End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.01,
|
| 3 |
+
"eval_accuracy": 1.0,
|
| 4 |
+
"eval_loss": 0.008221009746193886,
|
| 5 |
+
"eval_runtime": 0.5785,
|
| 6 |
+
"eval_samples_per_second": 3.457,
|
| 7 |
+
"eval_steps_per_second": 3.457
|
| 8 |
+
}
|
runs/Mar15_16-16-03_nphe645fj4/events.out.tfevents.1678897685.nphe645fj4.32.11
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f76253244640b05b42bdc1e4b4f3a32625107a32a4f908057433df352022cbb9
|
| 3 |
+
size 686
|
test_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.01,
|
| 3 |
+
"eval_accuracy": 1.0,
|
| 4 |
+
"eval_loss": 0.008221009746193886,
|
| 5 |
+
"eval_runtime": 0.5785,
|
| 6 |
+
"eval_samples_per_second": 3.457,
|
| 7 |
+
"eval_steps_per_second": 3.457
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,574 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.0,
|
| 3 |
+
"best_model_checkpoint": "videomae-base-finetuned-basketball-subset-v2/checkpoint-404",
|
| 4 |
+
"epoch": 4.009803921568627,
|
| 5 |
+
"global_step": 816,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.01,
|
| 12 |
+
"learning_rate": 6.0975609756097564e-06,
|
| 13 |
+
"loss": 0.6571,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.02,
|
| 18 |
+
"learning_rate": 1.2195121951219513e-05,
|
| 19 |
+
"loss": 0.6573,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.04,
|
| 24 |
+
"learning_rate": 1.8292682926829268e-05,
|
| 25 |
+
"loss": 0.7806,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.05,
|
| 30 |
+
"learning_rate": 2.4390243902439026e-05,
|
| 31 |
+
"loss": 0.8591,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.06,
|
| 36 |
+
"learning_rate": 3.048780487804878e-05,
|
| 37 |
+
"loss": 0.6489,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.07,
|
| 42 |
+
"learning_rate": 3.6585365853658535e-05,
|
| 43 |
+
"loss": 0.7533,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.09,
|
| 48 |
+
"learning_rate": 4.26829268292683e-05,
|
| 49 |
+
"loss": 0.576,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.1,
|
| 54 |
+
"learning_rate": 4.878048780487805e-05,
|
| 55 |
+
"loss": 1.0713,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.11,
|
| 60 |
+
"learning_rate": 4.945504087193461e-05,
|
| 61 |
+
"loss": 0.4372,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.12,
|
| 66 |
+
"learning_rate": 4.877384196185286e-05,
|
| 67 |
+
"loss": 0.5813,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.13,
|
| 72 |
+
"learning_rate": 4.809264305177112e-05,
|
| 73 |
+
"loss": 1.1109,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.15,
|
| 78 |
+
"learning_rate": 4.741144414168938e-05,
|
| 79 |
+
"loss": 1.4496,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.16,
|
| 84 |
+
"learning_rate": 4.673024523160763e-05,
|
| 85 |
+
"loss": 0.6032,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.17,
|
| 90 |
+
"learning_rate": 4.604904632152589e-05,
|
| 91 |
+
"loss": 2.0923,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.18,
|
| 96 |
+
"learning_rate": 4.5367847411444145e-05,
|
| 97 |
+
"loss": 1.4158,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.2,
|
| 102 |
+
"learning_rate": 4.46866485013624e-05,
|
| 103 |
+
"loss": 1.5632,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.21,
|
| 108 |
+
"learning_rate": 4.400544959128066e-05,
|
| 109 |
+
"loss": 1.1707,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.22,
|
| 114 |
+
"learning_rate": 4.332425068119891e-05,
|
| 115 |
+
"loss": 0.4736,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.23,
|
| 120 |
+
"learning_rate": 4.264305177111717e-05,
|
| 121 |
+
"loss": 2.0593,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.25,
|
| 126 |
+
"learning_rate": 4.196185286103542e-05,
|
| 127 |
+
"loss": 1.7747,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.25,
|
| 132 |
+
"eval_accuracy": 0.5,
|
| 133 |
+
"eval_loss": 2.5059080123901367,
|
| 134 |
+
"eval_runtime": 0.4908,
|
| 135 |
+
"eval_samples_per_second": 4.075,
|
| 136 |
+
"eval_steps_per_second": 4.075,
|
| 137 |
+
"step": 202
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.01,
