Update trainer_state.json
Browse files- trainer_state.json +1216 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1216 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.88762098828324,
|
| 3 |
+
"best_model_checkpoint": "test-glue/checkpoint-98176",
|
| 4 |
+
"epoch": 4.0,
|
| 5 |
+
"global_step": 98176,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.020371577574967405,
|
| 12 |
+
"learning_rate": 1.9898142112125166e-05,
|
| 13 |
+
"loss": 0.659133544921875,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.04074315514993481,
|
| 18 |
+
"learning_rate": 1.9796284224250328e-05,
|
| 19 |
+
"loss": 0.5110537109375,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.06111473272490222,
|
| 24 |
+
"learning_rate": 1.969442633637549e-05,
|
| 25 |
+
"loss": 0.485730224609375,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.08148631029986962,
|
| 30 |
+
"learning_rate": 1.9592568448500654e-05,
|
| 31 |
+
"loss": 0.48240771484375,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.10185788787483703,
|
| 36 |
+
"learning_rate": 1.9490710560625816e-05,
|
| 37 |
+
"loss": 0.46434228515625,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.12222946544980444,
|
| 42 |
+
"learning_rate": 1.938885267275098e-05,
|
| 43 |
+
"loss": 0.449291259765625,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.14260104302477183,
|
| 48 |
+
"learning_rate": 1.9286994784876142e-05,
|
| 49 |
+
"loss": 0.4585693359375,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.16297262059973924,
|
| 54 |
+
"learning_rate": 1.9185136897001307e-05,
|
| 55 |
+
"loss": 0.44105908203125,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.18334419817470665,
|
| 60 |
+
"learning_rate": 1.9083279009126468e-05,
|
| 61 |
+
"loss": 0.444558837890625,
|
| 62 |
+
"step": 4500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.20371577574967406,
|
| 66 |
+
"learning_rate": 1.898142112125163e-05,
|
| 67 |
+
"loss": 0.45100146484375,
|
| 68 |
+
"step": 5000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.22408735332464147,
|
| 72 |
+
"learning_rate": 1.8879563233376795e-05,
|
| 73 |
+
"loss": 0.448494140625,
|
| 74 |
+
"step": 5500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.24445893089960888,
|
| 78 |
+
"learning_rate": 1.8777705345501956e-05,
|
| 79 |
+
"loss": 0.4301044921875,
|
| 80 |
+
"step": 6000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.2648305084745763,
|
| 84 |
+
"learning_rate": 1.867584745762712e-05,
|
| 85 |
+
"loss": 0.42643359375,
|
| 86 |
+
"step": 6500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.28520208604954367,
|
| 90 |
+
"learning_rate": 1.8573989569752282e-05,
|
| 91 |
+
"loss": 0.41756689453125,
|
| 92 |
+
"step": 7000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.3055736636245111,
|
| 96 |
+
"learning_rate": 1.8472131681877447e-05,
|
| 97 |
+
"loss": 0.4308828125,
|
| 98 |
+
"step": 7500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.3259452411994785,
|
| 102 |
+
"learning_rate": 1.837027379400261e-05,
|
| 103 |
+
"loss": 0.41683154296875,
|
| 104 |
+
"step": 8000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.34631681877444587,
|
| 108 |
+
"learning_rate": 1.8268415906127773e-05,
|
| 109 |
+
"loss": 0.413720703125,
|
| 110 |
+
"step": 8500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.3666883963494133,
|
| 114 |
+
"learning_rate": 1.8166558018252935e-05,
|
| 115 |
+
"loss": 0.425869140625,
|
| 116 |
+
"step": 9000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.3870599739243807,
|
| 120 |
+
"learning_rate": 1.80647001303781e-05,
