
|
|
"learning_rate": 9.520469334150735e-06,
|
|
"loss": 1.2942,
|
|
"step": 1745
|
|
},
|
|
{
|
|
"epoch": 0.9182224559558243,
|
|
"grad_norm": 3.156067132949829,
|
|
"learning_rate": 9.519725617674826e-06,
|
|
"loss": 1.2563,
|
|
"step": 1746
|
|
},
|
|
{
|
|
"epoch": 0.91874835656061,
|
|
"grad_norm": 2.7733898162841797,
|
|
"learning_rate": 9.518981354016032e-06,
|
|
"loss": 1.293,
|
|
"step": 1747
|
|
},
|
|
{
|
|
"epoch": 0.9192742571653958,
|
|
"grad_norm": 2.6913201808929443,
|
|
"learning_rate": 9.518236543264459e-06,
|
|
"loss": 1.1867,
|
|
"step": 1748
|
|
},
|
|
{
|
|
"epoch": 0.9198001577701814,
|
|
"grad_norm": 2.658437490463257,
|
|
"learning_rate": 9.517491185510272e-06,
|
|
"loss": 1.2938,
|
|
"step": 1749
|
|
},
|
|
{
|
|
"epoch": 0.9203260583749672,
|
|
"grad_norm": 2.6281015872955322,
|
|
"learning_rate": 9.516745280843716e-06,
|
|
"loss": 1.1781,
|
|
"step": 1750
|
|
},
|
|
{
|
|
"epoch": 0.9208519589797528,
|
|
"grad_norm": 2.7434194087982178,
|
|
"learning_rate": 9.515998829355086e-06,
|
|
"loss": 1.1395,
|
|
"step": 1751
|
|
},
|
|
{
|
|
"epoch": 0.9213778595845385,
|
|
"grad_norm": 2.733818292617798,
|
|
"learning_rate": 9.515251831134757e-06,
|
|
"loss": 1.1722,
|
|
"step": 1752
|
|
},
|
|
{
|
|
"epoch": 0.9219037601893242,
|
|
"grad_norm": 2.8211395740509033,
|
|
"learning_rate": 9.514504286273165e-06,
|
|
"loss": 1.2039,
|
|
"step": 1753
|
|
},
|
|
{
|
|
"epoch": 0.9224296607941099,
|
|
"grad_norm": 2.6792776584625244,
|
|
"learning_rate": 9.51375619486081e-06,
|
|
"loss": 1.1356,
|
|
"step": 1754
|
|
},
|
|
{
|
|
"epoch": 0.9229555613988956,
|
|
"grad_norm": 2.739906072616577,
|
|
"learning_rate": 9.513007556988259e-06,
|
|
"loss": 1.2448,
|
|
"step": 1755
|
|
},
|
|
{
|
|
"epoch": 0.9234814620036813,
|
|
"grad_norm": 2.8193283081054688,
|
|
"learning_rate": 9.512258372746148e-06,
|
|
"loss": 1.3139,
|
|
"step": 1756
|
|
},
|
|
{
|
|
"epoch": 0.924007362608467,
|
|
"grad_norm": 2.7297017574310303,
|
|
"learning_rate": 9.511508642225177e-06,
|
|
"loss": 1.1189,
|
|
"step": 1757
|
|
},
|
|
{
|
|
"epoch": 0.9245332632132527,
|
|
"grad_norm": 2.986417770385742,
|
|
"learning_rate": 9.510758365516113e-06,
|
|
"loss": 1.2692,
|
|
"step": 1758
|
|
},
|
|
{
|
|
"epoch": 0.9250591638180384,
|
|
"grad_norm": 2.79677414894104,
|
|
"learning_rate": 9.510007542709789e-06,
|
|
"loss": 1.3251,
|
|
"step": 1759
|
|
},
|
|
{
|
|
"epoch": 0.9255850644228241,
|
|
"grad_norm": 2.8102521896362305,
|
|
"learning_rate": 9.509256173897103e-06,
|
|
"loss": 1.2665,
|
|
"step": 1760
|
|
},
|
|
{
|
|
"epoch": 0.9261109650276098,
|
|
"grad_norm": 2.7303950786590576,
|
|
"learning_rate": 9.508504259169019e-06,
|
|
"loss": 1.1804,
|
|
"step": 1761
|
|
},
|
|
{
|
|
"epoch": 0.9266368656323954,
|
|
"grad_norm": 2.836937665939331,
|
|
"learning_rate": 9.507751798616568e-06,
|
|
"loss": 1.2308,
|
|
"step": 1762
|
|
},
|
|
{
|
|
"epoch": 0.9271627662371812,
|
|
"grad_norm": 2.6794517040252686,
|
|
"learning_rate": 9.506998792330848e-06,
|
|
"loss": 1.2656,
|
|
"step": 1763
|
|
},
|
|
{
|
|
"epoch": 0.9276886668419668,
|
|
"grad_norm": 2.734978199005127,
|
|
"learning_rate": 9.506245240403021e-06,
|
|
"loss": 1.2584,
|
|
"step": 1764
|
|
},
|
|
{
|
|
"epoch": 0.9282145674467526,
|
|