|
| 141 |
+
"learning_rate": 4.128065395095368e-05,
|
| 142 |
+
"loss": 0.706,
|
| 143 |
+
"step": 210
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 1.02,
|
| 147 |
+
"learning_rate": 4.0599455040871935e-05,
|
| 148 |
+
"loss": 1.2504,
|
| 149 |
+
"step": 220
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.03,
|
| 153 |
+
"learning_rate": 3.9918256130790195e-05,
|
| 154 |
+
"loss": 1.8594,
|
| 155 |
+
"step": 230
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 1.05,
|
| 159 |
+
"learning_rate": 3.923705722070845e-05,
|
| 160 |
+
"loss": 0.8968,
|
| 161 |
+
"step": 240
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 1.06,
|
| 165 |
+
"learning_rate": 3.85558583106267e-05,
|
| 166 |
+
"loss": 0.739,
|
| 167 |
+
"step": 250
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 1.07,
|
| 171 |
+
"learning_rate": 3.787465940054496e-05,
|
| 172 |
+
"loss": 1.2789,
|
| 173 |
+
"step": 260
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 1.08,
|
| 177 |
+
"learning_rate": 3.719346049046322e-05,
|
| 178 |
+
"loss": 0.8226,
|
| 179 |
+
"step": 270
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.1,
|
| 183 |
+
"learning_rate": 3.651226158038147e-05,
|
| 184 |
+
"loss": 0.9322,
|
| 185 |
+
"step": 280
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.11,
|
| 189 |
+
"learning_rate": 3.583106267029973e-05,
|
| 190 |
+
"loss": 0.4803,
|
| 191 |
+
"step": 290
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.12,
|
| 195 |
+
"learning_rate": 3.5149863760217985e-05,
|
| 196 |
+
"loss": 0.4243,
|
| 197 |
+
"step": 300
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.13,
|
| 201 |
+
"learning_rate": 3.446866485013624e-05,
|
| 202 |
+
"loss": 0.0077,
|
| 203 |
+
"step": 310
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 1.14,
|
| 207 |
+
"learning_rate": 3.37874659400545e-05,
|
| 208 |
+
"loss": 0.5999,
|
| 209 |
+
"step": 320
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 1.16,
|
| 213 |
+
"learning_rate": 3.310626702997276e-05,
|
| 214 |
+
"loss": 1.2867,
|
| 215 |
+
"step": 330
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 1.17,
|
| 219 |
+
"learning_rate": 3.242506811989101e-05,
|
| 220 |
+
"loss": 2.3124,
|
| 221 |
+
"step": 340
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.18,
|
| 225 |
+
"learning_rate": 3.174386920980927e-05,
|
| 226 |
+
"loss": 0.9884,
|
| 227 |
+
"step": 350
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 1.19,
|
| 231 |
+
"learning_rate": 3.106267029972752e-05,
|
| 232 |
+
"loss": 0.2988,
|
| 233 |
+
"step": 360
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.21,
|
| 237 |
+
"learning_rate": 3.038147138964578e-05,
|
| 238 |
+
"loss": 0.9146,
|
| 239 |
+
"step": 370
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.22,
|
| 243 |
+
"learning_rate": 2.9700272479564035e-05,
|
| 244 |
+
"loss": 1.9739,
|
| 245 |
+
"step": 380
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 1.23,
|
| 249 |
+
"learning_rate": 2.901907356948229e-05,
|
| 250 |
+
"loss": 0.7482,
|
| 251 |
+
"step": 390
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.24,
|
| 255 |
+
"learning_rate": 2.8337874659400547e-05,
|
| 256 |
+
"loss": 0.0199,
|
| 257 |
+
"step": 400
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.25,
|
| 261 |
+
"eval_accuracy": 1.0,
|
| 262 |
+
"eval_loss": 0.008221009746193886,
|
| 263 |
+
"eval_runtime": 0.4787,
|
| 264 |
+
"eval_samples_per_second": 4.178,
|
| 265 |
+
"eval_steps_per_second": 4.178,
|
| 266 |
+
"step": 404
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 2.01,
|
| 270 |
+
"learning_rate": 2.76566757493188e-05,
|
| 271 |
+
"loss": 1.4992,
|
| 272 |
+
"step": 410
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.02,
|
| 276 |
+
"learning_rate": 2.6975476839237056e-05,
|
| 277 |
+
"loss": 0.085,
|
| 278 |
+
"step": 420
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 2.03,
|
| 282 |
+
"learning_rate": 2.6294277929155313e-05,
|
| 283 |
+
"loss": 1.2553,
|
| 284 |
+
"step": 430
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 2.04,
|
| 288 |
+