|
| 121 |
+
"loss": 0.42243359375,
|
| 122 |
+
"step": 9500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.4074315514993481,
|
| 126 |
+
"learning_rate": 1.796284224250326e-05,
|
| 127 |
+
"loss": 0.415326171875,
|
| 128 |
+
"step": 10000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.4278031290743155,
|
| 132 |
+
"learning_rate": 1.7860984354628423e-05,
|
| 133 |
+
"loss": 0.4072890625,
|
| 134 |
+
"step": 10500
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.44817470664928294,
|
| 138 |
+
"learning_rate": 1.7759126466753587e-05,
|
| 139 |
+
"loss": 0.4062841796875,
|
| 140 |
+
"step": 11000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.4685462842242503,
|
| 144 |
+
"learning_rate": 1.765726857887875e-05,
|
| 145 |
+
"loss": 0.3890205078125,
|
| 146 |
+
"step": 11500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.48891786179921776,
|
| 150 |
+
"learning_rate": 1.7555410691003914e-05,
|
| 151 |
+
"loss": 0.3969296875,
|
| 152 |
+
"step": 12000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.5092894393741851,
|
| 156 |
+
"learning_rate": 1.7453552803129075e-05,
|
| 157 |
+
"loss": 0.3952880859375,
|
| 158 |
+
"step": 12500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.5296610169491526,
|
| 162 |
+
"learning_rate": 1.7351694915254237e-05,
|
| 163 |
+
"loss": 0.41177734375,
|
| 164 |
+
"step": 13000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.5500325945241199,
|
| 168 |
+
"learning_rate": 1.72498370273794e-05,
|
| 169 |
+
"loss": 0.3807109375,
|
| 170 |
+
"step": 13500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.5704041720990873,
|
| 174 |
+
"learning_rate": 1.7147979139504566e-05,
|
| 175 |
+
"loss": 0.403853515625,
|
| 176 |
+
"step": 14000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.5907757496740548,
|
| 180 |
+
"learning_rate": 1.7046121251629728e-05,
|
| 181 |
+
"loss": 0.411607421875,
|
| 182 |
+
"step": 14500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.6111473272490222,
|
| 186 |
+
"learning_rate": 1.6944263363754893e-05,
|
| 187 |
+
"loss": 0.3996162109375,
|
| 188 |
+
"step": 15000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.6315189048239895,
|
| 192 |
+
"learning_rate": 1.6842405475880054e-05,
|
| 193 |
+
"loss": 0.3948984375,
|
| 194 |
+
"step": 15500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.651890482398957,
|
| 198 |
+
"learning_rate": 1.6740547588005215e-05,
|
| 199 |
+
"loss": 0.389115234375,
|
| 200 |
+
"step": 16000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.6722620599739244,
|
| 204 |
+
"learning_rate": 1.663868970013038e-05,
|
| 205 |
+
"loss": 0.403466796875,
|
| 206 |
+
"step": 16500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.6926336375488917,
|
| 210 |
+
"learning_rate": 1.6536831812255542e-05,
|
| 211 |
+
"loss": 0.4059140625,
|
| 212 |
+
"step": 17000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.7130052151238592,
|
| 216 |
+
"learning_rate": 1.6434973924380707e-05,
|
| 217 |
+
"loss": 0.3873095703125,
|
| 218 |
+
"step": 17500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.7333767926988266,
|
| 222 |
+
"learning_rate": 1.6333116036505868e-05,
|
| 223 |
+
"loss": 0.3887783203125,
|
| 224 |
+
"step": 18000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.753748370273794,
|
| 228 |
+
"learning_rate": 1.623125814863103e-05,
|
| 229 |
+
"loss": 0.395041015625,
|
| 230 |
+
"step": 18500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.7741199478487614,
|
| 234 |
+
"learning_rate": 1.6129400260756194e-05,
|
| 235 |
+
"loss": 0.39855078125,