"grad_norm": 2.639958381652832,
|
|
"learning_rate": 9.505491142924317e-06,
|
|
"loss": 1.1946,
|
|
"step": 1765
|
|
},
|
|
{
|
|
"epoch": 0.9287404680515382,
|
|
"grad_norm": 2.922431707382202,
|
|
"learning_rate": 9.504736499986033e-06,
|
|
"loss": 1.1951,
|
|
"step": 1766
|
|
},
|
|
{
|
|
"epoch": 0.929266368656324,
|
|
"grad_norm": 2.720578193664551,
|
|
"learning_rate": 9.503981311679526e-06,
|
|
"loss": 1.2109,
|
|
"step": 1767
|
|
},
|
|
{
|
|
"epoch": 0.9297922692611097,
|
|
"grad_norm": 2.9377944469451904,
|
|
"learning_rate": 9.503225578096228e-06,
|
|
"loss": 1.2838,
|
|
"step": 1768
|
|
},
|
|
{
|
|
"epoch": 0.9303181698658953,
|
|
"grad_norm": 2.8547558784484863,
|
|
"learning_rate": 9.50246929932763e-06,
|
|
"loss": 1.2121,
|
|
"step": 1769
|
|
},
|
|
{
|
|
"epoch": 0.9308440704706811,
|
|
"grad_norm": 2.966747283935547,
|
|
"learning_rate": 9.50171247546529e-06,
|
|
"loss": 1.3048,
|
|
"step": 1770
|
|
},
|
|
{
|
|
"epoch": 0.9313699710754667,
|
|
"grad_norm": 2.700716495513916,
|
|
"learning_rate": 9.500955106600832e-06,
|
|
"loss": 1.1526,
|
|
"step": 1771
|
|
},
|
|
{
|
|
"epoch": 0.9318958716802525,
|
|
"grad_norm": 2.661173105239868,
|
|
"learning_rate": 9.500197192825952e-06,
|
|
"loss": 1.2061,
|
|
"step": 1772
|
|
},
|
|
{
|
|
"epoch": 0.9324217722850381,
|
|
"grad_norm": 2.9022488594055176,
|
|
"learning_rate": 9.499438734232407e-06,
|
|
"loss": 1.1354,
|
|
"step": 1773
|
|
},
|
|
{
|
|
"epoch": 0.9329476728898238,
|
|
"grad_norm": 2.791740655899048,
|
|
"learning_rate": 9.498679730912015e-06,
|
|
"loss": 1.3512,
|
|
"step": 1774
|
|
},
|
|
{
|
|
"epoch": 0.9334735734946095,
|
|
"grad_norm": 2.462069034576416,
|
|
"learning_rate": 9.497920182956671e-06,
|
|
"loss": 1.1391,
|
|
"step": 1775
|
|
},
|
|
{
|
|
"epoch": 0.9339994740993952,
|
|
"grad_norm": 2.671215295791626,
|
|
"learning_rate": 9.497160090458325e-06,
|
|
"loss": 1.155,
|
|
"step": 1776
|
|
},
|
|
{
|
|
"epoch": 0.934525374704181,
|
|
"grad_norm": 2.7096974849700928,
|
|
"learning_rate": 9.496399453509e-06,
|
|
"loss": 1.1648,
|
|
"step": 1777
|
|
},
|
|
{
|
|
"epoch": 0.9350512753089666,
|
|
"grad_norm": 2.70662260055542,
|
|
"learning_rate": 9.495638272200785e-06,
|
|
"loss": 1.0678,
|
|
"step": 1778
|
|
},
|
|
{
|
|
"epoch": 0.9355771759137523,
|
|
"grad_norm": 2.7821078300476074,
|
|
"learning_rate": 9.494876546625828e-06,
|
|
"loss": 1.3183,
|
|
"step": 1779
|
|
},
|
|
{
|
|
"epoch": 0.936103076518538,
|
|
"grad_norm": 2.832709312438965,
|
|
"learning_rate": 9.494114276876354e-06,
|
|
"loss": 1.1714,
|
|
"step": 1780
|
|
},
|
|
{
|
|
"epoch": 0.9366289771233237,
|
|
"grad_norm": 2.9377658367156982,
|
|
"learning_rate": 9.493351463044642e-06,
|
|
"loss": 1.1614,
|
|
"step": 1781
|
|
},
|
|
{
|
|
"epoch": 0.9371548777281093,
|
|
"grad_norm": 2.7838244438171387,
|
|
"learning_rate": 9.492588105223046e-06,
|
|
"loss": 1.0868,
|
|
"step": 1782
|
|
},
|
|
{
|
|
"epoch": 0.9376807783328951,
|
|
"grad_norm": 2.7289419174194336,
|
|
"learning_rate": 9.49182420350398e-06,
|
|
"loss": 1.2301,
|
|
"step": 1783
|
|
},
|
|
{
|
|
"epoch": 0.9382066789376807,
|
|
"grad_norm": 2.851503610610962,
|
|
"learning_rate": 9.491059757979927e-06,
|
|
"loss": 1.2603,
|
|
"step": 1784
|
|
},
|
|
{
|
|
"epoch": 0.9387325795424665,
