llava_model / trainer_state.json
khaninza's picture
Upload trainer_state.json
c4af4b4 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 500,
"global_step": 363,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1.0526315789473685e-06,
"loss": 1.8901,
"step": 1
},
{
"epoch": 0.02,
"learning_rate": 2.105263157894737e-06,
"loss": 1.9537,
"step": 2
},
{
"epoch": 0.02,
"learning_rate": 3.157894736842105e-06,
"loss": 1.9589,
"step": 3
},
{
"epoch": 0.03,
"learning_rate": 4.210526315789474e-06,
"loss": 1.9071,
"step": 4
},
{
"epoch": 0.04,
"learning_rate": 5.263157894736842e-06,
"loss": 1.8846,
"step": 5
},
{
"epoch": 0.05,
"learning_rate": 6.31578947368421e-06,
"loss": 1.8558,
"step": 6
},
{
"epoch": 0.06,
"learning_rate": 7.368421052631579e-06,
"loss": 1.8128,
"step": 7
},
{
"epoch": 0.07,
"learning_rate": 8.421052631578948e-06,
"loss": 1.7392,
"step": 8
},
{
"epoch": 0.07,
"learning_rate": 9.473684210526315e-06,
"loss": 1.7129,
"step": 9
},
{
"epoch": 0.08,
"learning_rate": 1.0526315789473684e-05,
"loss": 1.6144,
"step": 10
},
{
"epoch": 0.09,
"learning_rate": 1.1578947368421053e-05,
"loss": 1.5617,
"step": 11
},
{
"epoch": 0.1,
"learning_rate": 1.263157894736842e-05,
"loss": 1.6225,
"step": 12
},
{
"epoch": 0.11,
"learning_rate": 1.3684210526315791e-05,
"loss": 1.4294,
"step": 13
},
{
"epoch": 0.12,
"learning_rate": 1.4736842105263159e-05,
"loss": 1.4533,
"step": 14
},
{
"epoch": 0.12,
"learning_rate": 1.578947368421053e-05,
"loss": 1.4398,
"step": 15
},
{
"epoch": 0.13,
"learning_rate": 1.6842105263157896e-05,
"loss": 1.3575,
"step": 16
},
{
"epoch": 0.14,
"learning_rate": 1.7894736842105264e-05,
"loss": 1.4498,
"step": 17
},
{
"epoch": 0.15,
"learning_rate": 1.894736842105263e-05,
"loss": 1.3377,
"step": 18
},
{
"epoch": 0.16,
"learning_rate": 2e-05,
"loss": 1.3156,
"step": 19
},
{
"epoch": 0.17,
"learning_rate": 1.999958298675784e-05,
"loss": 1.2075,
"step": 20
},
{
"epoch": 0.17,
"learning_rate": 1.999833198181137e-05,
"loss": 1.1799,
"step": 21
},
{
"epoch": 0.18,
"learning_rate": 1.9996247089497703e-05,
"loss": 1.1307,
"step": 22
},
{
"epoch": 0.19,
"learning_rate": 1.9993328483702393e-05,
"loss": 1.3141,
"step": 23
},
{
"epoch": 0.2,
"learning_rate": 1.9989576407844894e-05,
"loss": 1.2313,
"step": 24
},
{
"epoch": 0.21,
"learning_rate": 1.998499117485826e-05,
"loss": 1.1836,
"step": 25
},
{
"epoch": 0.21,
"learning_rate": 1.997957316716307e-05,
"loss": 1.1769,
"step": 26
},
{
"epoch": 0.22,
"learning_rate": 1.9973322836635517e-05,
"loss": 1.1203,
"step": 27
},
{
"epoch": 0.23,
"learning_rate": 1.9966240704569722e-05,
"loss": 1.1273,
"step": 28
},
{
"epoch": 0.24,
"learning_rate": 1.9958327361634248e-05,
"loss": 1.0204,
"step": 29
},
{
"epoch": 0.25,
"learning_rate": 1.9949583467822863e-05,
"loss": 1.1182,
"step": 30
},
{
"epoch": 0.26,
"learning_rate": 1.9940009752399462e-05,
"loss": 1.105,
"step": 31
},
{
"epoch": 0.26,
"learning_rate": 1.9929607013837268e-05,
"loss": 1.1114,
"step": 32
},
{
"epoch": 0.27,
"learning_rate": 1.991837611975223e-05,
"loss": 1.0798,
"step": 33
},
{
"epoch": 0.28,
"learning_rate": 1.990631800683066e-05,
"loss": 1.075,
"step": 34
},
{
"epoch": 0.29,
"learning_rate": 1.9893433680751105e-05,
"loss": 1.0588,
"step": 35
},
{
"epoch": 0.3,
"learning_rate": 1.9879724216100488e-05,
"loss": 1.0657,
"step": 36
},
{
"epoch": 0.31,
"learning_rate": 1.9865190756284467e-05,
"loss": 1.0628,
"step": 37
},
{
"epoch": 0.31,
"learning_rate": 1.9849834513432084e-05,
"loss": 1.0119,
"step": 38
},
{
"epoch": 0.32,
"learning_rate": 1.983365676829466e-05,
"loss": 1.0563,
"step": 39
},
{
"epoch": 0.33,
"learning_rate": 1.981665887013899e-05,
"loss": 0.9891,
"step": 40
},
{
"epoch": 0.34,
"learning_rate": 1.9798842236634797e-05,
"loss": 1.1157,
"step": 41
},
{
"epoch": 0.35,
"learning_rate": 1.9780208353736493e-05,
"loss": 1.1446,
"step": 42
},
{
"epoch": 0.36,
"learning_rate": 1.9760758775559275e-05,
"loss": 1.0876,
"step": 43
},
{
"epoch": 0.36,
"learning_rate": 1.9740495124249462e-05,
"loss": 1.0892,
"step": 44
},
{
"epoch": 0.37,
"learning_rate": 1.971941908984925e-05,