|
| 236 |
+
"step": 19000
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.7944915254237288,
|
| 240 |
+
"learning_rate": 1.6027542372881356e-05,
|
| 241 |
+
"loss": 0.390134765625,
|
| 242 |
+
"step": 19500
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.8148631029986962,
|
| 246 |
+
"learning_rate": 1.592568448500652e-05,
|
| 247 |
+
"loss": 0.3717578125,
|
| 248 |
+
"step": 20000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.8352346805736637,
|
| 252 |
+
"learning_rate": 1.5823826597131685e-05,
|
| 253 |
+
"loss": 0.387556640625,
|
| 254 |
+
"step": 20500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.855606258148631,
|
| 258 |
+
"learning_rate": 1.5721968709256844e-05,
|
| 259 |
+
"loss": 0.389892578125,
|
| 260 |
+
"step": 21000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.8759778357235984,
|
| 264 |
+
"learning_rate": 1.562011082138201e-05,
|
| 265 |
+
"loss": 0.38671875,
|
| 266 |
+
"step": 21500
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.8963494132985659,
|
| 270 |
+
"learning_rate": 1.5518252933507173e-05,
|
| 271 |
+
"loss": 0.36665234375,
|
| 272 |
+
"step": 22000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.9167209908735332,
|
| 276 |
+
"learning_rate": 1.5416395045632335e-05,
|
| 277 |
+
"loss": 0.3940859375,
|
| 278 |
+
"step": 22500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.9370925684485006,
|
| 282 |
+
"learning_rate": 1.53145371577575e-05,
|
| 283 |
+
"loss": 0.382552734375,
|
| 284 |
+
"step": 23000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.9574641460234681,
|
| 288 |
+
"learning_rate": 1.5212679269882663e-05,
|
| 289 |
+
"loss": 0.38312109375,
|
| 290 |
+
"step": 23500
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.9778357235984355,
|
| 294 |
+
"learning_rate": 1.5110821382007822e-05,
|
| 295 |
+
"loss": 0.3804765625,
|
| 296 |
+
"step": 24000
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.9982073011734028,
|
| 300 |
+
"learning_rate": 1.5008963494132987e-05,
|
| 301 |
+
"loss": 0.383294921875,
|
| 302 |
+
"step": 24500
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.0,
|
| 306 |
+
"eval_accuracy": 0.8807947019867549,
|
| 307 |
+
"eval_loss": 0.3417690396308899,
|
| 308 |
+
"step": 24544
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 1.0185788787483703,
|
| 312 |
+
"learning_rate": 1.490710560625815e-05,
|
| 313 |
+
"loss": 0.293595703125,
|
| 314 |
+
"step": 25000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 1.0389504563233376,
|
| 318 |
+
"learning_rate": 1.4805247718383314e-05,
|
| 319 |
+
"loss": 0.29225,
|
| 320 |
+
"step": 25500
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 1.0593220338983051,
|
| 324 |
+
"learning_rate": 1.4703389830508477e-05,
|
| 325 |
+
"loss": 0.294130859375,
|
| 326 |
+
"step": 26000
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 1.0796936114732725,
|
| 330 |
+
"learning_rate": 1.4601531942633638e-05,
|
| 331 |
+
"loss": 0.28494921875,
|
| 332 |
+
"step": 26500
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 1.1000651890482398,
|
| 336 |
+
"learning_rate": 1.4499674054758801e-05,
|
| 337 |
+
"loss": 0.28199609375,
|
| 338 |
+
"step": 27000
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 1.1204367666232073,
|
| 342 |
+
"learning_rate": 1.4397816166883964e-05,
|
| 343 |
+
"loss": 0.28830078125,
|
| 344 |
+
"step": 27500
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 1.1408083441981747,
|
| 348 |
+
"learning_rate": 1.4295958279009128e-05,
|
| 349 |
+
"loss": 0.283265625,
|
| 350 |
+
"step": 28000