|
|
"grad_norm": 2.735098123550415,
|
|
"learning_rate": 9.490294768743435e-06,
|
|
"loss": 1.1653,
|
|
"step": 1785
|
|
},
|
|
{
|
|
"epoch": 0.9392584801472522,
|
|
"grad_norm": 2.758702278137207,
|
|
"learning_rate": 9.48952923588712e-06,
|
|
"loss": 1.2216,
|
|
"step": 1786
|
|
},
|
|
{
|
|
"epoch": 0.9397843807520379,
|
|
"grad_norm": 2.878904342651367,
|
|
"learning_rate": 9.488763159503658e-06,
|
|
"loss": 1.1555,
|
|
"step": 1787
|
|
},
|
|
{
|
|
"epoch": 0.9403102813568236,
|
|
"grad_norm": 2.578371286392212,
|
|
"learning_rate": 9.487996539685796e-06,
|
|
"loss": 1.2949,
|
|
"step": 1788
|
|
},
|
|
{
|
|
"epoch": 0.9408361819616092,
|
|
"grad_norm": 2.8806192874908447,
|
|
"learning_rate": 9.487229376526344e-06,
|
|
"loss": 1.183,
|
|
"step": 1789
|
|
},
|
|
{
|
|
"epoch": 0.941362082566395,
|
|
"grad_norm": 2.727499485015869,
|
|
"learning_rate": 9.486461670118183e-06,
|
|
"loss": 1.1664,
|
|
"step": 1790
|
|
},
|
|
{
|
|
"epoch": 0.9418879831711806,
|
|
"grad_norm": 2.5423619747161865,
|
|
"learning_rate": 9.485693420554251e-06,
|
|
"loss": 1.1919,
|
|
"step": 1791
|
|
},
|
|
{
|
|
"epoch": 0.9424138837759664,
|
|
"grad_norm": 2.8886513710021973,
|
|
"learning_rate": 9.484924627927559e-06,
|
|
"loss": 1.1801,
|
|
"step": 1792
|
|
},
|
|
{
|
|
"epoch": 0.942939784380752,
|
|
"grad_norm": 3.0344011783599854,
|
|
"learning_rate": 9.48415529233118e-06,
|
|
"loss": 1.1934,
|
|
"step": 1793
|
|
},
|
|
{
|
|
"epoch": 0.9434656849855377,
|
|
"grad_norm": 2.6005403995513916,
|
|
"learning_rate": 9.483385413858255e-06,
|
|
"loss": 1.1283,
|
|
"step": 1794
|
|
},
|
|
{
|
|
"epoch": 0.9439915855903235,
|
|
"grad_norm": 2.687229633331299,
|
|
"learning_rate": 9.482614992601991e-06,
|
|
"loss": 1.2185,
|
|
"step": 1795
|
|
},
|
|
{
|
|
"epoch": 0.9445174861951091,
|
|
"grad_norm": 2.8815500736236572,
|
|
"learning_rate": 9.481844028655656e-06,
|
|
"loss": 1.2021,
|
|
"step": 1796
|
|
},
|
|
{
|
|
"epoch": 0.9450433867998949,
|
|
"grad_norm": 2.815253734588623,
|
|
"learning_rate": 9.481072522112589e-06,
|
|
"loss": 1.2677,
|
|
"step": 1797
|
|
},
|
|
{
|
|
"epoch": 0.9455692874046805,
|
|
"grad_norm": 2.5723671913146973,
|
|
"learning_rate": 9.480300473066195e-06,
|
|
"loss": 1.2222,
|
|
"step": 1798
|
|
},
|
|
{
|
|
"epoch": 0.9460951880094662,
|
|
"grad_norm": 2.702446937561035,
|
|
"learning_rate": 9.479527881609939e-06,
|
|
"loss": 1.2078,
|
|
"step": 1799
|
|
},
|
|
{
|
|
"epoch": 0.9466210886142519,
|
|
"grad_norm": 2.589468240737915,
|
|
"learning_rate": 9.478754747837355e-06,
|
|
"loss": 1.0752,
|
|
"step": 1800
|
|
},
|
|
{
|
|
"epoch": 0.9471469892190376,
|
|
"grad_norm": 2.7270820140838623,
|
|
"learning_rate": 9.477981071842048e-06,
|
|
"loss": 1.1876,
|
|
"step": 1801
|
|
},
|
|
{
|
|
"epoch": 0.9476728898238233,
|
|
"grad_norm": 2.7801074981689453,
|
|
"learning_rate": 9.477206853717678e-06,
|
|
"loss": 1.2658,
|
|
"step": 1802
|
|
},
|
|
{
|
|
"epoch": 0.948198790428609,
|
|
"grad_norm": 2.6370134353637695,
|
|
"learning_rate": 9.476432093557976e-06,
|
|
"loss": 1.2205,
|
|
"step": 1803
|
|
},
|
|
{
|
|
"epoch": 0.9487246910333947,
|
|
"grad_norm": 3.233083724975586,
|
|
"learning_rate": 9.475656791456745e-06,
|
|
"loss": 1.3612,
|
|
"step": 1804
|
|
},
|