"loss": 1.0958,
"step": 45
},
{
"epoch": 0.38,
"learning_rate": 1.9697532430155718e-05,
"loss": 1.1405,
"step": 46
},
{
"epoch": 0.39,
"learning_rate": 1.9674836970574253e-05,
"loss": 0.963,
"step": 47
},
{
"epoch": 0.4,
"learning_rate": 1.9651334603966298e-05,
"loss": 1.0596,
"step": 48
},
{
"epoch": 0.4,
"learning_rate": 1.962702729049146e-05,
"loss": 1.2334,
"step": 49
},
{
"epoch": 0.41,
"learning_rate": 1.960191705744407e-05,
"loss": 1.0826,
"step": 50
},
{
"epoch": 0.42,
"learning_rate": 1.957600599908406e-05,
"loss": 0.9767,
"step": 51
},
{
"epoch": 0.43,
"learning_rate": 1.9549296276462326e-05,
"loss": 0.9579,
"step": 52
},
{
"epoch": 0.44,
"learning_rate": 1.9521790117240472e-05,
"loss": 0.9171,
"step": 53
},
{
"epoch": 0.45,
"learning_rate": 1.949348981550502e-05,
"loss": 1.1456,
"step": 54
},
{
"epoch": 0.45,
"learning_rate": 1.9464397731576093e-05,
"loss": 1.0953,
"step": 55
},
{
"epoch": 0.46,
"learning_rate": 1.943451629181054e-05,
"loss": 1.137,
"step": 56
},
{
"epoch": 0.47,
"learning_rate": 1.940384798839957e-05,
"loss": 1.1126,
"step": 57
},
{
"epoch": 0.48,
"learning_rate": 1.9372395379160913e-05,
"loss": 0.9716,
"step": 58
},
{
"epoch": 0.49,
"learning_rate": 1.9340161087325483e-05,
"loss": 0.9391,
"step": 59
},
{
"epoch": 0.5,
"learning_rate": 1.9307147801318585e-05,
"loss": 1.0386,
"step": 60
},
{
"epoch": 0.5,
"learning_rate": 1.9273358274535703e-05,
"loss": 0.9707,
"step": 61
},
{
"epoch": 0.51,
"learning_rate": 1.9238795325112867e-05,
"loss": 0.989,
"step": 62
},
{
"epoch": 0.52,
"learning_rate": 1.9203461835691596e-05,
"loss": 0.9676,
"step": 63
},
{
"epoch": 0.53,
"learning_rate": 1.916736075317848e-05,
"loss": 1.0501,
"step": 64
},
{
"epoch": 0.54,
"learning_rate": 1.9130495088499417e-05,
"loss": 0.9988,
"step": 65
},
{
"epoch": 0.55,
"learning_rate": 1.909286791634848e-05,
"loss": 1.0927,
"step": 66
},
{
"epoch": 0.55,
"learning_rate": 1.905448237493147e-05,
"loss": 1.0531,
"step": 67
},
{
"epoch": 0.56,
"learning_rate": 1.9015341665704206e-05,
"loss": 1.0239,
"step": 68
},
{
"epoch": 0.57,
"learning_rate": 1.8975449053105505e-05,
"loss": 1.0,
"step": 69
},
{
"epoch": 0.58,
"learning_rate": 1.8934807864284904e-05,
"loss": 0.967,
"step": 70
},
{
"epoch": 0.59,
"learning_rate": 1.889342148882519e-05,
"loss": 1.0019,
"step": 71
},
{
"epoch": 0.6,
"learning_rate": 1.8851293378459685e-05,
"loss": 0.9319,
"step": 72
},
{
"epoch": 0.6,
"learning_rate": 1.8808427046784365e-05,
"loss": 0.9753,
"step": 73
},
{
"epoch": 0.61,
"learning_rate": 1.876482606896482e-05,
"loss": 0.9619,
"step": 74
},
{
"epoch": 0.62,
"learning_rate": 1.872049408143808e-05,
"loss": 0.9042,
"step": 75
},
{
"epoch": 0.63,
"learning_rate": 1.8675434781609305e-05,
"loss": 1.0871,
"step": 76
},
{
"epoch": 0.64,
"learning_rate": 1.8629651927543447e-05,
"loss": 0.9747,
"step": 77
},
{
"epoch": 0.64,
"learning_rate": 1.858314933765178e-05,
"loss": 1.0254,
"step": 78
},
{
"epoch": 0.65,
"learning_rate": 1.8535930890373467e-05,
"loss": 1.0032,
"step": 79
},
{
"epoch": 0.66,
"learning_rate": 1.848800052385206e-05,
"loss": 0.8816,
"step": 80
},
{
"epoch": 0.67,
"learning_rate": 1.8439362235607074e-05,
"loss": 0.9593,
"step": 81
},
{
"epoch": 0.68,
"learning_rate": 1.8390020082200553e-05,
"loss": 1.012,
"step": 82
},
{
"epoch": 0.69,
"learning_rate": 1.833997817889878e-05,
"loss": 1.0523,
"step": 83
},
{
"epoch": 0.69,
"learning_rate": 1.828924069932902e-05,
"loss": 1.0015,
"step": 84
},
{
"epoch": 0.7,
"learning_rate": 1.8237811875131446e-05,
"loss": 0.9485,
"step": 85
},
{
"epoch": 0.71,
"learning_rate": 1.8185695995606196e-05,
"loss": 0.8034,
"step": 86
},
{
"epoch": 0.72,
"learning_rate": 1.8132897407355657e-05,
"loss": 0.9674,
"step": 87
},
{
"epoch": 0.73,
"learning_rate": 1.8079420513921913e-05,
"loss": 1.0236,
"step": 88
},
{
"epoch": 0.74,
"learning_rate": 1.802526977541951e-05,
"loss": 0.9783,
"step": 89
},
{
"epoch": 0.74,
"learning_rate": 1.7970449708163455e-05,
"loss": 1.0106,
"step": 90
},
{
"epoch": 0.75,
"learning_rate": 1.7914964884292543e-05,
"loss": 1.0462,
"step": 91
},
{
