InternVL2_5_1b_finetuned / trainer_state.json
Aditya02's picture
Initial model upload
4da178b verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9988571428571429,
"eval_steps": 500,
"global_step": 437,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 2.8571428571428573e-06,
"loss": 1.8699,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 5.7142857142857145e-06,
"loss": 2.8546,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 8.571428571428571e-06,
"loss": 1.8072,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 1.1428571428571429e-05,
"loss": 2.4662,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 1.4285714285714287e-05,
"loss": 2.9095,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 1.7142857142857142e-05,
"loss": 3.8613,
"step": 6
},
{
"epoch": 0.02,
"learning_rate": 2e-05,
"loss": 2.1535,
"step": 7
},
{
"epoch": 0.02,
"learning_rate": 2.2857142857142858e-05,
"loss": 3.8349,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 2.5714285714285718e-05,
"loss": 1.5684,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 2.8571428571428574e-05,
"loss": 3.1305,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 3.142857142857143e-05,
"loss": 1.5548,
"step": 11
},
{
"epoch": 0.03,
"learning_rate": 3.4285714285714284e-05,
"loss": 1.4553,
"step": 12
},
{
"epoch": 0.03,
"learning_rate": 3.714285714285715e-05,
"loss": 1.1268,
"step": 13
},
{
"epoch": 0.03,
"learning_rate": 4e-05,
"loss": 1.1313,
"step": 14
},
{
"epoch": 0.03,
"learning_rate": 3.9999448409199496e-05,
"loss": 1.3773,
"step": 15
},
{
"epoch": 0.04,
"learning_rate": 3.999779366722321e-05,
"loss": 1.587,
"step": 16
},
{
"epoch": 0.04,
"learning_rate": 3.9995035865345184e-05,
"loss": 0.868,
"step": 17
},
{
"epoch": 0.04,
"learning_rate": 3.999117515568324e-05,
"loss": 1.1585,
"step": 18
},
{
"epoch": 0.04,
"learning_rate": 3.998621175119058e-05,
"loss": 4.0398,
"step": 19
},
{
"epoch": 0.05,
"learning_rate": 3.998014592564402e-05,
"loss": 0.7112,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 3.997297801362891e-05,
"loss": 0.724,
"step": 21
},
{
"epoch": 0.05,
"learning_rate": 3.996470841052069e-05,
"loss": 1.1229,
"step": 22
},
{
"epoch": 0.05,
"learning_rate": 3.995533757246307e-05,
"loss": 0.9368,
"step": 23
},
{
"epoch": 0.05,
"learning_rate": 3.9944866016342835e-05,
"loss": 0.7026,
"step": 24
},
{
"epoch": 0.06,
"learning_rate": 3.9933294319761405e-05,
"loss": 0.6165,
"step": 25
},
{
"epoch": 0.06,
"learning_rate": 3.9920623121002914e-05,
"loss": 0.5194,
"step": 26
},
{
"epoch": 0.06,
"learning_rate": 3.9906853118999016e-05,
"loss": 0.5856,
"step": 27
},
{
"epoch": 0.06,
"learning_rate": 3.9891985073290375e-05,
"loss": 0.8067,
"step": 28
},
{
"epoch": 0.07,
"learning_rate": 3.98760198039847e-05,
"loss": 0.4607,
"step": 29
},
{
"epoch": 0.07,
"learning_rate": 3.9858958191711566e-05,
"loss": 0.6095,
"step": 30
},
{
"epoch": 0.07,
"learning_rate": 3.98408011775738e-05,
"loss": 0.6495,
"step": 31
},
{
"epoch": 0.07,
"learning_rate": 3.9821549763095606e-05,
"loss": 0.589,
"step": 32
},
{
"epoch": 0.08,
"learning_rate": 3.98012050101673e-05,
"loss": 0.5097,
"step": 33
},
{
"epoch": 0.08,
"learning_rate": 3.9779768040986726e-05,
"loss": 0.4351,
"step": 34
},
{
"epoch": 0.08,
"learning_rate": 3.975724003799739e-05,
"loss": 0.6051,
"step": 35
},
{
"epoch": 0.08,
"learning_rate": 3.973362224382322e-05,
"loss": 0.9956,
"step": 36
},
{
"epoch": 0.08,
"learning_rate": 3.97089159612e-05,
"loss": 0.4976,
"step": 37
},
{
"epoch": 0.09,
"learning_rate": 3.968312255290356e-05,
"loss": 0.5614,
"step": 38
},
{
"epoch": 0.09,
"learning_rate": 3.9656243441674576e-05,
"loss": 0.8428,
"step": 39
},
{
"epoch": 0.09,
"learning_rate": 3.962828011014009e-05,
"loss": 0.7532,
"step": 40
},
{
"epoch": 0.09,
"learning_rate": 3.959923410073174e-05,
"loss": 0.4086,
"step": 41
},
{
"epoch": 0.1,
"learning_rate": 3.956910701560069e-05,
"loss": 0.4337,
"step": 42
},
{
"epoch": 0.1,
"learning_rate": 3.953790051652925e-05,
"loss": 0.369,
"step": 43
},
{
"epoch": 0.1,
"learning_rate": 3.950561632483918e-05,
"loss": 0.3884,
"step": 44
},
{
"epoch": 0.1,
"learning_rate": 3.9472256221296815e-05,
"loss": 0.4261,
"step": 45
},
{
"epoch": 0.11,
"learning_rate": 3.9437822046014764e-05,
"loss": 0.4273,
"step": 46
},
{
"epoch": 0.11,
"learning_rate": 3.940231569835046e-05,
"loss": 0.3396,
"step": 47
},
{
"epoch": 0.11,
"learning_rate": 3.9365739136801366e-05,
"loss": 0.4417,
"step": 48
},
{
"epoch": 0.11,
"learning_rate": 3.932809437889699e-05,
"loss": 0.62,
"step": 49
},
{
"epoch": 0.11,
"learning_rate": 3.928938350108754e-05,
"loss": 0.5719,
"step": 50
},
{
"epoch": 0.12,
"learning_rate": 3.924960863862941e-05,
"loss": 0.6614,
"step": 51
},
{
"epoch": 0.12,
"learning_rate": 3.9208771985467435e-05,
"loss": 0.3308,
"step": 52
},
{
"epoch": 0.12,
"learning_rate": 3.916687579411383e-05,
"loss": 0.3355,
"step": 53
},
{
"epoch": 0.12,