|
{
|
|
"epoch": 0.9492505916381804,
|
|
"grad_norm": 2.825835704803467,
|
|
"learning_rate": 9.474880947507841e-06,
|
|
"loss": 1.1251,
|
|
"step": 1805
|
|
},
|
|
{
|
|
"epoch": 0.9497764922429661,
|
|
"grad_norm": 2.539668560028076,
|
|
"learning_rate": 9.474104561805192e-06,
|
|
"loss": 1.1156,
|
|
"step": 1806
|
|
},
|
|
{
|
|
"epoch": 0.9503023928477518,
|
|
"grad_norm": 2.6709625720977783,
|
|
"learning_rate": 9.473327634442797e-06,
|
|
"loss": 1.1732,
|
|
"step": 1807
|
|
},
|
|
{
|
|
"epoch": 0.9508282934525375,
|
|
"grad_norm": 2.843092441558838,
|
|
"learning_rate": 9.47255016551471e-06,
|
|
"loss": 1.1526,
|
|
"step": 1808
|
|
},
|
|
{
|
|
"epoch": 0.9513541940573231,
|
|
"grad_norm": 2.7349181175231934,
|
|
"learning_rate": 9.471772155115058e-06,
|
|
"loss": 1.1036,
|
|
"step": 1809
|
|
},
|
|
{
|
|
"epoch": 0.9518800946621089,
|
|
"grad_norm": 2.7796084880828857,
|
|
"learning_rate": 9.47099360333803e-06,
|
|
"loss": 1.2184,
|
|
"step": 1810
|
|
},
|
|
{
|
|
"epoch": 0.9524059952668945,
|
|
"grad_norm": 2.6712260246276855,
|
|
"learning_rate": 9.470214510277882e-06,
|
|
"loss": 1.1489,
|
|
"step": 1811
|
|
},
|
|
{
|
|
"epoch": 0.9529318958716803,
|
|
"grad_norm": 2.596186876296997,
|
|
"learning_rate": 9.469434876028937e-06,
|
|
"loss": 1.0813,
|
|
"step": 1812
|
|
},
|
|
{
|
|
"epoch": 0.953457796476466,
|
|
"grad_norm": 2.816624402999878,
|
|
"learning_rate": 9.468654700685578e-06,
|
|
"loss": 1.2567,
|
|
"step": 1813
|
|
},
|
|
{
|
|
"epoch": 0.9539836970812516,
|
|
"grad_norm": 2.8059024810791016,
|
|
"learning_rate": 9.467873984342262e-06,
|
|
"loss": 1.1788,
|
|
"step": 1814
|
|
},
|
|
{
|
|
"epoch": 0.9545095976860374,
|
|
"grad_norm": 2.7334797382354736,
|
|
"learning_rate": 9.467092727093505e-06,
|
|
"loss": 1.2629,
|
|
"step": 1815
|
|
},
|
|
{
|
|
"epoch": 0.955035498290823,
|
|
"grad_norm": 2.774820566177368,
|
|
"learning_rate": 9.466310929033889e-06,
|
|
"loss": 1.2188,
|
|
"step": 1816
|
|
},
|
|
{
|
|
"epoch": 0.9555613988956088,
|
|
"grad_norm": 2.794215202331543,
|
|
"learning_rate": 9.465528590258063e-06,
|
|
"loss": 1.2917,
|
|
"step": 1817
|
|
},
|
|
{
|
|
"epoch": 0.9560872995003944,
|
|
"grad_norm": 2.7117912769317627,
|
|
"learning_rate": 9.464745710860743e-06,
|
|
"loss": 1.1912,
|
|
"step": 1818
|
|
},
|
|
{
|
|
"epoch": 0.9566132001051801,
|
|
"grad_norm": 2.7747268676757812,
|
|
"learning_rate": 9.463962290936707e-06,
|
|
"loss": 1.1096,
|
|
"step": 1819
|
|
},
|
|
{
|
|
"epoch": 0.9571391007099658,
|
|
"grad_norm": 2.9322032928466797,
|
|
"learning_rate": 9.4631783305808e-06,
|
|
"loss": 1.3064,
|
|
"step": 1820
|
|
},
|
|
{
|
|
"epoch": 0.9576650013147515,
|
|
"grad_norm": 2.687162160873413,
|
|
"learning_rate": 9.462393829887935e-06,
|
|
"loss": 1.2218,
|
|
"step": 1821
|
|
},
|
|
{
|
|
"epoch": 0.9581909019195372,
|
|
"grad_norm": 2.6757469177246094,
|
|
"learning_rate": 9.461608788953085e-06,
|
|
"loss": 1.2643,
|
|
"step": 1822
|
|
},
|
|
{
|
|
"epoch": 0.9587168025243229,
|
|
"grad_norm": 2.718979597091675,
|
|
"learning_rate": 9.460823207871295e-06,
|
|
"loss": 1.1976,
|
|
"step": 1823
|
|
},
|
|
{
|
|
"epoch": 0.9592427031291086,
|
|
"grad_norm": 2.878077745437622,
|
|
"learning_rate": 9.460037086737667e-06,
|
|
"loss": 1.2524,