"learning_rate": 2.5613079019073572e-05,
|
| 289 |
+
"loss": 0.5826,
|
| 290 |
+
"step": 440
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 2.06,
|
| 294 |
+
"learning_rate": 2.4931880108991825e-05,
|
| 295 |
+
"loss": 1.4869,
|
| 296 |
+
"step": 450
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.07,
|
| 300 |
+
"learning_rate": 2.425068119891008e-05,
|
| 301 |
+
"loss": 0.5711,
|
| 302 |
+
"step": 460
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 2.08,
|
| 306 |
+
"learning_rate": 2.356948228882834e-05,
|
| 307 |
+
"loss": 0.0089,
|
| 308 |
+
"step": 470
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 2.09,
|
| 312 |
+
"learning_rate": 2.2888283378746594e-05,
|
| 313 |
+
"loss": 0.5027,
|
| 314 |
+
"step": 480
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.11,
|
| 318 |
+
"learning_rate": 2.220708446866485e-05,
|
| 319 |
+
"loss": 0.4668,
|
| 320 |
+
"step": 490
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 2.12,
|
| 324 |
+
"learning_rate": 2.1525885558583106e-05,
|
| 325 |
+
"loss": 0.4391,
|
| 326 |
+
"step": 500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 2.13,
|
| 330 |
+
"learning_rate": 2.0844686648501363e-05,
|
| 331 |
+
"loss": 0.8851,
|
| 332 |
+
"step": 510
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 2.14,
|
| 336 |
+
"learning_rate": 2.016348773841962e-05,
|
| 337 |
+
"loss": 0.3284,
|
| 338 |
+
"step": 520
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 2.15,
|
| 342 |
+
"learning_rate": 1.9482288828337875e-05,
|
| 343 |
+
"loss": 0.5255,
|
| 344 |
+
"step": 530
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 2.17,
|
| 348 |
+
"learning_rate": 1.880108991825613e-05,
|
| 349 |
+
"loss": 1.0862,
|
| 350 |
+
"step": 540
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 2.18,
|
| 354 |
+
"learning_rate": 1.8119891008174387e-05,
|
| 355 |
+
"loss": 1.5676,
|
| 356 |
+
"step": 550
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.19,
|
| 360 |
+
"learning_rate": 1.7438692098092644e-05,
|
| 361 |
+
"loss": 1.2709,
|
| 362 |
+
"step": 560
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.2,
|
| 366 |
+
"learning_rate": 1.67574931880109e-05,
|
| 367 |
+
"loss": 0.4326,
|
| 368 |
+
"step": 570
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 2.22,
|
| 372 |
+
"learning_rate": 1.6076294277929156e-05,
|
| 373 |
+
"loss": 0.0092,
|
| 374 |
+
"step": 580
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 2.23,
|
| 378 |
+
"learning_rate": 1.5395095367847412e-05,
|
| 379 |
+
"loss": 0.7223,
|
| 380 |
+
"step": 590
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.24,
|
| 384 |
+
"learning_rate": 1.4713896457765669e-05,
|
| 385 |
+
"loss": 0.4094,
|
| 386 |
+
"step": 600
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 2.25,
|
| 390 |
+
"eval_accuracy": 1.0,
|
| 391 |
+
"eval_loss": 0.03134175017476082,
|
| 392 |
+
"eval_runtime": 0.4687,
|
| 393 |
+
"eval_samples_per_second": 4.267,
|
| 394 |
+
"eval_steps_per_second": 4.267,
|
| 395 |
+
"step": 606
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 3.0,
|
| 399 |
+
"learning_rate": 1.4032697547683923e-05,
|
| 400 |
+
"loss": 0.2763,
|
| 401 |
+
"step": 610
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 3.02,
|
| 405 |
+
"learning_rate": 1.3351498637602181e-05,
|
| 406 |
+
"loss": 0.8591,
|
| 407 |
+
"step": 620
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 3.03,
|
| 411 |
+
"learning_rate": 1.2670299727520437e-05,
|
| 412 |
+
"loss": 0.0039,
|
| 413 |
+
"step": 630
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 3.04,
|
| 417 |
+
"learning_rate": 1.1989100817438692e-05,
|
| 418 |
+
"loss": 0.0026,
|
| 419 |
+
"step": 640
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 3.05,
|
| 423 |
+
"learning_rate": 1.130790190735695e-05,
|
| 424 |
+
"loss": 0.0036,
|
| 425 |
+
"step": 650
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 3.07,
|
| 429 |
+
"learning_rate": 1.0626702997275204e-05,
|
| 430 |
+
"loss": 0.0025,
|
| 431 |
+
"step": 660
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 3.08,
|
| 435 |
+