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 1.161179921773142,
|
| 354 |
+
"learning_rate": 1.419410039113429e-05,
|
| 355 |
+
"loss": 0.284984375,
|
| 356 |
+
"step": 28500
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 1.1815514993481095,
|
| 360 |
+
"learning_rate": 1.4092242503259454e-05,
|
| 361 |
+
"loss": 0.2741875,
|
| 362 |
+
"step": 29000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 1.2019230769230769,
|
| 366 |
+
"learning_rate": 1.3990384615384615e-05,
|
| 367 |
+
"loss": 0.293828125,
|
| 368 |
+
"step": 29500
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 1.2222946544980444,
|
| 372 |
+
"learning_rate": 1.3888526727509778e-05,
|
| 373 |
+
"loss": 0.291970703125,
|
| 374 |
+
"step": 30000
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 1.2426662320730117,
|
| 378 |
+
"learning_rate": 1.3786668839634942e-05,
|
| 379 |
+
"loss": 0.2823515625,
|
| 380 |
+
"step": 30500
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 1.263037809647979,
|
| 384 |
+
"learning_rate": 1.3684810951760106e-05,
|
| 385 |
+
"loss": 0.2898671875,
|
| 386 |
+
"step": 31000
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 1.2834093872229466,
|
| 390 |
+
"learning_rate": 1.358295306388527e-05,
|
| 391 |
+
"loss": 0.291859375,
|
| 392 |
+
"step": 31500
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 1.303780964797914,
|
| 396 |
+
"learning_rate": 1.3481095176010431e-05,
|
| 397 |
+
"loss": 0.292462890625,
|
| 398 |
+
"step": 32000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 1.3241525423728815,
|
| 402 |
+
"learning_rate": 1.3379237288135594e-05,
|
| 403 |
+
"loss": 0.29555859375,
|
| 404 |
+
"step": 32500
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 1.3445241199478488,
|
| 408 |
+
"learning_rate": 1.3277379400260757e-05,
|
| 409 |
+
"loss": 0.288212890625,
|
| 410 |
+
"step": 33000
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 1.3648956975228161,
|
| 414 |
+
"learning_rate": 1.317552151238592e-05,
|
| 415 |
+
"loss": 0.294849609375,
|
| 416 |
+
"step": 33500
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 1.3852672750977835,
|
| 420 |
+
"learning_rate": 1.3073663624511084e-05,
|
| 421 |
+
"loss": 0.289576171875,
|
| 422 |
+
"step": 34000
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 1.405638852672751,
|
| 426 |
+
"learning_rate": 1.2971805736636247e-05,
|
| 427 |
+
"loss": 0.289185546875,
|
| 428 |
+
"step": 34500
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 1.4260104302477183,
|
| 432 |
+
"learning_rate": 1.2869947848761408e-05,
|
| 433 |
+
"loss": 0.283068359375,
|
| 434 |
+
"step": 35000
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 1.4463820078226859,
|
| 438 |
+
"learning_rate": 1.2768089960886571e-05,
|
| 439 |
+
"loss": 0.28625390625,
|
| 440 |
+
"step": 35500
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 1.4667535853976532,
|
| 444 |
+
"learning_rate": 1.2666232073011735e-05,
|
| 445 |
+
"loss": 0.282048828125,
|
| 446 |
+
"step": 36000
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 1.4871251629726205,
|
| 450 |
+
"learning_rate": 1.2564374185136898e-05,
|
| 451 |
+
"loss": 0.28896484375,
|
| 452 |
+
"step": 36500
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.5074967405475879,
|
| 456 |
+
"learning_rate": 1.2462516297262063e-05,
|
| 457 |
+
"loss": 0.276443359375,
|
| 458 |
+
"step": 37000
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.5278683181225554,
|
| 462 |
+
"learning_rate": 1.2360658409387226e-05,
|
| 463 |
+
"loss": 0.310044921875,
|
| 464 |
+
"step": 37500