|
|
"step": 1824
|
|
},
|
|
{
|
|
"epoch": 0.9597686037338943,
|
|
"grad_norm": 2.6501734256744385,
|
|
"learning_rate": 9.459250425647376e-06,
|
|
"loss": 1.2038,
|
|
"step": 1825
|
|
},
|
|
{
|
|
"epoch": 0.96029450433868,
|
|
"grad_norm": 2.6035494804382324,
|
|
"learning_rate": 9.45846322469566e-06,
|
|
"loss": 1.1412,
|
|
"step": 1826
|
|
},
|
|
{
|
|
"epoch": 0.9608204049434657,
|
|
"grad_norm": 2.589569568634033,
|
|
"learning_rate": 9.45767548397782e-06,
|
|
"loss": 1.1572,
|
|
"step": 1827
|
|
},
|
|
{
|
|
"epoch": 0.9613463055482514,
|
|
"grad_norm": 2.6510744094848633,
|
|
"learning_rate": 9.456887203589226e-06,
|
|
"loss": 1.2127,
|
|
"step": 1828
|
|
},
|
|
{
|
|
"epoch": 0.961872206153037,
|
|
"grad_norm": 2.813180446624756,
|
|
"learning_rate": 9.45609838362531e-06,
|
|
"loss": 1.0575,
|
|
"step": 1829
|
|
},
|
|
{
|
|
"epoch": 0.9623981067578228,
|
|
"grad_norm": 2.9063849449157715,
|
|
"learning_rate": 9.455309024181573e-06,
|
|
"loss": 1.2185,
|
|
"step": 1830
|
|
},
|
|
{
|
|
"epoch": 0.9629240073626084,
|
|
"grad_norm": 2.73730206489563,
|
|
"learning_rate": 9.454519125353577e-06,
|
|
"loss": 1.2147,
|
|
"step": 1831
|
|
},
|
|
{
|
|
"epoch": 0.9634499079673942,
|
|
"grad_norm": 2.669572591781616,
|
|
"learning_rate": 9.453728687236953e-06,
|
|
"loss": 1.2742,
|
|
"step": 1832
|
|
},
|
|
{
|
|
"epoch": 0.9639758085721799,
|
|
"grad_norm": 2.7116668224334717,
|
|
"learning_rate": 9.452937709927395e-06,
|
|
"loss": 1.3213,
|
|
"step": 1833
|
|
},
|
|
{
|
|
"epoch": 0.9645017091769655,
|
|
"grad_norm": 2.6835808753967285,
|
|
"learning_rate": 9.452146193520663e-06,
|
|
"loss": 1.1738,
|
|
"step": 1834
|
|
},
|
|
{
|
|
"epoch": 0.9650276097817513,
|
|
"grad_norm": 2.710160493850708,
|
|
"learning_rate": 9.451354138112583e-06,
|
|
"loss": 1.2754,
|
|
"step": 1835
|
|
},
|
|
{
|
|
"epoch": 0.9655535103865369,
|
|
"grad_norm": 2.8004534244537354,
|
|
"learning_rate": 9.450561543799044e-06,
|
|
"loss": 1.2435,
|
|
"step": 1836
|
|
},
|
|
{
|
|
"epoch": 0.9660794109913227,
|
|
"grad_norm": 2.7181503772735596,
|
|
"learning_rate": 9.449768410676004e-06,
|
|
"loss": 1.1298,
|
|
"step": 1837
|
|
},
|
|
{
|
|
"epoch": 0.9666053115961083,
|
|
"grad_norm": 2.6982367038726807,
|
|
"learning_rate": 9.448974738839483e-06,
|
|
"loss": 1.2353,
|
|
"step": 1838
|
|
},
|
|
{
|
|
"epoch": 0.967131212200894,
|
|
"grad_norm": 2.7241265773773193,
|
|
"learning_rate": 9.448180528385569e-06,
|
|
"loss": 1.2734,
|
|
"step": 1839
|
|
},
|
|
{
|
|
"epoch": 0.9676571128056797,
|
|
"grad_norm": 2.6802639961242676,
|
|
"learning_rate": 9.44738577941041e-06,
|
|
"loss": 1.1996,
|
|
"step": 1840
|
|
},
|
|
{
|
|
"epoch": 0.9681830134104654,
|
|
"grad_norm": 2.829801559448242,
|
|
"learning_rate": 9.446590492010226e-06,
|
|
"loss": 1.2567,
|
|
"step": 1841
|
|
},
|
|
{
|
|
"epoch": 0.9687089140152512,
|
|
"grad_norm": 2.7040324211120605,
|
|
"learning_rate": 9.445794666281296e-06,
|
|
"loss": 1.2764,
|
|
"step": 1842
|
|
},
|
|
{
|
|
"epoch": 0.9692348146200368,
|
|
"grad_norm": 2.7584850788116455,
|
|
"learning_rate": 9.444998302319967e-06,
|
|
"loss": 1.1367,
|
|
"step": 1843
|
|
},
|
|
{
|
|
"epoch": 0.9697607152248225,
|
|
"grad_norm": 2.7660248279571533,
|
|
"learning_rate": 9.444201400222657e-06,