"learning_rate": 3.912392237552397e-05,
"loss": 0.3821,
"step": 54
},
{
"epoch": 0.13,
"learning_rate": 3.907991409896892e-05,
"loss": 0.3476,
"step": 55
},
{
"epoch": 0.13,
"learning_rate": 3.9034853391904715e-05,
"loss": 0.3094,
"step": 56
},
{
"epoch": 0.13,
"learning_rate": 3.8988742739838505e-05,
"loss": 0.3785,
"step": 57
},
{
"epoch": 0.13,
"learning_rate": 3.8941584686191436e-05,
"loss": 0.4117,
"step": 58
},
{
"epoch": 0.13,
"learning_rate": 3.889338183215838e-05,
"loss": 0.3387,
"step": 59
},
{
"epoch": 0.14,
"learning_rate": 3.884413683656441e-05,
"loss": 0.3845,
"step": 60
},
{
"epoch": 0.14,
"learning_rate": 3.879385241571817e-05,
"loss": 0.3232,
"step": 61
},
{
"epoch": 0.14,
"learning_rate": 3.8742531343262076e-05,
"loss": 0.3627,
"step": 62
},
{
"epoch": 0.14,
"learning_rate": 3.869017645001925e-05,
"loss": 0.3797,
"step": 63
},
{
"epoch": 0.15,
"learning_rate": 3.863679062383746e-05,
"loss": 0.2881,
"step": 64
},
{
"epoch": 0.15,
"learning_rate": 3.858237680942975e-05,
"loss": 0.8201,
"step": 65
},
{
"epoch": 0.15,
"learning_rate": 3.852693800821208e-05,
"loss": 0.2627,
"step": 66
},
{
"epoch": 0.15,
"learning_rate": 3.847047727813771e-05,
"loss": 0.3228,
"step": 67
},
{
"epoch": 0.16,
"learning_rate": 3.841299773352858e-05,
"loss": 0.2972,
"step": 68
},
{
"epoch": 0.16,
"learning_rate": 3.8354502544903484e-05,
"loss": 0.5059,
"step": 69
},
{
"epoch": 0.16,
"learning_rate": 3.829499493880322e-05,
"loss": 0.3438,
"step": 70
},
{
"epoch": 0.16,
"learning_rate": 3.823447819761259e-05,
"loss": 0.9574,
"step": 71
},
{
"epoch": 0.16,
"learning_rate": 3.8172955659379376e-05,
"loss": 0.3248,
"step": 72
},
{
"epoch": 0.17,
"learning_rate": 3.811043071763018e-05,
"loss": 0.3328,
"step": 73
},
{
"epoch": 0.17,
"learning_rate": 3.804690682118328e-05,
"loss": 0.5611,
"step": 74
},
{
"epoch": 0.17,
"learning_rate": 3.798238747395835e-05,
"loss": 0.355,
"step": 75
},
{
"epoch": 0.17,
"learning_rate": 3.791687623478324e-05,
"loss": 0.6809,
"step": 76
},
{
"epoch": 0.18,
"learning_rate": 3.785037671719763e-05,
"loss": 0.3312,
"step": 77
},
{
"epoch": 0.18,
"learning_rate": 3.7782892589253735e-05,
"loss": 0.36,
"step": 78
},
{
"epoch": 0.18,
"learning_rate": 3.771442757331397e-05,
"loss": 0.3337,
"step": 79
},
{
"epoch": 0.18,
"learning_rate": 3.764498544584564e-05,
"loss": 0.3287,
"step": 80
},
{
"epoch": 0.19,
"learning_rate": 3.75745700372126e-05,
"loss": 0.3119,
"step": 81
},
{
"epoch": 0.19,
"learning_rate": 3.7503185231464014e-05,
"loss": 0.2839,
"step": 82
},
{
"epoch": 0.19,
"learning_rate": 3.74308349661201e-05,
"loss": 0.9241,
"step": 83
},
{
"epoch": 0.19,
"learning_rate": 3.735752323195494e-05,
"loss": 0.3167,
"step": 84
},
{
"epoch": 0.19,
"learning_rate": 3.7283254072776336e-05,
"loss": 0.3737,
"step": 85
},
{
"epoch": 0.2,
"learning_rate": 3.720803158520279e-05,
"loss": 0.2678,
"step": 86
},
{
"epoch": 0.2,
"learning_rate": 3.713185991843752e-05,
"loss": 0.2083,
"step": 87
},
{
"epoch": 0.2,
"learning_rate": 3.705474327403958e-05,
"loss": 0.7429,
"step": 88
},
{
"epoch": 0.2,
"learning_rate": 3.697668590569215e-05,
"loss": 0.4499,
"step": 89
},
{
"epoch": 0.21,
"learning_rate": 3.689769211896784e-05,
"loss": 0.4594,
"step": 90
},
{
"epoch": 0.21,
"learning_rate": 3.681776627109127e-05,
"loss": 0.379,
"step": 91
},
{
"epoch": 0.21,
"learning_rate": 3.673691277069867e-05,
"loss": 0.4662,
"step": 92
},
{
"epoch": 0.21,
"learning_rate": 3.665513607759474e-05,
"loss": 0.357,
"step": 93
},
{
"epoch": 0.21,
"learning_rate": 3.657244070250665e-05,
"loss": 0.7296,
"step": 94
},
{
"epoch": 0.22,
"learning_rate": 3.648883120683521e-05,
"loss": 0.3254,
"step": 95
},
{
"epoch": 0.22,
"learning_rate": 3.640431220240329e-05,
"loss": 0.3623,
"step": 96
},
{
"epoch": 0.22,
"learning_rate": 3.631888835120141e-05,
"loss": 0.6485,
"step": 97
},
{
"epoch": 0.22,
"learning_rate": 3.6232564365130626e-05,
"loss": 0.3066,
"step": 98
},
{
"epoch": 0.23,
"learning_rate": 3.614534500574259e-05,
"loss": 0.3152,
"step": 99
},
{
"epoch": 0.23,
"learning_rate": 3.605723508397694e-05,
"loss": 0.5848,
"step": 100
},
{
"epoch": 0.23,
"learning_rate": 3.596823945989589e-05,
"loss": 0.3347,
"step": 101
},
{
"epoch": 0.23,
"learning_rate": 3.58783630424162e-05,
"loss": 0.3636,
"step": 102
},
{
"epoch": 0.24,
"learning_rate": 3.5787610789038373e-05,
"loss": 0.2241,
"step": 103
},
{
"epoch": 0.24,
"learning_rate": 3.5695987705573223e-05,
"loss": 0.4166,
"step": 104
},
{
"epoch": 0.24,
"learning_rate": 3.560349884586574e-05,
"loss": 0.6224,
"step": 105
},
{
"epoch": 0.24,
"learning_rate": 3.551014931151635e-05,
"loss": 0.3595,
"step": 106
},
{
"epoch": 0.24,
"learning_rate": 3.5415944251599476e-05,
"loss": 0.2988,
"step": 107
},
{
"epoch": 0.25,
"learning_rate": 3.532088886237956e-05,
"loss": 0.3415,
"step": 108
},
{
"epoch": 0.25,
"learning_rate": 3.522498838702444e-05,
"loss": 0.3732,
"step": 109
},
{
"epoch": 0.25,