"epoch": 0.76,
"learning_rate": 1.785881993138803e-05,
"loss": 0.9723,
"step": 92
},
{
"epoch": 0.77,
"learning_rate": 1.7802019532087692e-05,
"loss": 0.9475,
"step": 93
},
{
"epoch": 0.78,
"learning_rate": 1.7744568423695257e-05,
"loss": 0.9122,
"step": 94
},
{
"epoch": 0.79,
"learning_rate": 1.7686471397785322e-05,
"loss": 0.9502,
"step": 95
},
{
"epoch": 0.79,
"learning_rate": 1.7627733299803714e-05,
"loss": 0.9869,
"step": 96
},
{
"epoch": 0.8,
"learning_rate": 1.7568359028663365e-05,
"loss": 0.8826,
"step": 97
},
{
"epoch": 0.81,
"learning_rate": 1.750835353633574e-05,
"loss": 0.9557,
"step": 98
},
{
"epoch": 0.82,
"learning_rate": 1.744772182743782e-05,
"loss": 0.979,
"step": 99
},
{
"epoch": 0.83,
"learning_rate": 1.7386468958814706e-05,
"loss": 1.0202,
"step": 100
},
{
"epoch": 0.83,
"learning_rate": 1.7324600039117862e-05,
"loss": 1.0266,
"step": 101
},
{
"epoch": 0.84,
"learning_rate": 1.7262120228379053e-05,
"loss": 0.9527,
"step": 102
},
{
"epoch": 0.85,
"learning_rate": 1.7199034737579962e-05,
"loss": 0.9514,
"step": 103
},
{
"epoch": 0.86,
"learning_rate": 1.71353488282176e-05,
"loss": 0.9188,
"step": 104
},
{
"epoch": 0.87,
"learning_rate": 1.7071067811865477e-05,
"loss": 0.8725,
"step": 105
},
{
"epoch": 0.88,
"learning_rate": 1.70061970497306e-05,
"loss": 0.9311,
"step": 106
},
{
"epoch": 0.88,
"learning_rate": 1.6940741952206342e-05,
"loss": 0.9228,
"step": 107
},
{
"epoch": 0.89,
"learning_rate": 1.687470797842118e-05,
"loss": 1.0714,
"step": 108
},
{
"epoch": 0.9,
"learning_rate": 1.680810063578342e-05,
"loss": 0.8246,
"step": 109
},
{
"epoch": 0.91,
"learning_rate": 1.6740925479521844e-05,
"loss": 0.9465,
"step": 110
},
{
"epoch": 0.92,
"learning_rate": 1.6673188112222394e-05,
"loss": 0.9637,
"step": 111
},
{
"epoch": 0.93,
"learning_rate": 1.66048941833609e-05,
"loss": 0.8188,
"step": 112
},
{
"epoch": 0.93,
"learning_rate": 1.6536049388831897e-05,
"loss": 0.864,
"step": 113
},
{
"epoch": 0.94,
"learning_rate": 1.646665947047358e-05,
"loss": 0.8149,
"step": 114
},
{
"epoch": 0.95,
"learning_rate": 1.6396730215588913e-05,
"loss": 0.9013,
"step": 115
},
{
"epoch": 0.96,
"learning_rate": 1.6326267456462965e-05,
"loss": 0.9409,
"step": 116
},
{
"epoch": 0.97,
"learning_rate": 1.6255277069876454e-05,
"loss": 0.9824,
"step": 117
},
{
"epoch": 0.98,
"learning_rate": 1.618376497661564e-05,
"loss": 0.9688,
"step": 118
},
{
"epoch": 0.98,
"learning_rate": 1.6111737140978495e-05,
"loss": 0.8832,
"step": 119
},
{
"epoch": 0.99,
"learning_rate": 1.603919957027727e-05,
"loss": 1.015,
"step": 120
},
{
"epoch": 1.0,
"learning_rate": 1.5966158314337472e-05,
"loss": 0.9867,
"step": 121
},
{
"epoch": 1.01,
"learning_rate": 1.589261946499329e-05,
"loss": 0.9525,
"step": 122
},
{
"epoch": 1.02,
"learning_rate": 1.581858915557953e-05,
"loss": 0.8725,
"step": 123
},
{
"epoch": 1.02,
"learning_rate": 1.574407356042005e-05,
"loss": 0.8788,
"step": 124
},
{
"epoch": 1.03,
"learning_rate": 1.5669078894312848e-05,
"loss": 0.997,
"step": 125
},
{
"epoch": 1.04,
"learning_rate": 1.5593611412011685e-05,
"loss": 0.9908,
"step": 126
},
{
"epoch": 1.05,
"learning_rate": 1.551767740770446e-05,
"loss": 0.8651,
"step": 127
},
{
"epoch": 1.06,
"learning_rate": 1.544128321448824e-05,
"loss": 0.9856,
"step": 128
},
{
"epoch": 1.07,
"learning_rate": 1.5364435203841058e-05,
"loss": 0.9477,
"step": 129
},
{
"epoch": 1.07,
"learning_rate": 1.5287139785090534e-05,
"loss": 0.872,
"step": 130
},
{
"epoch": 1.08,
"learning_rate": 1.5209403404879305e-05,
"loss": 0.882,
"step": 131
},
{
"epoch": 1.09,
"learning_rate": 1.5131232546627355e-05,
"loss": 0.9071,
"step": 132
},
{
"epoch": 1.1,
"learning_rate": 1.5052633729991296e-05,
"loss": 0.9204,
"step": 133
},
{
"epoch": 1.11,
"learning_rate": 1.4973613510320595e-05,
"loss": 0.9915,
"step": 134
},
{
"epoch": 1.12,
"learning_rate": 1.4894178478110856e-05,
"loss": 0.8548,
"step": 135
},
{
"epoch": 1.12,
"learning_rate": 1.4814335258454144e-05,
"loss": 0.8516,
"step": 136
},
{
"epoch": 1.13,
"learning_rate": 1.4734090510486435e-05,
"loss": 0.8408,
"step": 137
},
{