|
|
"loss": 1.2053,
|
|
"step": 1844
|
|
},
|
|
{
|
|
"epoch": 0.9702866158296082,
|
|
"grad_norm": 3.078641653060913,
|
|
"learning_rate": 9.443403960085835e-06,
|
|
"loss": 1.2846,
|
|
"step": 1845
|
|
},
|
|
{
|
|
"epoch": 0.9708125164343939,
|
|
"grad_norm": 2.6587343215942383,
|
|
"learning_rate": 9.44260598200605e-06,
|
|
"loss": 1.1719,
|
|
"step": 1846
|
|
},
|
|
{
|
|
"epoch": 0.9713384170391796,
|
|
"grad_norm": 2.7403454780578613,
|
|
"learning_rate": 9.441807466079907e-06,
|
|
"loss": 1.2232,
|
|
"step": 1847
|
|
},
|
|
{
|
|
"epoch": 0.9718643176439653,
|
|
"grad_norm": 2.625275135040283,
|
|
"learning_rate": 9.441008412404077e-06,
|
|
"loss": 1.1232,
|
|
"step": 1848
|
|
},
|
|
{
|
|
"epoch": 0.9723902182487509,
|
|
"grad_norm": 2.6796512603759766,
|
|
"learning_rate": 9.440208821075302e-06,
|
|
"loss": 1.1821,
|
|
"step": 1849
|
|
},
|
|
{
|
|
"epoch": 0.9729161188535367,
|
|
"grad_norm": 2.8642115592956543,
|
|
"learning_rate": 9.439408692190383e-06,
|
|
"loss": 1.3042,
|
|
"step": 1850
|
|
},
|
|
{
|
|
"epoch": 0.9734420194583224,
|
|
"grad_norm": 2.7071051597595215,
|
|
"learning_rate": 9.438608025846185e-06,
|
|
"loss": 1.1399,
|
|
"step": 1851
|
|
},
|
|
{
|
|
"epoch": 0.9739679200631081,
|
|
"grad_norm": 2.786220073699951,
|
|
"learning_rate": 9.437806822139646e-06,
|
|
"loss": 1.1746,
|
|
"step": 1852
|
|
},
|
|
{
|
|
"epoch": 0.9744938206678938,
|
|
"grad_norm": 2.838064193725586,
|
|
"learning_rate": 9.437005081167762e-06,
|
|
"loss": 1.1721,
|
|
"step": 1853
|
|
},
|
|
{
|
|
"epoch": 0.9750197212726794,
|
|
"grad_norm": 2.6426138877868652,
|
|
"learning_rate": 9.436202803027595e-06,
|
|
"loss": 1.2819,
|
|
"step": 1854
|
|
},
|
|
{
|
|
"epoch": 0.9755456218774652,
|
|
"grad_norm": 2.7467615604400635,
|
|
"learning_rate": 9.435399987816274e-06,
|
|
"loss": 1.2588,
|
|
"step": 1855
|
|
},
|
|
{
|
|
"epoch": 0.9760715224822508,
|
|
"grad_norm": 2.631035804748535,
|
|
"learning_rate": 9.434596635630995e-06,
|
|
"loss": 1.1511,
|
|
"step": 1856
|
|
},
|
|
{
|
|
"epoch": 0.9765974230870366,
|
|
"grad_norm": 2.7912638187408447,
|
|
"learning_rate": 9.433792746569011e-06,
|
|
"loss": 1.0962,
|
|
"step": 1857
|
|
},
|
|
{
|
|
"epoch": 0.9771233236918222,
|
|
"grad_norm": 2.820770025253296,
|
|
"learning_rate": 9.43298832072765e-06,
|
|
"loss": 1.2672,
|
|
"step": 1858
|
|
},
|
|
{
|
|
"epoch": 0.9776492242966079,
|
|
"grad_norm": 2.7003540992736816,
|
|
"learning_rate": 9.432183358204297e-06,
|
|
"loss": 1.1681,
|
|
"step": 1859
|
|
},
|
|
{
|
|
"epoch": 0.9781751249013937,
|
|
"grad_norm": 2.645509958267212,
|
|
"learning_rate": 9.431377859096407e-06,
|
|
"loss": 1.2267,
|
|
"step": 1860
|
|
},
|
|
{
|
|
"epoch": 0.9787010255061793,
|
|
"grad_norm": 2.822660446166992,
|
|
"learning_rate": 9.430571823501498e-06,
|
|
"loss": 1.2248,
|
|
"step": 1861
|
|
},
|
|
{
|
|
"epoch": 0.9792269261109651,
|
|
"grad_norm": 2.5736472606658936,
|
|
"learning_rate": 9.429765251517152e-06,
|
|
"loss": 1.1422,
|
|
"step": 1862
|
|
},
|
|
{
|
|
"epoch": 0.9797528267157507,
|
|
"grad_norm": 2.647050619125366,
|
|
"learning_rate": 9.428958143241017e-06,
|
|
"loss": 1.2432,
|
|
"step": 1863
|
|
},
|
|
{
|
|
"epoch": 0.9802787273205364,
|
|
"grad_norm": 2.5553689002990723,
|
|