"learning_rate": 3.5128248115316114e-05,
"loss": 0.2805,
"step": 110
},
{
"epoch": 0.25,
"learning_rate": 3.5030673383358956e-05,
"loss": 0.5045,
"step": 111
},
{
"epoch": 0.26,
"learning_rate": 3.493226957328542e-05,
"loss": 0.3575,
"step": 112
},
{
"epoch": 0.26,
"learning_rate": 3.483304211295915e-05,
"loss": 0.3387,
"step": 113
},
{
"epoch": 0.26,
"learning_rate": 3.473299647567558e-05,
"loss": 0.5167,
"step": 114
},
{
"epoch": 0.26,
"learning_rate": 3.4632138179860013e-05,
"loss": 0.252,
"step": 115
},
{
"epoch": 0.27,
"learning_rate": 3.453047278876327e-05,
"loss": 0.299,
"step": 116
},
{
"epoch": 0.27,
"learning_rate": 3.442800591015479e-05,
"loss": 0.4623,
"step": 117
},
{
"epoch": 0.27,
"learning_rate": 3.4324743196013335e-05,
"loss": 0.3507,
"step": 118
},
{
"epoch": 0.27,
"learning_rate": 3.422069034221522e-05,
"loss": 0.3827,
"step": 119
},
{
"epoch": 0.27,
"learning_rate": 3.411585308822014e-05,
"loss": 0.3607,
"step": 120
},
{
"epoch": 0.28,
"learning_rate": 3.4010237216754575e-05,
"loss": 0.3312,
"step": 121
},
{
"epoch": 0.28,
"learning_rate": 3.390384855349285e-05,
"loss": 0.3545,
"step": 122
},
{
"epoch": 0.28,
"learning_rate": 3.379669296673574e-05,
"loss": 0.3093,
"step": 123
},
{
"epoch": 0.28,
"learning_rate": 3.3688776367086837e-05,
"loss": 0.5048,
"step": 124
},
{
"epoch": 0.29,
"learning_rate": 3.358010470712651e-05,
"loss": 0.2603,
"step": 125
},
{
"epoch": 0.29,
"learning_rate": 3.347068398108353e-05,
"loss": 0.2749,
"step": 126
},
{
"epoch": 0.29,
"learning_rate": 3.3360520224504505e-05,
"loss": 0.3122,
"step": 127
},
{
"epoch": 0.29,
"learning_rate": 3.324961951392089e-05,
"loss": 0.3832,
"step": 128
},
{
"epoch": 0.29,
"learning_rate": 3.313798796651386e-05,
"loss": 0.2636,
"step": 129
},
{
"epoch": 0.3,
"learning_rate": 3.302563173977688e-05,
"loss": 0.3716,
"step": 130
},
{
"epoch": 0.3,
"learning_rate": 3.291255703117605e-05,
"loss": 0.7091,
"step": 131
},
{
"epoch": 0.3,
"learning_rate": 3.279877007780828e-05,
"loss": 0.491,
"step": 132
},
{
"epoch": 0.3,
"learning_rate": 3.268427715605723e-05,
"loss": 0.4181,
"step": 133
},
{
"epoch": 0.31,
"learning_rate": 3.2569084581247137e-05,
"loss": 0.3262,
"step": 134
},
{
"epoch": 0.31,
"learning_rate": 3.245319870729446e-05,
"loss": 0.3126,
"step": 135
},
{
"epoch": 0.31,
"learning_rate": 3.23366259263574e-05,
"loss": 0.2856,
"step": 136
},
{
"epoch": 0.31,
"learning_rate": 3.221937266848331e-05,
"loss": 0.7066,
"step": 137
},
{
"epoch": 0.32,
"learning_rate": 3.210144540125402e-05,
"loss": 0.2677,
"step": 138
},
{
"epoch": 0.32,
"learning_rate": 3.198285062942912e-05,
"loss": 0.2525,
"step": 139
},
{
"epoch": 0.32,
"learning_rate": 3.1863594894587106e-05,
"loss": 0.3914,
"step": 140
},
{
"epoch": 0.32,
"learning_rate": 3.174368477476461e-05,
"loss": 0.3101,
"step": 141
},
{
"epoch": 0.32,
"learning_rate": 3.1623126884093536e-05,
"loss": 0.3466,
"step": 142
},
{
"epoch": 0.33,
"learning_rate": 3.150192787243621e-05,
"loss": 0.2589,
"step": 143
},
{
"epoch": 0.33,
"learning_rate": 3.1380094425018634e-05,
"loss": 0.2615,
"step": 144
},
{
"epoch": 0.33,
"learning_rate": 3.1257633262061674e-05,
"loss": 0.4989,
"step": 145
},
{
"epoch": 0.33,
"learning_rate": 3.113455113841043e-05,
"loss": 0.3507,
"step": 146
},
{
"epoch": 0.34,
"learning_rate": 3.101085484316162e-05,
"loss": 0.7505,
"step": 147
},
{
"epoch": 0.34,
"learning_rate": 3.0886551199289074e-05,
"loss": 0.304,
"step": 148
},
{
"epoch": 0.34,
"learning_rate": 3.0761647063267455e-05,
"loss": 0.2395,
"step": 149
},
{
"epoch": 0.34,
"learning_rate": 3.063614932469399e-05,
"loss": 0.2855,
"step": 150
},
{
"epoch": 0.35,
"learning_rate": 3.0510064905908492e-05,
"loss": 0.3145,
"step": 151
},
{
"epoch": 0.35,
"learning_rate": 3.038340076161151e-05,
"loss": 0.6382,
"step": 152
},
{
"epoch": 0.35,
"learning_rate": 3.0256163878480715e-05,
"loss": 0.2493,
"step": 153
},
{
"epoch": 0.35,
"learning_rate": 3.012836127478553e-05,
"loss": 0.2916,
"step": 154
},
{
"epoch": 0.35,
"learning_rate": 3.0000000000000004e-05,
"loss": 0.2677,
"step": 155
},
{
"epoch": 0.36,
"learning_rate": 2.9871087134413968e-05,
"loss": 0.346,
"step": 156
},
{
"epoch": 0.36,
"learning_rate": 2.974162978874249e-05,
"loss": 0.2209,
"step": 157
},
{
"epoch": 0.36,
"learning_rate": 2.9611635103733677e-05,
"loss": 0.4053,
"step": 158
},
{
"epoch": 0.36,
"learning_rate": 2.9481110249774755e-05,
"loss": 0.6514,
"step": 159
},
{
"epoch": 0.37,
"learning_rate": 2.935006242649659e-05,
"loss": 0.3108,
"step": 160
},
{
"epoch": 0.37,
"learning_rate": 2.921849886237656e-05,
"loss": 0.7936,
"step": 161
},
{
"epoch": 0.37,
"learning_rate": 2.9086426814339838e-05,
"loss": 0.3377,
"step": 162
},
{
"epoch": 0.37,
"learning_rate": 2.895385356735908e-05,
"loss": 0.3292,
"step": 163
},
{
"epoch": 0.37,
"learning_rate": 2.8820786434052634e-05,
"loss": 0.6782,
"step": 164
},
{
"epoch": 0.38,
"learning_rate": 2.8687232754281162e-05,