"epoch": 1.14,
"learning_rate": 1.4653450926832236e-05,
"loss": 0.9548,
"step": 138
},
{
"epoch": 1.15,
"learning_rate": 1.4572423233046386e-05,
"loss": 0.8366,
"step": 139
},
{
"epoch": 1.16,
"learning_rate": 1.449101418705315e-05,
"loss": 0.9097,
"step": 140
},
{
"epoch": 1.17,
"learning_rate": 1.4409230578582566e-05,
"loss": 0.9336,
"step": 141
},
{
"epoch": 1.17,
"learning_rate": 1.4327079228604177e-05,
"loss": 0.9178,
"step": 142
},
{
"epoch": 1.18,
"learning_rate": 1.4244566988758152e-05,
"loss": 0.9283,
"step": 143
},
{
"epoch": 1.19,
"learning_rate": 1.4161700740783815e-05,
"loss": 0.8792,
"step": 144
},
{
"epoch": 1.2,
"learning_rate": 1.4078487395945712e-05,
"loss": 0.9653,
"step": 145
},
{
"epoch": 1.21,
"learning_rate": 1.3994933894457193e-05,
"loss": 0.8916,
"step": 146
},
{
"epoch": 1.21,
"learning_rate": 1.391104720490156e-05,
"loss": 0.9913,
"step": 147
},
{
"epoch": 1.22,
"learning_rate": 1.3826834323650899e-05,
"loss": 0.8563,
"step": 148
},
{
"epoch": 1.23,
"learning_rate": 1.3742302274282532e-05,
"loss": 0.9575,
"step": 149
},
{
"epoch": 1.24,
"learning_rate": 1.3657458106993257e-05,
"loss": 0.9216,
"step": 150
},
{
"epoch": 1.25,
"learning_rate": 1.3572308898011328e-05,
"loss": 0.9395,
"step": 151
},
{
"epoch": 1.26,
"learning_rate": 1.3486861749006286e-05,
"loss": 0.8954,
"step": 152
},
{
"epoch": 1.26,
"learning_rate": 1.3401123786496664e-05,
"loss": 0.9628,
"step": 153
},
{
"epoch": 1.27,
"learning_rate": 1.3315102161255603e-05,
"loss": 0.8323,
"step": 154
},
{
"epoch": 1.28,
"learning_rate": 1.3228804047714462e-05,
"loss": 0.8951,
"step": 155
},
{
"epoch": 1.29,
"learning_rate": 1.3142236643364481e-05,
"loss": 0.9588,
"step": 156
},
{
"epoch": 1.3,
"learning_rate": 1.3055407168156438e-05,
"loss": 0.8521,
"step": 157
},
{
"epoch": 1.31,
"learning_rate": 1.2968322863898533e-05,
"loss": 0.9239,
"step": 158
},
{
"epoch": 1.31,
"learning_rate": 1.2880990993652379e-05,
"loss": 0.8886,
"step": 159
},
{
"epoch": 1.32,
"learning_rate": 1.2793418841127242e-05,
"loss": 0.8379,
"step": 160
},
{
"epoch": 1.33,
"learning_rate": 1.2705613710072575e-05,
"loss": 0.9035,
"step": 161
},
{
"epoch": 1.34,
"learning_rate": 1.2617582923668855e-05,
"loss": 0.8227,
"step": 162
},
{
"epoch": 1.35,
"learning_rate": 1.2529333823916807e-05,
"loss": 0.9045,
"step": 163
},
{
"epoch": 1.36,
"learning_rate": 1.2440873771025079e-05,
"loss": 0.8571,
"step": 164
},
{
"epoch": 1.36,
"learning_rate": 1.2352210142796359e-05,
"loss": 0.8233,
"step": 165
},
{
"epoch": 1.37,
"learning_rate": 1.2263350334012059e-05,
"loss": 0.9579,
"step": 166
},
{
"epoch": 1.38,
"learning_rate": 1.2174301755815572e-05,
"loss": 0.9478,
"step": 167
},
{
"epoch": 1.39,
"learning_rate": 1.208507183509416e-05,
"loss": 0.8041,
"step": 168
},
{
"epoch": 1.4,
"learning_rate": 1.199566801385953e-05,
"loss": 0.8119,
"step": 169
},
{
"epoch": 1.4,
"learning_rate": 1.190609774862715e-05,
"loss": 0.9701,
"step": 170
},
{
"epoch": 1.41,
"learning_rate": 1.1816368509794365e-05,
"loss": 0.9351,
"step": 171
},
{
"epoch": 1.42,
"learning_rate": 1.1726487781017337e-05,
"loss": 0.8202,
"step": 172
},
{
"epoch": 1.43,
"learning_rate": 1.1636463058586882e-05,
"loss": 0.8849,
"step": 173
},
{
"epoch": 1.44,
"learning_rate": 1.1546301850803283e-05,
"loss": 0.9304,
"step": 174
},
{
"epoch": 1.45,
"learning_rate": 1.1456011677350052e-05,
"loss": 0.9295,
"step": 175
},
{
"epoch": 1.45,
"learning_rate": 1.1365600068666781e-05,
"loss": 0.9849,
"step": 176
},
{
"epoch": 1.46,
"learning_rate": 1.127507456532108e-05,
"loss": 0.7768,
"step": 177
},
{
"epoch": 1.47,
"learning_rate": 1.1184442717379687e-05,
"loss": 1.0072,
"step": 178
},
{
"epoch": 1.48,
"learning_rate": 1.1093712083778748e-05,
"loss": 0.9502,
"step": 179
},
{
"epoch": 1.49,
"learning_rate": 1.1002890231693395e-05,
"loss": 0.7969,
"step": 180
},
{
"epoch": 1.5,
"learning_rate": 1.0911984735906635e-05,
"loss": 0.8685,
"step": 181
},
{
"epoch": 1.5,
"learning_rate": 1.0821003178177572e-05,
"loss": 0.8699,
"step": 182
},
{
"epoch": 1.51,
"learning_rate": 1.0729953146609076e-05,
"loss": 0.9276,