"learning_rate": 9.42815049877081e-06,
|
|
"loss": 1.1576,
|
|
"step": 1864
|
|
},
|
|
{
|
|
"epoch": 0.9808046279253221,
|
|
"grad_norm": 2.692744255065918,
|
|
"learning_rate": 9.427342318204302e-06,
|
|
"loss": 1.2077,
|
|
"step": 1865
|
|
},
|
|
{
|
|
"epoch": 0.9813305285301078,
|
|
"grad_norm": 2.4748384952545166,
|
|
"learning_rate": 9.426533601639341e-06,
|
|
"loss": 1.1213,
|
|
"step": 1866
|
|
},
|
|
{
|
|
"epoch": 0.9818564291348935,
|
|
"grad_norm": 2.8118677139282227,
|
|
"learning_rate": 9.425724349173833e-06,
|
|
"loss": 1.2363,
|
|
"step": 1867
|
|
},
|
|
{
|
|
"epoch": 0.9823823297396792,
|
|
"grad_norm": 2.794780731201172,
|
|
"learning_rate": 9.424914560905751e-06,
|
|
"loss": 1.2495,
|
|
"step": 1868
|
|
},
|
|
{
|
|
"epoch": 0.982908230344465,
|
|
"grad_norm": 2.575608491897583,
|
|
"learning_rate": 9.424104236933131e-06,
|
|
"loss": 1.0972,
|
|
"step": 1869
|
|
},
|
|
{
|
|
"epoch": 0.9834341309492506,
|
|
"grad_norm": 2.74013614654541,
|
|
"learning_rate": 9.423293377354077e-06,
|
|
"loss": 1.2093,
|
|
"step": 1870
|
|
},
|
|
{
|
|
"epoch": 0.9839600315540363,
|
|
"grad_norm": 2.6837282180786133,
|
|
"learning_rate": 9.422481982266756e-06,
|
|
"loss": 1.2029,
|
|
"step": 1871
|
|
},
|
|
{
|
|
"epoch": 0.984485932158822,
|
|
"grad_norm": 2.7307164669036865,
|
|
"learning_rate": 9.4216700517694e-06,
|
|
"loss": 1.1545,
|
|
"step": 1872
|
|
},
|
|
{
|
|
"epoch": 0.9850118327636077,
|
|
"grad_norm": 2.6432669162750244,
|
|
"learning_rate": 9.420857585960304e-06,
|
|
"loss": 1.1801,
|
|
"step": 1873
|
|
},
|
|
{
|
|
"epoch": 0.9855377333683933,
|
|
"grad_norm": 2.5711894035339355,
|
|
"learning_rate": 9.420044584937831e-06,
|
|
"loss": 1.2286,
|
|
"step": 1874
|
|
},
|
|
{
|
|
"epoch": 0.9860636339731791,
|
|
"grad_norm": 2.4743969440460205,
|
|
"learning_rate": 9.419231048800408e-06,
|
|
"loss": 1.0981,
|
|
"step": 1875
|
|
},
|
|
{
|
|
"epoch": 0.9865895345779647,
|
|
"grad_norm": 2.8026185035705566,
|
|
"learning_rate": 9.418416977646525e-06,
|
|
"loss": 1.3001,
|
|
"step": 1876
|
|
},
|
|
{
|
|
"epoch": 0.9871154351827505,
|
|
"grad_norm": 2.6530983448028564,
|
|
"learning_rate": 9.417602371574736e-06,
|
|
"loss": 1.1737,
|
|
"step": 1877
|
|
},
|
|
{
|
|
"epoch": 0.9876413357875361,
|
|
"grad_norm": 2.53423810005188,
|
|
"learning_rate": 9.416787230683669e-06,
|
|
"loss": 1.1396,
|
|
"step": 1878
|
|
},
|
|
{
|
|
"epoch": 0.9881672363923218,
|
|
"grad_norm": 2.8699934482574463,
|
|
"learning_rate": 9.415971555072002e-06,
|
|
"loss": 1.285,
|
|
"step": 1879
|
|
},
|
|
{
|
|
"epoch": 0.9886931369971076,
|
|
"grad_norm": 2.6265554428100586,
|
|
"learning_rate": 9.415155344838486e-06,
|
|
"loss": 1.1935,
|
|
"step": 1880
|
|
},
|
|
{
|
|
"epoch": 0.9892190376018932,
|
|
"grad_norm": 2.727611541748047,
|
|
"learning_rate": 9.414338600081938e-06,
|
|
"loss": 1.1886,
|
|
"step": 1881
|
|
},
|
|
{
|
|
"epoch": 0.989744938206679,
|
|
"grad_norm": 2.7758800983428955,
|
|
"learning_rate": 9.413521320901239e-06,
|
|
"loss": 1.1854,
|
|
"step": 1882
|
|
},
|
|
{
|
|
"epoch": 0.9902708388114646,
|
|
"grad_norm": 2.671818494796753,
|
|
"learning_rate": 9.412703507395331e-06,
|
|
"loss": 1.2591,
|
|
"step": 1883
|
|
},
|
|
{
|
|
"epoch": 0.9907967394162503,
|
|