"loss": 0.3089,
"step": 165
},
{
"epoch": 0.38,
"learning_rate": 2.8553199894742776e-05,
"loss": 0.3611,
"step": 166
},
{
"epoch": 0.38,
"learning_rate": 2.8418695248566703e-05,
"loss": 0.3621,
"step": 167
},
{
"epoch": 0.38,
"learning_rate": 2.8283726234905492e-05,
"loss": 0.2988,
"step": 168
},
{
"epoch": 0.39,
"learning_rate": 2.8148300298525764e-05,
"loss": 0.3555,
"step": 169
},
{
"epoch": 0.39,
"learning_rate": 2.8012424909397597e-05,
"loss": 0.2199,
"step": 170
},
{
"epoch": 0.39,
"learning_rate": 2.7876107562282447e-05,
"loss": 0.2688,
"step": 171
},
{
"epoch": 0.39,
"learning_rate": 2.7739355776319775e-05,
"loss": 0.2502,
"step": 172
},
{
"epoch": 0.4,
"learning_rate": 2.7602177094612298e-05,
"loss": 0.5406,
"step": 173
},
{
"epoch": 0.4,
"learning_rate": 2.7464579083809896e-05,
"loss": 0.4893,
"step": 174
},
{
"epoch": 0.4,
"learning_rate": 2.732656933369227e-05,
"loss": 0.2604,
"step": 175
},
{
"epoch": 0.4,
"learning_rate": 2.7188155456750257e-05,
"loss": 0.2974,
"step": 176
},
{
"epoch": 0.4,
"learning_rate": 2.7049345087765995e-05,
"loss": 0.3004,
"step": 177
},
{
"epoch": 0.41,
"learning_rate": 2.6910145883391726e-05,
"loss": 0.3131,
"step": 178
},
{
"epoch": 0.41,
"learning_rate": 2.6770565521727516e-05,
"loss": 0.305,
"step": 179
},
{
"epoch": 0.41,
"learning_rate": 2.6630611701897696e-05,
"loss": 0.2949,
"step": 180
},
{
"epoch": 0.41,
"learning_rate": 2.6490292143626228e-05,
"loss": 0.9866,
"step": 181
},
{
"epoch": 0.42,
"learning_rate": 2.6349614586810854e-05,
"loss": 0.2713,
"step": 182
},
{
"epoch": 0.42,
"learning_rate": 2.6208586791096196e-05,
"loss": 0.4035,
"step": 183
},
{
"epoch": 0.42,
"learning_rate": 2.606721653544572e-05,
"loss": 1.2587,
"step": 184
},
{
"epoch": 0.42,
"learning_rate": 2.5925511617712683e-05,
"loss": 0.2273,
"step": 185
},
{
"epoch": 0.43,
"learning_rate": 2.5783479854209983e-05,
"loss": 0.2003,
"step": 186
},
{
"epoch": 0.43,
"learning_rate": 2.5641129079279025e-05,
"loss": 0.8117,
"step": 187
},
{
"epoch": 0.43,
"learning_rate": 2.549846714485761e-05,
"loss": 0.7241,
"step": 188
},
{
"epoch": 0.43,
"learning_rate": 2.5355501920046792e-05,
"loss": 0.854,
"step": 189
},
{
"epoch": 0.43,
"learning_rate": 2.5212241290676857e-05,
"loss": 0.2085,
"step": 190
},
{
"epoch": 0.44,
"learning_rate": 2.5068693158872315e-05,
"loss": 0.3107,
"step": 191
},
{
"epoch": 0.44,
"learning_rate": 2.492486544261608e-05,
"loss": 0.4577,
"step": 192
},
{
"epoch": 0.44,
"learning_rate": 2.4780766075312652e-05,
"loss": 0.4021,
"step": 193
},
{
"epoch": 0.44,
"learning_rate": 2.463640300535057e-05,
"loss": 0.7204,
"step": 194
},
{
"epoch": 0.45,
"learning_rate": 2.4491784195663965e-05,
"loss": 0.2075,
"step": 195
},
{
"epoch": 0.45,
"learning_rate": 2.434691762329334e-05,
"loss": 0.2459,
"step": 196
},
{
"epoch": 0.45,
"learning_rate": 2.420181127894556e-05,
"loss": 0.3556,
"step": 197
},
{
"epoch": 0.45,
"learning_rate": 2.4056473166553087e-05,
"loss": 0.4762,
"step": 198
},
{
"epoch": 0.45,
"learning_rate": 2.391091130283249e-05,
"loss": 0.2244,
"step": 199
},
{
"epoch": 0.46,
"learning_rate": 2.3765133716842273e-05,
"loss": 0.7107,
"step": 200
},
{
"epoch": 0.46,
"learning_rate": 2.3619148449539965e-05,
"loss": 0.4242,
"step": 201
},
{
"epoch": 0.46,
"learning_rate": 2.3472963553338614e-05,
"loss": 0.3044,
"step": 202
},
{
"epoch": 0.46,
"learning_rate": 2.3326587091662605e-05,
"loss": 0.2674,
"step": 203
},
{
"epoch": 0.47,
"learning_rate": 2.3180027138502913e-05,
"loss": 0.2977,
"step": 204
},
{
"epoch": 0.47,
"learning_rate": 2.303329177797172e-05,
"loss": 0.3201,
"step": 205
},
{
"epoch": 0.47,
"learning_rate": 2.2886389103856534e-05,
"loss": 0.2589,
"step": 206
},
{
"epoch": 0.47,
"learning_rate": 2.2739327219173707e-05,
"loss": 0.2949,
"step": 207
},
{
"epoch": 0.48,
"learning_rate": 2.259211423572152e-05,
"loss": 0.2762,
"step": 208
},
{
"epoch": 0.48,
"learning_rate": 2.2444758273632693e-05,
"loss": 0.3394,
"step": 209
},
{
"epoch": 0.48,
"learning_rate": 2.2297267460926548e-05,
"loss": 0.2609,
"step": 210
},
{
"epoch": 0.48,
"learning_rate": 2.2149649933060625e-05,
"loss": 0.295,
"step": 211
},
{
"epoch": 0.48,
"learning_rate": 2.200191383248197e-05,
"loss": 0.3794,
"step": 212
},
{
"epoch": 0.49,
"learning_rate": 2.1854067308177967e-05,
"loss": 0.4467,
"step": 213
},
{
"epoch": 0.49,
"learning_rate": 2.1706118515226894e-05,
"loss": 0.2828,
"step": 214
},
{
"epoch": 0.49,
"learning_rate": 2.1558075614348065e-05,
"loss": 0.3457,
"step": 215
},
{
"epoch": 0.49,
"learning_rate": 2.1409946771451705e-05,
"loss": 0.2655,
"step": 216
},
{
"epoch": 0.5,
"learning_rate": 2.1261740157188498e-05,
"loss": 0.3977,
"step": 217
},
{
"epoch": 0.5,
"learning_rate": 2.111346394649897e-05,
"loss": 0.274,
"step": 218
},
{
"epoch": 0.5,
"learning_rate": 2.0965126318162476e-05,
"loss": 0.4635,
"step": 219
},
{
"epoch": 0.5,
"learning_rate": 2.0816735454346134e-05,
"loss": 0.3107,