"step": 183
},
{
"epoch": 1.52,
"learning_rate": 1.0638842235014923e-05,
"loss": 0.8302,
"step": 184
},
{
"epoch": 1.53,
"learning_rate": 1.0547678042286435e-05,
"loss": 0.7993,
"step": 185
},
{
"epoch": 1.54,
"learning_rate": 1.045646817175874e-05,
"loss": 0.8794,
"step": 186
},
{
"epoch": 1.55,
"learning_rate": 1.0365220230576592e-05,
"loss": 0.8809,
"step": 187
},
{
"epoch": 1.55,
"learning_rate": 1.027394182905995e-05,
"loss": 0.9458,
"step": 188
},
{
"epoch": 1.56,
"learning_rate": 1.0182640580069249e-05,
"loss": 0.7746,
"step": 189
},
{
"epoch": 1.57,
"learning_rate": 1.009132409837046e-05,
"loss": 0.8053,
"step": 190
},
{
"epoch": 1.58,
"learning_rate": 1e-05,
"loss": 0.8711,
"step": 191
},
{
"epoch": 1.59,
"learning_rate": 9.908675901629542e-06,
"loss": 0.8419,
"step": 192
},
{
"epoch": 1.6,
"learning_rate": 9.817359419930753e-06,
"loss": 0.8391,
"step": 193
},
{
"epoch": 1.6,
"learning_rate": 9.726058170940053e-06,
"loss": 0.8785,
"step": 194
},
{
"epoch": 1.61,
"learning_rate": 9.634779769423412e-06,
"loss": 0.8864,
"step": 195
},
{
"epoch": 1.62,
"learning_rate": 9.543531828241263e-06,
"loss": 0.9405,
"step": 196
},
{
"epoch": 1.63,
"learning_rate": 9.452321957713563e-06,
"loss": 0.8087,
"step": 197
},
{
"epoch": 1.64,
"learning_rate": 9.361157764985079e-06,
"loss": 0.9207,
"step": 198
},
{
"epoch": 1.64,
"learning_rate": 9.270046853390924e-06,
"loss": 0.8302,
"step": 199
},
{
"epoch": 1.65,
"learning_rate": 9.17899682182243e-06,
"loss": 0.9462,
"step": 200
},
{
"epoch": 1.66,
"learning_rate": 9.088015264093365e-06,
"loss": 1.0384,
"step": 201
},
{
"epoch": 1.67,
"learning_rate": 8.997109768306607e-06,
"loss": 0.9399,
"step": 202
},
{
"epoch": 1.68,
"learning_rate": 8.906287916221259e-06,
"loss": 0.8899,
"step": 203
},
{
"epoch": 1.69,
"learning_rate": 8.81555728262032e-06,
"loss": 0.9902,
"step": 204
},
{
"epoch": 1.69,
"learning_rate": 8.724925434678923e-06,
"loss": 0.8855,
"step": 205
},
{
"epoch": 1.7,
"learning_rate": 8.634399931333226e-06,
"loss": 0.8417,
"step": 206
},
{
"epoch": 1.71,
"learning_rate": 8.543988322649954e-06,
"loss": 0.8521,
"step": 207
},
{
"epoch": 1.72,
"learning_rate": 8.45369814919672e-06,
"loss": 0.867,
"step": 208
},
{
"epoch": 1.73,
"learning_rate": 8.363536941413121e-06,
"loss": 0.8874,
"step": 209
},
{
"epoch": 1.74,
"learning_rate": 8.273512218982666e-06,
"loss": 0.8213,
"step": 210
},
{
"epoch": 1.74,
"learning_rate": 8.183631490205636e-06,
"loss": 0.7704,
"step": 211
},
{
"epoch": 1.75,
"learning_rate": 8.093902251372854e-06,
"loss": 0.8162,
"step": 212
},
{
"epoch": 1.76,
"learning_rate": 8.004331986140474e-06,
"loss": 0.8831,
"step": 213
},
{
"epoch": 1.77,
"learning_rate": 7.914928164905844e-06,
"loss": 0.7004,
"step": 214
},
{
"epoch": 1.78,
"learning_rate": 7.825698244184432e-06,
"loss": 0.8225,
"step": 215
},
{
"epoch": 1.79,
"learning_rate": 7.736649665987944e-06,
"loss": 0.823,
"step": 216
},
{
"epoch": 1.79,
"learning_rate": 7.647789857203644e-06,
"loss": 0.8252,
"step": 217
},
{
"epoch": 1.8,
"learning_rate": 7.559126228974921e-06,
"loss": 0.8729,
"step": 218
},
{
"epoch": 1.81,
"learning_rate": 7.470666176083193e-06,
"loss": 0.9415,
"step": 219
},
{
"epoch": 1.82,
"learning_rate": 7.382417076331148e-06,
"loss": 0.9198,
"step": 220
},
{
"epoch": 1.83,
"learning_rate": 7.294386289927425e-06,
"loss": 0.9263,
"step": 221
},
{
"epoch": 1.83,
"learning_rate": 7.206581158872761e-06,
"loss": 0.9364,
"step": 222
},
{
"epoch": 1.84,
"learning_rate": 7.119009006347625e-06,
"loss": 0.8889,
"step": 223
},
{
"epoch": 1.85,
"learning_rate": 7.031677136101471e-06,
"loss": 0.8712,
"step": 224
},
{
"epoch": 1.86,
"learning_rate": 6.944592831843566e-06,
"loss": 0.895,
"step": 225
},
{
"epoch": 1.87,
"learning_rate": 6.857763356635525e-06,
"loss": 0.7805,
"step": 226
},
{
"epoch": 1.88,
"learning_rate": 6.771195952285541e-06,
"loss": 0.7608,
"step": 227
},
{
"epoch": 1.88,
"learning_rate": 6.684897838744403e-06,
"loss": 0.8335,
"step": 228
},
{
"epoch": 1.89,
"learning_rate": 6.5988762135033405e-06,
"loss": 0.8861,
"step": 229
},
{
"epoch": 1.9,