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 1.548239895697523,
|
| 468 |
+
"learning_rate": 1.2258800521512385e-05,
|
| 469 |
+
"loss": 0.285341796875,
|
| 470 |
+
"step": 38000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.5686114732724903,
|
| 474 |
+
"learning_rate": 1.215694263363755e-05,
|
| 475 |
+
"loss": 0.282486328125,
|
| 476 |
+
"step": 38500
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 1.5889830508474576,
|
| 480 |
+
"learning_rate": 1.2055084745762713e-05,
|
| 481 |
+
"loss": 0.295353515625,
|
| 482 |
+
"step": 39000
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 1.609354628422425,
|
| 486 |
+
"learning_rate": 1.1953226857887877e-05,
|
| 487 |
+
"loss": 0.284724609375,
|
| 488 |
+
"step": 39500
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 1.6297262059973925,
|
| 492 |
+
"learning_rate": 1.185136897001304e-05,
|
| 493 |
+
"loss": 0.286890625,
|
| 494 |
+
"step": 40000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.6500977835723598,
|
| 498 |
+
"learning_rate": 1.1749511082138201e-05,
|
| 499 |
+
"loss": 0.278595703125,
|
| 500 |
+
"step": 40500
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.6704693611473274,
|
| 504 |
+
"learning_rate": 1.1647653194263364e-05,
|
| 505 |
+
"loss": 0.30148046875,
|
| 506 |
+
"step": 41000
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.6908409387222947,
|
| 510 |
+
"learning_rate": 1.1545795306388527e-05,
|
| 511 |
+
"loss": 0.274353515625,
|
| 512 |
+
"step": 41500
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 1.711212516297262,
|
| 516 |
+
"learning_rate": 1.144393741851369e-05,
|
| 517 |
+
"loss": 0.29472265625,
|
| 518 |
+
"step": 42000
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 1.7315840938722293,
|
| 522 |
+
"learning_rate": 1.1342079530638854e-05,
|
| 523 |
+
"loss": 0.2853359375,
|
| 524 |
+
"step": 42500
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 1.7519556714471969,
|
| 528 |
+
"learning_rate": 1.1240221642764017e-05,
|
| 529 |
+
"loss": 0.286759765625,
|
| 530 |
+
"step": 43000
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 1.7723272490221644,
|
| 534 |
+
"learning_rate": 1.1138363754889178e-05,
|
| 535 |
+
"loss": 0.28123046875,
|
| 536 |
+
"step": 43500
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.7926988265971318,
|
| 540 |
+
"learning_rate": 1.1036505867014341e-05,
|
| 541 |
+
"loss": 0.2833515625,
|
| 542 |
+
"step": 44000
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.813070404172099,
|
| 546 |
+
"learning_rate": 1.0934647979139506e-05,
|
| 547 |
+
"loss": 0.281861328125,
|
| 548 |
+
"step": 44500
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 1.8334419817470664,
|
| 552 |
+
"learning_rate": 1.083279009126467e-05,
|
| 553 |
+
"loss": 0.2866171875,
|
| 554 |
+
"step": 45000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 1.8538135593220337,
|
| 558 |
+
"learning_rate": 1.0730932203389833e-05,
|
| 559 |
+
"loss": 0.28151953125,
|
| 560 |
+
"step": 45500
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 1.8741851368970013,
|
| 564 |
+
"learning_rate": 1.0629074315514994e-05,
|
| 565 |
+
"loss": 0.285787109375,
|
| 566 |
+
"step": 46000
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 1.8945567144719688,
|
| 570 |
+
"learning_rate": 1.0527216427640157e-05,
|
| 571 |
+
"loss": 0.28494921875,
|
| 572 |
+
"step": 46500
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 1.9149282920469362,
|
| 576 |
+
"learning_rate": 1.042535853976532e-05,
|
| 577 |
+
"loss": 0.281390625,
|
| 578 |
+