"step": 220
},
{
"epoch": 0.51,
"learning_rate": 2.0668299540153494e-05,
"loss": 0.4861,
"step": 221
},
{
"epoch": 0.51,
"learning_rate": 2.051982676317302e-05,
"loss": 0.2846,
"step": 222
},
{
"epoch": 0.51,
"learning_rate": 2.0371325313026502e-05,
"loss": 0.3598,
"step": 223
},
{
"epoch": 0.51,
"learning_rate": 2.022280338091731e-05,
"loss": 0.2647,
"step": 224
},
{
"epoch": 0.51,
"learning_rate": 2.0074269159178606e-05,
"loss": 0.4459,
"step": 225
},
{
"epoch": 0.52,
"learning_rate": 1.9925730840821404e-05,
"loss": 0.5016,
"step": 226
},
{
"epoch": 0.52,
"learning_rate": 1.9777196619082693e-05,
"loss": 0.3645,
"step": 227
},
{
"epoch": 0.52,
"learning_rate": 1.9628674686973508e-05,
"loss": 0.5755,
"step": 228
},
{
"epoch": 0.52,
"learning_rate": 1.948017323682699e-05,
"loss": 0.3132,
"step": 229
},
{
"epoch": 0.53,
"learning_rate": 1.9331700459846516e-05,
"loss": 0.5478,
"step": 230
},
{
"epoch": 0.53,
"learning_rate": 1.918326454565387e-05,
"loss": 0.274,
"step": 231
},
{
"epoch": 0.53,
"learning_rate": 1.9034873681837534e-05,
"loss": 0.3535,
"step": 232
},
{
"epoch": 0.53,
"learning_rate": 1.8886536053501042e-05,
"loss": 0.2747,
"step": 233
},
{
"epoch": 0.53,
"learning_rate": 1.873825984281151e-05,
"loss": 0.2108,
"step": 234
},
{
"epoch": 0.54,
"learning_rate": 1.8590053228548305e-05,
"loss": 0.3552,
"step": 235
},
{
"epoch": 0.54,
"learning_rate": 1.844192438565194e-05,
"loss": 0.3144,
"step": 236
},
{
"epoch": 0.54,
"learning_rate": 1.829388148477311e-05,
"loss": 0.2116,
"step": 237
},
{
"epoch": 0.54,
"learning_rate": 1.814593269182204e-05,
"loss": 0.3492,
"step": 238
},
{
"epoch": 0.55,
"learning_rate": 1.7998086167518033e-05,
"loss": 0.2632,
"step": 239
},
{
"epoch": 0.55,
"learning_rate": 1.785035006693938e-05,
"loss": 0.4334,
"step": 240
},
{
"epoch": 0.55,
"learning_rate": 1.770273253907346e-05,
"loss": 0.2417,
"step": 241
},
{
"epoch": 0.55,
"learning_rate": 1.7555241726367317e-05,
"loss": 0.2955,
"step": 242
},
{
"epoch": 0.56,
"learning_rate": 1.7407885764278488e-05,
"loss": 0.538,
"step": 243
},
{
"epoch": 0.56,
"learning_rate": 1.7260672780826296e-05,
"loss": 0.2689,
"step": 244
},
{
"epoch": 0.56,
"learning_rate": 1.7113610896143473e-05,
"loss": 0.2849,
"step": 245
},
{
"epoch": 0.56,
"learning_rate": 1.6966708222028284e-05,
"loss": 0.5197,
"step": 246
},
{
"epoch": 0.56,
"learning_rate": 1.681997286149709e-05,
"loss": 0.4282,
"step": 247
},
{
"epoch": 0.57,
"learning_rate": 1.6673412908337402e-05,
"loss": 0.4306,
"step": 248
},
{
"epoch": 0.57,
"learning_rate": 1.6527036446661396e-05,
"loss": 0.2655,
"step": 249
},
{
"epoch": 0.57,
"learning_rate": 1.638085155046004e-05,
"loss": 0.2738,
"step": 250
},
{
"epoch": 0.57,
"learning_rate": 1.623486628315773e-05,
"loss": 0.4238,
"step": 251
},
{
"epoch": 0.58,
"learning_rate": 1.608908869716751e-05,
"loss": 0.2554,
"step": 252
},
{
"epoch": 0.58,
"learning_rate": 1.5943526833446917e-05,
"loss": 0.295,
"step": 253
},
{
"epoch": 0.58,
"learning_rate": 1.579818872105444e-05,
"loss": 0.3214,
"step": 254
},
{
"epoch": 0.58,
"learning_rate": 1.565308237670666e-05,
"loss": 0.2114,
"step": 255
},
{
"epoch": 0.59,
"learning_rate": 1.550821580433604e-05,
"loss": 0.3037,
"step": 256
},
{
"epoch": 0.59,
"learning_rate": 1.5363596994649433e-05,
"loss": 0.3683,
"step": 257
},
{
"epoch": 0.59,
"learning_rate": 1.5219233924687351e-05,
"loss": 0.4236,
"step": 258
},
{
"epoch": 0.59,
"learning_rate": 1.5075134557383931e-05,
"loss": 0.2787,
"step": 259
},
{
"epoch": 0.59,
"learning_rate": 1.4931306841127691e-05,
"loss": 0.588,
"step": 260
},
{
"epoch": 0.6,
"learning_rate": 1.4787758709323155e-05,
"loss": 0.2892,
"step": 261
},
{
"epoch": 0.6,
"learning_rate": 1.4644498079953215e-05,
"loss": 0.5691,
"step": 262
},
{
"epoch": 0.6,
"learning_rate": 1.45015328551424e-05,
"loss": 0.2227,
"step": 263
},
{
"epoch": 0.6,
"learning_rate": 1.4358870920720982e-05,
"loss": 0.5576,
"step": 264
},
{
"epoch": 0.61,
"learning_rate": 1.4216520145790027e-05,
"loss": 0.6571,
"step": 265
},
{
"epoch": 0.61,
"learning_rate": 1.4074488382287324e-05,
"loss": 0.4547,
"step": 266
},
{
"epoch": 0.61,
"learning_rate": 1.3932783464554286e-05,
"loss": 0.3705,
"step": 267
},
{
"epoch": 0.61,
"learning_rate": 1.379141320890381e-05,
"loss": 0.4904,
"step": 268
},
{
"epoch": 0.61,
"learning_rate": 1.3650385413189151e-05,
"loss": 0.4299,
"step": 269
},
{
"epoch": 0.62,
"learning_rate": 1.3509707856373779e-05,
"loss": 0.4775,
"step": 270
},
{
"epoch": 0.62,
"learning_rate": 1.3369388298102312e-05,
"loss": 0.2769,
"step": 271
},
{
"epoch": 0.62,
"learning_rate": 1.3229434478272492e-05,
"loss": 0.2478,
"step": 272
},
{
"epoch": 0.62,
"learning_rate": 1.3089854116608279e-05,
"loss": 0.3056,
"step": 273
},
{
"epoch": 0.63,
"learning_rate": 1.295065491223401e-05,
"loss": 0.4498,
"step": 274
},
{
"epoch": 0.63,
"learning_rate": 1.2811844543249748e-05,
"loss": 0.5963,
"step": 275