"learning_rate": 6.513138250993716e-06,
"loss": 0.9386,
"step": 230
},
{
"epoch": 1.91,
"learning_rate": 6.427691101988673e-06,
"loss": 0.9067,
"step": 231
},
{
"epoch": 1.92,
"learning_rate": 6.342541893006746e-06,
"loss": 0.9105,
"step": 232
},
{
"epoch": 1.93,
"learning_rate": 6.257697725717469e-06,
"loss": 0.8314,
"step": 233
},
{
"epoch": 1.93,
"learning_rate": 6.173165676349103e-06,
"loss": 0.8538,
"step": 234
},
{
"epoch": 1.94,
"learning_rate": 6.088952795098442e-06,
"loss": 0.7411,
"step": 235
},
{
"epoch": 1.95,
"learning_rate": 6.00506610554281e-06,
"loss": 0.8632,
"step": 236
},
{
"epoch": 1.96,
"learning_rate": 5.921512604054289e-06,
"loss": 0.8846,
"step": 237
},
{
"epoch": 1.97,
"learning_rate": 5.838299259216187e-06,
"loss": 0.8266,
"step": 238
},
{
"epoch": 1.98,
"learning_rate": 5.755433011241851e-06,
"loss": 0.9274,
"step": 239
},
{
"epoch": 1.98,
"learning_rate": 5.672920771395822e-06,
"loss": 0.9221,
"step": 240
},
{
"epoch": 1.99,
"learning_rate": 5.590769421417435e-06,
"loss": 0.8928,
"step": 241
},
{
"epoch": 2.0,
"learning_rate": 5.50898581294685e-06,
"loss": 0.8753,
"step": 242
},
{
"epoch": 2.01,
"learning_rate": 5.427576766953615e-06,
"loss": 0.8497,
"step": 243
},
{
"epoch": 2.02,
"learning_rate": 5.346549073167766e-06,
"loss": 0.7893,
"step": 244
},
{
"epoch": 2.02,
"learning_rate": 5.265909489513568e-06,
"loss": 0.8574,
"step": 245
},
{
"epoch": 2.03,
"learning_rate": 5.185664741545862e-06,
"loss": 0.7664,
"step": 246
},
{
"epoch": 2.04,
"learning_rate": 5.105821521889147e-06,
"loss": 0.7805,
"step": 247
},
{
"epoch": 2.05,
"learning_rate": 5.026386489679408e-06,
"loss": 0.7907,
"step": 248
},
{
"epoch": 2.06,
"learning_rate": 4.947366270008708e-06,
"loss": 0.9125,
"step": 249
},
{
"epoch": 2.07,
"learning_rate": 4.868767453372649e-06,
"loss": 0.9681,
"step": 250
},
{
"epoch": 2.07,
"learning_rate": 4.790596595120699e-06,
"loss": 0.9316,
"step": 251
},
{
"epoch": 2.08,
"learning_rate": 4.712860214909466e-06,
"loss": 0.8419,
"step": 252
},
{
"epoch": 2.09,
"learning_rate": 4.635564796158946e-06,
"loss": 0.8667,
"step": 253
},
{
"epoch": 2.1,
"learning_rate": 4.558716785511764e-06,
"loss": 0.753,
"step": 254
},
{
"epoch": 2.11,
"learning_rate": 4.482322592295541e-06,
"loss": 0.8154,
"step": 255
},
{
"epoch": 2.12,
"learning_rate": 4.406388587988318e-06,
"loss": 0.8494,
"step": 256
},
{
"epoch": 2.12,
"learning_rate": 4.330921105687155e-06,
"loss": 0.7057,
"step": 257
},
{
"epoch": 2.13,
"learning_rate": 4.255926439579948e-06,
"loss": 0.8995,
"step": 258
},
{
"epoch": 2.14,
"learning_rate": 4.181410844420473e-06,
"loss": 0.8642,
"step": 259
},
{
"epoch": 2.15,
"learning_rate": 4.107380535006709e-06,
"loss": 0.7852,
"step": 260
},
{
"epoch": 2.16,
"learning_rate": 4.033841685662529e-06,
"loss": 0.7933,
"step": 261
},
{
"epoch": 2.17,
"learning_rate": 3.960800429722734e-06,
"loss": 0.8312,
"step": 262
},
{
"epoch": 2.17,
"learning_rate": 3.888262859021508e-06,
"loss": 0.8643,
"step": 263
},
{
"epoch": 2.18,
"learning_rate": 3.81623502338436e-06,
"loss": 0.775,
"step": 264
},
{
"epoch": 2.19,
"learning_rate": 3.7447229301235443e-06,
"loss": 0.8607,
"step": 265
},
{
"epoch": 2.2,
"learning_rate": 3.6737325435370376e-06,
"loss": 0.8145,
"step": 266
},
{
"epoch": 2.21,
"learning_rate": 3.6032697844110896e-06,
"loss": 0.8945,
"step": 267
},
{
"epoch": 2.21,
"learning_rate": 3.5333405295264255e-06,
"loss": 0.9856,
"step": 268
},
{
"epoch": 2.22,
"learning_rate": 3.463950611168111e-06,
"loss": 0.8377,
"step": 269
},
{
"epoch": 2.23,
"learning_rate": 3.395105816639106e-06,
"loss": 0.915,
"step": 270
},
{
"epoch": 2.24,
"learning_rate": 3.326811887777607e-06,
"loss": 0.8007,
"step": 271
},
{
"epoch": 2.25,
"learning_rate": 3.2590745204781537e-06,
"loss": 0.8139,
"step": 272
},
{
"epoch": 2.26,
"learning_rate": 3.191899364216581e-06,
"loss": 0.8724,
"step": 273
},
{
"epoch": 2.26,
"learning_rate": 3.125292021578822e-06,
"loss": 0.8596,
"step": 274
},
{
"epoch": 2.27,
"learning_rate": 3.0592580477936606e-06,
"loss": 0.8309,
"step": 275
},
{
"epoch": 2.28,