"step": 47000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.9352998696219035,
|
| 582 |
+
"learning_rate": 1.0323500651890483e-05,
|
| 583 |
+
"loss": 0.2822734375,
|
| 584 |
+
"step": 47500
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 1.9556714471968708,
|
| 588 |
+
"learning_rate": 1.0221642764015647e-05,
|
| 589 |
+
"loss": 0.2885546875,
|
| 590 |
+
"step": 48000
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 1.9760430247718384,
|
| 594 |
+
"learning_rate": 1.011978487614081e-05,
|
| 595 |
+
"loss": 0.2759296875,
|
| 596 |
+
"step": 48500
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 1.996414602346806,
|
| 600 |
+
"learning_rate": 1.0017926988265971e-05,
|
| 601 |
+
"loss": 0.27166796875,
|
| 602 |
+
"step": 49000
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 2.0,
|
| 606 |
+
"eval_accuracy": 0.8865002547121752,
|
| 607 |
+
"eval_loss": 0.34813082218170166,
|
| 608 |
+
"step": 49088
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 2.0167861799217732,
|
| 612 |
+
"learning_rate": 9.916069100391134e-06,
|
| 613 |
+
"loss": 0.2067265625,
|
| 614 |
+
"step": 49500
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 2.0371577574967406,
|
| 618 |
+
"learning_rate": 9.814211212516298e-06,
|
| 619 |
+
"loss": 0.18627734375,
|
| 620 |
+
"step": 50000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 2.057529335071708,
|
| 624 |
+
"learning_rate": 9.71235332464146e-06,
|
| 625 |
+
"loss": 0.20591796875,
|
| 626 |
+
"step": 50500
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 2.077900912646675,
|
| 630 |
+
"learning_rate": 9.610495436766624e-06,
|
| 631 |
+
"loss": 0.207765625,
|
| 632 |
+
"step": 51000
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 2.098272490221643,
|
| 636 |
+
"learning_rate": 9.508637548891787e-06,
|
| 637 |
+
"loss": 0.200046875,
|
| 638 |
+
"step": 51500
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 2.1186440677966103,
|
| 642 |
+
"learning_rate": 9.40677966101695e-06,
|
| 643 |
+
"loss": 0.20955078125,
|
| 644 |
+
"step": 52000
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 2.1390156453715776,
|
| 648 |
+
"learning_rate": 9.304921773142113e-06,
|
| 649 |
+
"loss": 0.20408203125,
|
| 650 |
+
"step": 52500
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 2.159387222946545,
|
| 654 |
+
"learning_rate": 9.203063885267276e-06,
|
| 655 |
+
"loss": 0.1967578125,
|
| 656 |
+
"step": 53000
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 2.1797588005215123,
|
| 660 |
+
"learning_rate": 9.101205997392438e-06,
|
| 661 |
+
"loss": 0.2086171875,
|
| 662 |
+
"step": 53500
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 2.2001303780964796,
|
| 666 |
+
"learning_rate": 8.999348109517601e-06,
|
| 667 |
+
"loss": 0.19387890625,
|
| 668 |
+
"step": 54000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 2.2205019556714474,
|
| 672 |
+
"learning_rate": 8.897490221642766e-06,
|
| 673 |
+
"loss": 0.2149140625,
|
| 674 |
+
"step": 54500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 2.2408735332464147,
|
| 678 |
+
"learning_rate": 8.795632333767927e-06,
|
| 679 |
+
"loss": 0.20816015625,
|
| 680 |
+
"step": 55000
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 2.261245110821382,
|
| 684 |
+
"learning_rate": 8.69377444589309e-06,
|
| 685 |
+
"loss": 0.1981796875,
|
| 686 |
+
"step": 55500
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 2.2816166883963493,
|
| 690 |
+
"learning_rate": 8.591916558018254e-06,
|
| 691 |
+
"loss": 0.20524609375,
|
| 692 |
+
"step": 56000
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 2.3019882659713167,