},
{
"epoch": 0.63,
"learning_rate": 1.2673430666307738e-05,
"loss": 0.4305,
"step": 276
},
{
"epoch": 0.63,
"learning_rate": 1.2535420916190106e-05,
"loss": 0.3375,
"step": 277
},
{
"epoch": 0.64,
"learning_rate": 1.2397822905387707e-05,
"loss": 0.3299,
"step": 278
},
{
"epoch": 0.64,
"learning_rate": 1.2260644223680228e-05,
"loss": 0.537,
"step": 279
},
{
"epoch": 0.64,
"learning_rate": 1.212389243771756e-05,
"loss": 0.2732,
"step": 280
},
{
"epoch": 0.64,
"learning_rate": 1.1987575090602408e-05,
"loss": 0.325,
"step": 281
},
{
"epoch": 0.64,
"learning_rate": 1.185169970147424e-05,
"loss": 1.0755,
"step": 282
},
{
"epoch": 0.65,
"learning_rate": 1.1716273765094517e-05,
"loss": 0.6557,
"step": 283
},
{
"epoch": 0.65,
"learning_rate": 1.1581304751433305e-05,
"loss": 0.4481,
"step": 284
},
{
"epoch": 0.65,
"learning_rate": 1.1446800105257232e-05,
"loss": 0.3972,
"step": 285
},
{
"epoch": 0.65,
"learning_rate": 1.1312767245718836e-05,
"loss": 0.3317,
"step": 286
},
{
"epoch": 0.66,
"learning_rate": 1.1179213565947366e-05,
"loss": 0.2758,
"step": 287
},
{
"epoch": 0.66,
"learning_rate": 1.1046146432640923e-05,
"loss": 0.2696,
"step": 288
},
{
"epoch": 0.66,
"learning_rate": 1.0913573185660167e-05,
"loss": 0.3465,
"step": 289
},
{
"epoch": 0.66,
"learning_rate": 1.078150113762344e-05,
"loss": 0.3807,
"step": 290
},
{
"epoch": 0.67,
"learning_rate": 1.0649937573503419e-05,
"loss": 0.3017,
"step": 291
},
{
"epoch": 0.67,
"learning_rate": 1.051888975022525e-05,
"loss": 0.3269,
"step": 292
},
{
"epoch": 0.67,
"learning_rate": 1.0388364896266326e-05,
"loss": 0.1849,
"step": 293
},
{
"epoch": 0.67,
"learning_rate": 1.0258370211257511e-05,
"loss": 0.4315,
"step": 294
},
{
"epoch": 0.67,
"learning_rate": 1.0128912865586038e-05,
"loss": 0.2808,
"step": 295
},
{
"epoch": 0.68,
"learning_rate": 1.0000000000000006e-05,
"loss": 0.292,
"step": 296
},
{
"epoch": 0.68,
"learning_rate": 9.871638725214481e-06,
"loss": 0.2757,
"step": 297
},
{
"epoch": 0.68,
"learning_rate": 9.743836121519297e-06,
"loss": 0.2608,
"step": 298
},
{
"epoch": 0.68,
"learning_rate": 9.616599238388501e-06,
"loss": 0.2605,
"step": 299
},
{
"epoch": 0.69,
"learning_rate": 9.48993509409151e-06,
"loss": 0.2572,
"step": 300
},
{
"epoch": 0.69,
"learning_rate": 9.363850675306013e-06,
"loss": 0.2759,
"step": 301
},
{
"epoch": 0.69,
"learning_rate": 9.238352936732549e-06,
"loss": 0.2789,
"step": 302
},
{
"epoch": 0.69,
"learning_rate": 9.113448800710929e-06,
"loss": 0.2701,
"step": 303
},
{
"epoch": 0.69,
"learning_rate": 8.989145156838387e-06,
"loss": 0.2623,
"step": 304
},
{
"epoch": 0.7,
"learning_rate": 8.865448861589572e-06,
"loss": 0.2679,
"step": 305
},
{
"epoch": 0.7,
"learning_rate": 8.74236673793833e-06,
"loss": 0.4513,
"step": 306
},
{
"epoch": 0.7,
"learning_rate": 8.619905574981378e-06,
"loss": 0.2597,
"step": 307
},
{
"epoch": 0.7,
"learning_rate": 8.498072127563793e-06,
"loss": 0.2465,
"step": 308
},
{
"epoch": 0.71,
"learning_rate": 8.37687311590647e-06,
"loss": 0.2899,
"step": 309
},
{
"epoch": 0.71,
"learning_rate": 8.256315225235392e-06,
"loss": 0.321,
"step": 310
},
{
"epoch": 0.71,
"learning_rate": 8.136405105412897e-06,
"loss": 0.5139,
"step": 311
},
{
"epoch": 0.71,
"learning_rate": 8.017149370570884e-06,
"loss": 0.4021,
"step": 312
},
{
"epoch": 0.72,
"learning_rate": 7.89855459874598e-06,
"loss": 0.4034,
"step": 313
},
{
"epoch": 0.72,
"learning_rate": 7.780627331516697e-06,
"loss": 0.3049,
"step": 314
},
{
"epoch": 0.72,
"learning_rate": 7.6633740736426e-06,
"loss": 0.2607,
"step": 315
},
{
"epoch": 0.72,
"learning_rate": 7.546801292705539e-06,
"loss": 0.4599,
"step": 316
},
{
"epoch": 0.72,
"learning_rate": 7.430915418752867e-06,
"loss": 0.3117,
"step": 317
},
{
"epoch": 0.73,
"learning_rate": 7.3157228439427765e-06,
"loss": 0.4483,
"step": 318
},
{
"epoch": 0.73,
"learning_rate": 7.201229922191726e-06,
"loss": 0.2225,
"step": 319
},
{
"epoch": 0.73,
"learning_rate": 7.087442968823952e-06,
"loss": 0.467,
"step": 320
},
{
"epoch": 0.73,
"learning_rate": 6.974368260223123e-06,
"loss": 0.4134,
"step": 321
},
{
"epoch": 0.74,
"learning_rate": 6.862012033486145e-06,
"loss": 0.3155,
"step": 322
},
{
"epoch": 0.74,
"learning_rate": 6.7503804860791115e-06,
"loss": 0.4041,
"step": 323
},
{
"epoch": 0.74,
"learning_rate": 6.6394797754955055e-06,
"loss": 0.2644,
"step": 324
},
{
"epoch": 0.74,
"learning_rate": 6.529316018916478e-06,
"loss": 0.269,
"step": 325
},
{
"epoch": 0.75,
"learning_rate": 6.419895292873504e-06,
"loss": 0.2354,
"step": 326
},
{
"epoch": 0.75,
"learning_rate": 6.311223632913173e-06,
"loss": 0.2503,
"step": 327
},
{
"epoch": 0.75,
"learning_rate": 6.203307033264272e-06,
"loss": 0.2419,
"step": 328
},
{
"epoch": 0.75,
"learning_rate": 6.096151446507155e-06,
"loss": 0.3827,
"step": 329
},
{
"epoch": 0.75,
"learning_rate": 5.989762783245423e-06,
"loss": 0.2289,
"step": 330
},
{
"epoch": 0.76,