"learning_rate": 2.993802950269402e-06,
"loss": 0.8712,
"step": 276
},
{
"epoch": 2.29,
"learning_rate": 2.9289321881345257e-06,
"loss": 0.8146,
"step": 277
},
{
"epoch": 2.3,
"learning_rate": 2.864651171782402e-06,
"loss": 0.814,
"step": 278
},
{
"epoch": 2.31,
"learning_rate": 2.8009652624200436e-06,
"loss": 0.7858,
"step": 279
},
{
"epoch": 2.31,
"learning_rate": 2.7378797716209506e-06,
"loss": 0.8521,
"step": 280
},
{
"epoch": 2.32,
"learning_rate": 2.675399960882138e-06,
"loss": 0.9599,
"step": 281
},
{
"epoch": 2.33,
"learning_rate": 2.6135310411852977e-06,
"loss": 0.7842,
"step": 282
},
{
"epoch": 2.34,
"learning_rate": 2.5522781725621814e-06,
"loss": 0.8024,
"step": 283
},
{
"epoch": 2.35,
"learning_rate": 2.491646463664261e-06,
"loss": 0.7793,
"step": 284
},
{
"epoch": 2.36,
"learning_rate": 2.4316409713366353e-06,
"loss": 0.8248,
"step": 285
},
{
"epoch": 2.36,
"learning_rate": 2.3722667001962898e-06,
"loss": 0.7695,
"step": 286
},
{
"epoch": 2.37,
"learning_rate": 2.3135286022146785e-06,
"loss": 0.8053,
"step": 287
},
{
"epoch": 2.38,
"learning_rate": 2.255431576304744e-06,
"loss": 0.8474,
"step": 288
},
{
"epoch": 2.39,
"learning_rate": 2.1979804679123108e-06,
"loss": 0.9379,
"step": 289
},
{
"epoch": 2.4,
"learning_rate": 2.141180068611971e-06,
"loss": 0.8415,
"step": 290
},
{
"epoch": 2.4,
"learning_rate": 2.08503511570746e-06,
"loss": 0.7226,
"step": 291
},
{
"epoch": 2.41,
"learning_rate": 2.0295502918365473e-06,
"loss": 0.8963,
"step": 292
},
{
"epoch": 2.42,
"learning_rate": 1.9747302245804944e-06,
"loss": 0.8423,
"step": 293
},
{
"epoch": 2.43,
"learning_rate": 1.920579486078091e-06,
"loss": 0.7995,
"step": 294
},
{
"epoch": 2.44,
"learning_rate": 1.8671025926443464e-06,
"loss": 0.8051,
"step": 295
},
{
"epoch": 2.45,
"learning_rate": 1.8143040043938054e-06,
"loss": 0.7731,
"step": 296
},
{
"epoch": 2.45,
"learning_rate": 1.7621881248685569e-06,
"loss": 0.8458,
"step": 297
},
{
"epoch": 2.46,
"learning_rate": 1.7107593006709799e-06,
"loss": 0.7302,
"step": 298
},
{
"epoch": 2.47,
"learning_rate": 1.660021821101222e-06,
"loss": 0.881,
"step": 299
},
{
"epoch": 2.48,
"learning_rate": 1.6099799177994491e-06,
"loss": 0.7728,
"step": 300
},
{
"epoch": 2.49,
"learning_rate": 1.5606377643929305e-06,
"loss": 0.8095,
"step": 301
},
{
"epoch": 2.5,
"learning_rate": 1.5119994761479429e-06,
"loss": 0.8324,
"step": 302
},
{
"epoch": 2.5,
"learning_rate": 1.4640691096265358e-06,
"loss": 0.8192,
"step": 303
},
{
"epoch": 2.51,
"learning_rate": 1.4168506623482202e-06,
"loss": 0.7712,
"step": 304
},
{
"epoch": 2.52,
"learning_rate": 1.3703480724565577e-06,
"loss": 0.7861,
"step": 305
},
{
"epoch": 2.53,
"learning_rate": 1.3245652183906965e-06,
"loss": 0.7886,
"step": 306
},
{
"epoch": 2.54,
"learning_rate": 1.279505918561923e-06,
"loss": 0.8942,
"step": 307
},
{
"epoch": 2.55,
"learning_rate": 1.2351739310351796e-06,
"loss": 0.911,
"step": 308
},
{
"epoch": 2.55,
"learning_rate": 1.1915729532156372e-06,
"loss": 0.7743,
"step": 309
},
{
"epoch": 2.56,
"learning_rate": 1.1487066215403186e-06,
"loss": 0.8076,
"step": 310
},
{
"epoch": 2.57,
"learning_rate": 1.1065785111748117e-06,
"loss": 0.7571,
"step": 311
},
{
"epoch": 2.58,
"learning_rate": 1.0651921357150997e-06,
"loss": 0.8203,
"step": 312
},
{
"epoch": 2.59,
"learning_rate": 1.0245509468944992e-06,
"loss": 0.7785,
"step": 313
},
{
"epoch": 2.6,
"learning_rate": 9.84658334295796e-07,
"loss": 0.8648,
"step": 314
},
{
"epoch": 2.6,
"learning_rate": 9.455176250685338e-07,
"loss": 0.8401,
"step": 315
},
{
"epoch": 2.61,
"learning_rate": 9.071320836515263e-07,
"loss": 0.8123,
"step": 316
},
{
"epoch": 2.62,
"learning_rate": 8.695049115005838e-07,
"loss": 0.7879,
"step": 317
},
{
"epoch": 2.63,
"learning_rate": 8.326392468215206e-07,
"loss": 0.8958,
"step": 318
},
{
"epoch": 2.64,
"learning_rate": 7.965381643084069e-07,
"loss": 0.8691,
"step": 319
},
{
"epoch": 2.64,
"learning_rate": 7.612046748871327e-07,
"loss": 0.7841,
"step": 320
},
{
"epoch": 2.65,
"learning_rate": 7.266417254642966e-07,
"loss": 0.7674,
"step": 321
},
{
"epoch": 2.66,