"learning_rate": 9.94550408719346e-06,
|
| 436 |
+
"loss": 1.1266,
|
| 437 |
+
"step": 670
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 3.09,
|
| 441 |
+
"learning_rate": 9.264305177111717e-06,
|
| 442 |
+
"loss": 0.497,
|
| 443 |
+
"step": 680
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 3.1,
|
| 447 |
+
"learning_rate": 8.583106267029973e-06,
|
| 448 |
+
"loss": 0.0028,
|
| 449 |
+
"step": 690
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.12,
|
| 453 |
+
"learning_rate": 7.90190735694823e-06,
|
| 454 |
+
"loss": 0.901,
|
| 455 |
+
"step": 700
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 3.13,
|
| 459 |
+
"learning_rate": 7.220708446866485e-06,
|
| 460 |
+
"loss": 0.2089,
|
| 461 |
+
"step": 710
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 3.14,
|
| 465 |
+
"learning_rate": 6.539509536784741e-06,
|
| 466 |
+
"loss": 1.0508,
|
| 467 |
+
"step": 720
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 3.15,
|
| 471 |
+
"learning_rate": 5.858310626702997e-06,
|
| 472 |
+
"loss": 0.004,
|
| 473 |
+
"step": 730
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 3.16,
|
| 477 |
+
"learning_rate": 5.1771117166212534e-06,
|
| 478 |
+
"loss": 0.4922,
|
| 479 |
+
"step": 740
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 3.18,
|
| 483 |
+
"learning_rate": 4.49591280653951e-06,
|
| 484 |
+
"loss": 0.3897,
|
| 485 |
+
"step": 750
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 3.19,
|
| 489 |
+
"learning_rate": 3.814713896457766e-06,
|
| 490 |
+
"loss": 0.003,
|
| 491 |
+
"step": 760
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 3.2,
|
| 495 |
+
"learning_rate": 3.1335149863760217e-06,
|
| 496 |
+
"loss": 0.0027,
|
| 497 |
+
"step": 770
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 3.21,
|
| 501 |
+
"learning_rate": 2.452316076294278e-06,
|
| 502 |
+
"loss": 0.0159,
|
| 503 |
+
"step": 780
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 3.23,
|
| 507 |
+
"learning_rate": 1.7711171662125342e-06,
|
| 508 |
+
"loss": 0.429,
|
| 509 |
+
"step": 790
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 3.24,
|
| 513 |
+
"learning_rate": 1.0899182561307902e-06,
|
| 514 |
+
"loss": 0.2331,
|
| 515 |
+
"step": 800
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 3.25,
|
| 519 |
+
"eval_accuracy": 0.5,
|
| 520 |
+
"eval_loss": 2.790487051010132,
|
| 521 |
+
"eval_runtime": 0.4807,
|
| 522 |
+
"eval_samples_per_second": 4.161,
|
| 523 |
+
"eval_steps_per_second": 4.161,
|
| 524 |
+
"step": 808
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 4.0,
|
| 528 |
+
"learning_rate": 4.087193460490463e-07,
|
| 529 |
+
"loss": 0.0021,
|
| 530 |
+
"step": 810
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 4.01,
|
| 534 |
+
"eval_accuracy": 0.5,
|
| 535 |
+
"eval_loss": 2.7916269302368164,
|
| 536 |
+
"eval_runtime": 0.9748,
|
| 537 |
+
"eval_samples_per_second": 2.052,
|
| 538 |
+
"eval_steps_per_second": 2.052,
|
| 539 |
+
"step": 816
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 4.01,
|
| 543 |
+
"step": 816,
|
| 544 |
+
"total_flos": 1.0167887050089431e+18,
|
| 545 |
+
"train_loss": 0.734293045413306,
|
| 546 |
+
"train_runtime": 501.5751,
|
| 547 |
+
"train_samples_per_second": 1.627,
|
| 548 |
+
"train_steps_per_second": 1.627
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 4.01,
|
| 552 |
+
"eval_accuracy": 1.0,
|
| 553 |
+
"eval_loss": 0.008221009746193886,
|
| 554 |
+
"eval_runtime": 0.5089,
|
| 555 |
+
"eval_samples_per_second": 3.93,
|
| 556 |
+
"eval_steps_per_second": 3.93,
|
| 557 |
+
"step": 816
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 4.01,
|
| 561 |
+
"eval_accuracy": 1.0,
|
| 562 |
+
"eval_loss": 0.008221009746193886,
|
| 563 |
+
"eval_runtime": 0.5785,
|
| 564 |
+
"eval_samples_per_second": 3.457,
|
| 565 |
+
"eval_steps_per_second": 3.457,
|
| 566 |
+
"step": 816
|
| 567 |
+
}
|
| 568 |
+
],
|
| 569 |
+
"max_steps": 816,
|
| 570 |
+
"num_train_epochs": 9223372036854775807,
|
| 571 |
+
"total_flos": 1.0167887050089431e+18,
|
| 572 |
+
"trial_name": null,
|
| 573 |
+
"trial_params": null
|
| 574 |
+
}
|