"learning_rate": 5.884146911779865e-06,
"loss": 0.3037,
"step": 331
},
{
"epoch": 0.76,
"learning_rate": 5.779309657784786e-06,
"loss": 0.3167,
"step": 332
},
{
"epoch": 0.76,
"learning_rate": 5.6752568039866754e-06,
"loss": 0.5754,
"step": 333
},
{
"epoch": 0.76,
"learning_rate": 5.5719940898452205e-06,
"loss": 0.2943,
"step": 334
},
{
"epoch": 0.77,
"learning_rate": 5.46952721123674e-06,
"loss": 0.376,
"step": 335
},
{
"epoch": 0.77,
"learning_rate": 5.367861820139995e-06,
"loss": 0.4093,
"step": 336
},
{
"epoch": 0.77,
"learning_rate": 5.267003524324423e-06,
"loss": 0.4033,
"step": 337
},
{
"epoch": 0.77,
"learning_rate": 5.166957887040849e-06,
"loss": 0.3671,
"step": 338
},
{
"epoch": 0.77,
"learning_rate": 5.067730426714583e-06,
"loss": 0.2423,
"step": 339
},
{
"epoch": 0.78,
"learning_rate": 4.969326616641052e-06,
"loss": 0.2651,
"step": 340
},
{
"epoch": 0.78,
"learning_rate": 4.871751884683895e-06,
"loss": 0.5764,
"step": 341
},
{
"epoch": 0.78,
"learning_rate": 4.775011612975562e-06,
"loss": 0.3681,
"step": 342
},
{
"epoch": 0.78,
"learning_rate": 4.679111137620442e-06,
"loss": 0.3841,
"step": 343
},
{
"epoch": 0.79,
"learning_rate": 4.5840557484005355e-06,
"loss": 0.2937,
"step": 344
},
{
"epoch": 0.79,
"learning_rate": 4.4898506884836565e-06,
"loss": 0.2957,
"step": 345
},
{
"epoch": 0.79,
"learning_rate": 4.3965011541342606e-06,
"loss": 0.2435,
"step": 346
},
{
"epoch": 0.79,
"learning_rate": 4.304012294426781e-06,
"loss": 0.3995,
"step": 347
},
{
"epoch": 0.8,
"learning_rate": 4.212389210961629e-06,
"loss": 0.2983,
"step": 348
},
{
"epoch": 0.8,
"learning_rate": 4.121636957583805e-06,
"loss": 0.4387,
"step": 349
},
{
"epoch": 0.8,
"learning_rate": 4.031760540104115e-06,
"loss": 0.3402,
"step": 350
},
{
"epoch": 0.8,
"learning_rate": 3.942764916023067e-06,
"loss": 0.3425,
"step": 351
},
{
"epoch": 0.8,
"learning_rate": 3.854654994257412e-06,
"loss": 0.34,
"step": 352
},
{
"epoch": 0.81,
"learning_rate": 3.7674356348693764e-06,
"loss": 0.3359,
"step": 353
},
{
"epoch": 0.81,
"learning_rate": 3.681111648798592e-06,
"loss": 0.393,
"step": 354
},
{
"epoch": 0.81,
"learning_rate": 3.5956877975967163e-06,
"loss": 0.4085,
"step": 355
},
{
"epoch": 0.81,
"learning_rate": 3.5111687931647984e-06,
"loss": 0.2618,
"step": 356
},
{
"epoch": 0.82,
"learning_rate": 3.427559297493359e-06,
"loss": 0.2674,
"step": 357
},
{
"epoch": 0.82,
"learning_rate": 3.3448639224052703e-06,
"loss": 0.251,
"step": 358
},
{
"epoch": 0.82,
"learning_rate": 3.2630872293013403e-06,
"loss": 0.2529,
"step": 359
},
{
"epoch": 0.82,
"learning_rate": 3.182233728908741e-06,
"loss": 0.1951,
"step": 360
},
{
"epoch": 0.83,
"learning_rate": 3.102307881032165e-06,
"loss": 0.3467,
"step": 361
},
{
"epoch": 0.83,
"learning_rate": 3.023314094307859e-06,
"loss": 0.3827,
"step": 362
},
{
"epoch": 0.83,
"learning_rate": 2.9452567259604215e-06,
"loss": 0.2283,
"step": 363
},
{
"epoch": 0.83,
"learning_rate": 2.868140081562487e-06,
"loss": 0.3542,
"step": 364
},
{
"epoch": 0.83,
"learning_rate": 2.791968414797217e-06,
"loss": 0.2871,
"step": 365
},
{
"epoch": 0.84,
"learning_rate": 2.7167459272236718e-06,
"loss": 0.323,
"step": 366
},
{
"epoch": 0.84,
"learning_rate": 2.6424767680450657e-06,
"loss": 0.9199,
"step": 367
},
{
"epoch": 0.84,
"learning_rate": 2.5691650338799012e-06,
"loss": 0.3337,
"step": 368
},
{
"epoch": 0.84,
"learning_rate": 2.496814768535989e-06,
"loss": 0.4007,
"step": 369
},
{
"epoch": 0.85,
"learning_rate": 2.4254299627874045e-06,
"loss": 0.3352,
"step": 370
},
{
"epoch": 0.85,
"learning_rate": 2.3550145541543666e-06,
"loss": 0.3592,
"step": 371
},
{
"epoch": 0.85,
"learning_rate": 2.2855724266860314e-06,
"loss": 0.2621,
"step": 372
},
{
"epoch": 0.85,
"learning_rate": 2.217107410746271e-06,
"loss": 0.2718,
"step": 373
},
{
"epoch": 0.85,
"learning_rate": 2.149623282802378e-06,
"loss": 0.1917,
"step": 374
},
{
"epoch": 0.86,
"learning_rate": 2.0831237652167656e-06,
"loss": 0.2845,
"step": 375
},
{
"epoch": 0.86,
"learning_rate": 2.0176125260416544e-06,
"loss": 0.2991,
"step": 376
},
{
"epoch": 0.86,
"learning_rate": 1.9530931788167274e-06,
"loss": 0.3188,
"step": 377
},
{
"epoch": 0.86,
"learning_rate": 1.889569282369823e-06,
"loss": 0.2927,
"step": 378
},
{
"epoch": 0.87,
"learning_rate": 1.8270443406206273e-06,
"loss": 0.2112,
"step": 379
},
{
"epoch": 0.87,
"learning_rate": 1.7655218023874131e-06,
"loss": 0.2621,
"step": 380
},
{
"epoch": 0.87,
"learning_rate": 1.7050050611967872e-06,
"loss": 0.2712,
"step": 381
},
{
"epoch": 0.87,
"learning_rate": 1.6454974550965185e-06,
"loss": 0.527,
"step": 382
},
{
"epoch": 0.88,
"learning_rate": 1.5870022664714225e-06,
"loss": 0.3912,
"step": 383
},
{
"epoch": 0.88,
"learning_rate": 1.529522721862291e-06,
"loss": 0.4273,
"step": 384
},
{
"epoch": 0.88,
"learning_rate": 1.473061991787923e-06,
"loss": 0.2741,
"step": 385
},
{
"epoch": 0.88,
"learning_rate": 1.4176231905702476e-06,