"learning_rate": 6.928521986814196e-07,
"loss": 0.7693,
"step": 322
},
{
"epoch": 2.67,
"learning_rate": 6.598389126745209e-07,
"loss": 0.8653,
"step": 323
},
{
"epoch": 2.68,
"learning_rate": 6.276046208390873e-07,
"loss": 0.8795,
"step": 324
},
{
"epoch": 2.69,
"learning_rate": 5.961520116004326e-07,
"loss": 0.8787,
"step": 325
},
{
"epoch": 2.69,
"learning_rate": 5.654837081894626e-07,
"loss": 0.8687,
"step": 326
},
{
"epoch": 2.7,
"learning_rate": 5.35602268423906e-07,
"loss": 0.8541,
"step": 327
},
{
"epoch": 2.71,
"learning_rate": 5.065101844949793e-07,
"loss": 0.9161,
"step": 328
},
{
"epoch": 2.72,
"learning_rate": 4.782098827595305e-07,
"loss": 0.7652,
"step": 329
},
{
"epoch": 2.73,
"learning_rate": 4.5070372353767543e-07,
"loss": 0.8382,
"step": 330
},
{
"epoch": 2.74,
"learning_rate": 4.2399400091594154e-07,
"loss": 0.9081,
"step": 331
},
{
"epoch": 2.74,
"learning_rate": 3.9808294255593293e-07,
"loss": 0.8119,
"step": 332
},
{
"epoch": 2.75,
"learning_rate": 3.7297270950854224e-07,
"loss": 1.0043,
"step": 333
},
{
"epoch": 2.76,
"learning_rate": 3.486653960337061e-07,
"loss": 0.8723,
"step": 334
},
{
"epoch": 2.77,
"learning_rate": 3.2516302942574794e-07,
"loss": 0.7953,
"step": 335
},
{
"epoch": 2.78,
"learning_rate": 3.024675698442858e-07,
"loss": 0.8223,
"step": 336
},
{
"epoch": 2.79,
"learning_rate": 2.805809101507539e-07,
"loss": 0.9616,
"step": 337
},
{
"epoch": 2.79,
"learning_rate": 2.595048757505392e-07,
"loss": 0.7757,
"step": 338
},
{
"epoch": 2.8,
"learning_rate": 2.392412244407294e-07,
"loss": 0.833,
"step": 339
},
{
"epoch": 2.81,
"learning_rate": 2.1979164626350745e-07,
"loss": 0.8236,
"step": 340
},
{
"epoch": 2.82,
"learning_rate": 2.0115776336520622e-07,
"loss": 0.8654,
"step": 341
},
{
"epoch": 2.83,
"learning_rate": 1.8334112986100994e-07,
"loss": 0.818,
"step": 342
},
{
"epoch": 2.83,
"learning_rate": 1.6634323170533928e-07,
"loss": 0.8791,
"step": 343
},
{
"epoch": 2.84,
"learning_rate": 1.5016548656791697e-07,
"loss": 0.9477,
"step": 344
},
{
"epoch": 2.85,
"learning_rate": 1.348092437155346e-07,
"loss": 0.8957,
"step": 345
},
{
"epoch": 2.86,
"learning_rate": 1.20275783899515e-07,
"loss": 0.8467,
"step": 346
},
{
"epoch": 2.87,
"learning_rate": 1.0656631924889749e-07,
"loss": 0.8196,
"step": 347
},
{
"epoch": 2.88,
"learning_rate": 9.368199316934446e-08,
"loss": 0.7764,
"step": 348
},
{
"epoch": 2.88,
"learning_rate": 8.162388024777202e-08,
"loss": 0.8987,
"step": 349
},
{
"epoch": 2.89,
"learning_rate": 7.039298616273393e-08,
"loss": 0.9182,
"step": 350
},
{
"epoch": 2.9,
"learning_rate": 5.999024760054095e-08,
"loss": 0.8248,
"step": 351
},
{
"epoch": 2.91,
"learning_rate": 5.041653217713993e-08,
"loss": 0.8595,
"step": 352
},
{
"epoch": 2.92,
"learning_rate": 4.167263836575286e-08,
"loss": 0.6834,
"step": 353
},
{
"epoch": 2.93,
"learning_rate": 3.3759295430281226e-08,
"loss": 0.6846,
"step": 354
},
{
"epoch": 2.93,
"learning_rate": 2.667716336448356e-08,
"loss": 0.8627,
"step": 355
},
{
"epoch": 2.94,
"learning_rate": 2.0426832836930588e-08,
"loss": 0.8074,
"step": 356
},
{
"epoch": 2.95,
"learning_rate": 1.50088251417424e-08,
"loss": 0.8568,
"step": 357
},
{
"epoch": 2.96,
"learning_rate": 1.0423592155108798e-08,
"loss": 0.8512,
"step": 358
},
{
"epoch": 2.97,
"learning_rate": 6.671516297606095e-09,
"loss": 0.8249,
"step": 359
},
{
"epoch": 2.98,
"learning_rate": 3.7529105022970915e-09,
"loss": 0.8835,
"step": 360
},
{
"epoch": 2.98,
"learning_rate": 1.6680181886352676e-09,
"loss": 0.8837,
"step": 361
},
{
"epoch": 2.99,
"learning_rate": 4.170132421610351e-10,
"loss": 0.8092,
"step": 362
},
{
"epoch": 3.0,
"learning_rate": 0.0,
"loss": 0.8638,
"step": 363
},
{
"epoch": 3.0,
"step": 363,
"total_flos": 457885308682240.0,
"train_loss": 0.9454823605942003,
"train_runtime": 6836.4148,
"train_samples_per_second": 5.088,
"train_steps_per_second": 0.053
}
],
"logging_steps": 1.0,
"max_steps": 363,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"total_flos": 457885308682240.0,
"train_batch_size": 12,
"trial_name": null,
"trial_params": null
}