"grad_norm": 2.907283067703247,
|
|
"learning_rate": 9.411885159663223e-06,
|
|
"loss": 1.3222,
|
|
"step": 1884
|
|
},
|
|
{
|
|
"epoch": 0.991322640021036,
|
|
"grad_norm": 2.5878562927246094,
|
|
"learning_rate": 9.411066277803988e-06,
|
|
"loss": 1.1456,
|
|
"step": 1885
|
|
},
|
|
{
|
|
"epoch": 0.9918485406258217,
|
|
"grad_norm": 2.639932870864868,
|
|
"learning_rate": 9.410246861916767e-06,
|
|
"loss": 1.2065,
|
|
"step": 1886
|
|
},
|
|
{
|
|
"epoch": 0.9923744412306074,
|
|
"grad_norm": 2.9076333045959473,
|
|
"learning_rate": 9.409426912100762e-06,
|
|
"loss": 1.2487,
|
|
"step": 1887
|
|
},
|
|
{
|
|
"epoch": 0.9929003418353931,
|
|
"grad_norm": 2.6688828468322754,
|
|
"learning_rate": 9.40860642845524e-06,
|
|
"loss": 1.2603,
|
|
"step": 1888
|
|
},
|
|
{
|
|
"epoch": 0.9934262424401789,
|
|
"grad_norm": 2.7303013801574707,
|
|
"learning_rate": 9.407785411079534e-06,
|
|
"loss": 1.2114,
|
|
"step": 1889
|
|
},
|
|
{
|
|
"epoch": 0.9939521430449645,
|
|
"grad_norm": 2.608018636703491,
|
|
"learning_rate": 9.40696386007304e-06,
|
|
"loss": 1.1125,
|
|
"step": 1890
|
|
},
|
|
{
|
|
"epoch": 0.9944780436497502,
|
|
"grad_norm": 2.5597963333129883,
|
|
"learning_rate": 9.406141775535218e-06,
|
|
"loss": 1.1237,
|
|
"step": 1891
|
|
},
|
|
{
|
|
"epoch": 0.9950039442545359,
|
|
"grad_norm": 2.7863688468933105,
|
|
"learning_rate": 9.405319157565597e-06,
|
|
"loss": 1.2916,
|
|
"step": 1892
|
|
},
|
|
{
|
|
"epoch": 0.9955298448593216,
|
|
"grad_norm": 2.7089035511016846,
|
|
"learning_rate": 9.404496006263768e-06,
|
|
"loss": 1.2233,
|
|
"step": 1893
|
|
},
|
|
{
|
|
"epoch": 0.9960557454641072,
|
|
"grad_norm": 2.6158456802368164,
|
|
"learning_rate": 9.403672321729383e-06,
|
|
"loss": 1.1048,
|
|
"step": 1894
|
|
},
|
|
{
|
|
"epoch": 0.996581646068893,
|
|
"grad_norm": 2.716733932495117,
|
|
"learning_rate": 9.402848104062165e-06,
|
|
"loss": 1.228,
|
|
"step": 1895
|
|
},
|
|
{
|
|
"epoch": 0.9971075466736786,
|
|
"grad_norm": 2.541886568069458,
|
|
"learning_rate": 9.402023353361897e-06,
|
|
"loss": 1.2334,
|
|
"step": 1896
|
|
},
|
|
{
|
|
"epoch": 0.9976334472784644,
|
|
"grad_norm": 2.7960169315338135,
|
|
"learning_rate": 9.401198069728427e-06,
|
|
"loss": 1.2283,
|
|
"step": 1897
|
|
},
|
|
{
|
|
"epoch": 0.9981593478832501,
|
|
"grad_norm": 2.682161808013916,
|
|
"learning_rate": 9.40037225326167e-06,
|
|
"loss": 1.1661,
|
|
"step": 1898
|
|
},
|
|
{
|
|
"epoch": 0.9986852484880357,
|
|
"grad_norm": 2.6004488468170166,
|
|
"learning_rate": 9.399545904061602e-06,
|
|
"loss": 1.1228,
|
|
"step": 1899
|
|
},
|
|
{
|
|
"epoch": 0.9992111490928215,
|
|
"grad_norm": 2.839430570602417,
|
|
"learning_rate": 9.398719022228266e-06,
|
|
"loss": 1.2116,
|
|
"step": 1900
|
|
},
|
|
{
|
|
"epoch": 0.9997370496976071,
|
|
"grad_norm": 2.763639211654663,
|
|
"learning_rate": 9.39789160786177e-06,
|
|
"loss": 1.2587,
|
|
"step": 1901
|
|
}
|
|
],
|
|
"logging_steps": 1.0,
|
|
"max_steps": 9505,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 5,
|
|
"save_steps": 500.0,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": false
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 1.1933493078272246e+18,
|
|
"train_batch_size": 2,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|