"loss": 0.4739,
"step": 386
},
{
"epoch": 0.88,
"learning_rate": 1.363209376162542e-06,
"loss": 0.7205,
"step": 387
},
{
"epoch": 0.89,
"learning_rate": 1.309823549980751e-06,
"loss": 0.3439,
"step": 388
},
{
"epoch": 0.89,
"learning_rate": 1.2574686567379324e-06,
"loss": 0.3784,
"step": 389
},
{
"epoch": 0.89,
"learning_rate": 1.2061475842818337e-06,
"loss": 0.3971,
"step": 390
},
{
"epoch": 0.89,
"learning_rate": 1.1558631634356e-06,
"loss": 0.2784,
"step": 391
},
{
"epoch": 0.9,
"learning_rate": 1.1066181678416266e-06,
"loss": 0.3908,
"step": 392
},
{
"epoch": 0.9,
"learning_rate": 1.058415313808565e-06,
"loss": 0.2788,
"step": 393
},
{
"epoch": 0.9,
"learning_rate": 1.0112572601615022e-06,
"loss": 0.2195,
"step": 394
},
{
"epoch": 0.9,
"learning_rate": 9.65146608095293e-07,
"loss": 0.2877,
"step": 395
},
{
"epoch": 0.91,
"learning_rate": 9.200859010310847e-07,
"loss": 0.1981,
"step": 396
},
{
"epoch": 0.91,
"learning_rate": 8.760776244760283e-07,
"loss": 0.2398,
"step": 397
},
{
"epoch": 0.91,
"learning_rate": 8.33124205886171e-07,
"loss": 0.217,
"step": 398
},
{
"epoch": 0.91,
"learning_rate": 7.912280145325702e-07,
"loss": 0.3294,
"step": 399
},
{
"epoch": 0.91,
"learning_rate": 7.503913613705971e-07,
"loss": 0.3187,
"step": 400
},
{
"epoch": 0.92,
"learning_rate": 7.106164989124708e-07,
"loss": 0.2297,
"step": 401
},
{
"epoch": 0.92,
"learning_rate": 6.719056211030128e-07,
"loss": 0.4698,
"step": 402
},
{
"epoch": 0.92,
"learning_rate": 6.342608631986346e-07,
"loss": 0.2209,
"step": 403
},
{
"epoch": 0.92,
"learning_rate": 5.976843016495482e-07,
"loss": 0.3496,
"step": 404
},
{
"epoch": 0.93,
"learning_rate": 5.621779539852435e-07,
"loss": 0.3261,
"step": 405
},
{
"epoch": 0.93,
"learning_rate": 5.277437787031892e-07,
"loss": 0.2379,
"step": 406
},
{
"epoch": 0.93,
"learning_rate": 4.943836751608211e-07,
"loss": 0.347,
"step": 407
},
{
"epoch": 0.93,
"learning_rate": 4.6209948347075483e-07,
"loss": 0.3664,
"step": 408
},
{
"epoch": 0.93,
"learning_rate": 4.308929843993115e-07,
"loss": 0.2325,
"step": 409
},
{
"epoch": 0.94,
"learning_rate": 4.0076589926826503e-07,
"loss": 0.3251,
"step": 410
},
{
"epoch": 0.94,
"learning_rate": 3.7171988985991835e-07,
"loss": 0.3222,
"step": 411
},
{
"epoch": 0.94,
"learning_rate": 3.4375655832542763e-07,
"loss": 0.3088,
"step": 412
},
{
"epoch": 0.94,
"learning_rate": 3.1687744709644197e-07,
"loss": 0.3003,
"step": 413
},
{
"epoch": 0.95,
"learning_rate": 2.9108403880000247e-07,
"loss": 0.359,
"step": 414
},
{
"epoch": 0.95,
"learning_rate": 2.663777561767855e-07,
"loss": 0.4534,
"step": 415
},
{
"epoch": 0.95,
"learning_rate": 2.4275996200261e-07,
"loss": 0.3747,
"step": 416
},
{
"epoch": 0.95,
"learning_rate": 2.2023195901327731e-07,
"loss": 0.2838,
"step": 417
},
{
"epoch": 0.96,
"learning_rate": 1.9879498983270685e-07,
"loss": 0.2783,
"step": 418
},
{
"epoch": 0.96,
"learning_rate": 1.7845023690439944e-07,
"loss": 0.4112,
"step": 419
},
{
"epoch": 0.96,
"learning_rate": 1.591988224262053e-07,
"loss": 0.3825,
"step": 420
},
{
"epoch": 0.96,
"learning_rate": 1.4104180828844237e-07,
"loss": 0.496,
"step": 421
},
{
"epoch": 0.96,
"learning_rate": 1.239801960153053e-07,
"loss": 0.4617,
"step": 422
},
{
"epoch": 0.97,
"learning_rate": 1.0801492670962976e-07,
"loss": 0.2298,
"step": 423
},
{
"epoch": 0.97,
"learning_rate": 9.314688100098502e-08,
"loss": 0.2499,
"step": 424
},
{
"epoch": 0.97,
"learning_rate": 7.9376878997095e-08,
"loss": 0.5606,
"step": 425
},
{
"epoch": 0.97,
"learning_rate": 6.670568023859902e-08,
"loss": 0.2629,
"step": 426
},
{
"epoch": 0.98,
"learning_rate": 5.5133983657167376e-08,
"loss": 0.4196,
"step": 427
},
{
"epoch": 0.98,
"learning_rate": 4.4662427536936727e-08,
"loss": 0.2699,
"step": 428
},
{
"epoch": 0.98,
"learning_rate": 3.529158947930933e-08,
"loss": 0.3988,
"step": 429
},
{
"epoch": 0.98,
"learning_rate": 2.70219863710941e-08,
"loss": 0.2276,
"step": 430
},
{
"epoch": 0.99,
"learning_rate": 1.9854074355987186e-08,
"loss": 0.243,
"step": 431
},
{
"epoch": 0.99,
"learning_rate": 1.37882488094232e-08,
"loss": 0.3782,
"step": 432
},
{
"epoch": 0.99,
"learning_rate": 8.82484431675712e-09,
"loss": 0.2823,
"step": 433
},
{
"epoch": 0.99,
"learning_rate": 4.9641346548190415e-09,
"loss": 0.3616,
"step": 434
},
{
"epoch": 0.99,
"learning_rate": 2.206332776797382e-09,
"loss": 0.359,
"step": 435
},
{
"epoch": 1.0,
"learning_rate": 5.51590800510482e-10,
"loss": 0.5555,
"step": 436
},
{
"epoch": 1.0,
"learning_rate": 0.0,
"loss": 0.5426,
"step": 437
},
{
"epoch": 1.0,
"step": 437,
"total_flos": 4.122039303326254e+20,
"train_loss": 0.46482172025286633,
"train_runtime": 2900.7464,
"train_samples_per_second": 2.413,
"train_steps_per_second": 0.151
}
],
"logging_steps": 1.0,
"max_steps": 437,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 200,
"total_flos": 4.122039303326254e+20,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}