|
| 696 |
+
"learning_rate": 8.490058670143417e-06,
|
| 697 |
+
"loss": 0.20031640625,
|
| 698 |
+
"step": 56500
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 2.322359843546284,
|
| 702 |
+
"learning_rate": 8.38820078226858e-06,
|
| 703 |
+
"loss": 0.2170234375,
|
| 704 |
+
"step": 57000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 2.3427314211212518,
|
| 708 |
+
"learning_rate": 8.286342894393743e-06,
|
| 709 |
+
"loss": 0.202625,
|
| 710 |
+
"step": 57500
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 2.363102998696219,
|
| 714 |
+
"learning_rate": 8.184485006518904e-06,
|
| 715 |
+
"loss": 0.21112890625,
|
| 716 |
+
"step": 58000
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 2.3834745762711864,
|
| 720 |
+
"learning_rate": 8.08262711864407e-06,
|
| 721 |
+
"loss": 0.19984375,
|
| 722 |
+
"step": 58500
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 2.4038461538461537,
|
| 726 |
+
"learning_rate": 7.980769230769232e-06,
|
| 727 |
+
"loss": 0.2016015625,
|
| 728 |
+
"step": 59000
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 2.424217731421121,
|
| 732 |
+
"learning_rate": 7.878911342894394e-06,
|
| 733 |
+
"loss": 0.20296875,
|
| 734 |
+
"step": 59500
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 2.444589308996089,
|
| 738 |
+
"learning_rate": 7.777053455019557e-06,
|
| 739 |
+
"loss": 0.20494140625,
|
| 740 |
+
"step": 60000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 2.464960886571056,
|
| 744 |
+
"learning_rate": 7.67519556714472e-06,
|
| 745 |
+
"loss": 0.19887890625,
|
| 746 |
+
"step": 60500
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 2.4853324641460235,
|
| 750 |
+
"learning_rate": 7.573337679269883e-06,
|
| 751 |
+
"loss": 0.2058984375,
|
| 752 |
+
"step": 61000
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 2.505704041720991,
|
| 756 |
+
"learning_rate": 7.4714797913950464e-06,
|
| 757 |
+
"loss": 0.2057265625,
|
| 758 |
+
"step": 61500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 2.526075619295958,
|
| 762 |
+
"learning_rate": 7.369621903520209e-06,
|
| 763 |
+
"loss": 0.1991484375,
|
| 764 |
+
"step": 62000
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 2.5464471968709255,
|
| 768 |
+
"learning_rate": 7.267764015645372e-06,
|
| 769 |
+
"loss": 0.2016328125,
|
| 770 |
+
"step": 62500
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 2.5668187744458932,
|
| 774 |
+
"learning_rate": 7.165906127770536e-06,
|
| 775 |
+
"loss": 0.2006640625,
|
| 776 |
+
"step": 63000
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 2.5871903520208606,
|
| 780 |
+
"learning_rate": 7.064048239895698e-06,
|
| 781 |
+
"loss": 0.20136328125,
|
| 782 |
+
"step": 63500
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 2.607561929595828,
|
| 786 |
+
"learning_rate": 6.962190352020861e-06,
|
| 787 |
+
"loss": 0.210953125,
|
| 788 |
+
"step": 64000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 2.627933507170795,
|
| 792 |
+
"learning_rate": 6.8603324641460245e-06,
|
| 793 |
+
"loss": 0.20728515625,
|
| 794 |
+
"step": 64500
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 2.648305084745763,
|
| 798 |
+
"learning_rate": 6.758474576271187e-06,
|
| 799 |
+
"loss": 0.19809765625,
|
| 800 |
+
"step": 65000
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 2.6686766623207303,
|
| 804 |
+
"learning_rate": 6.65661668839635e-06,
|
| 805 |
+
"loss": 0.20791796875,
|
| 806 |
+
"step": 65500
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 2.6890482398956976,
|
| 810 |
+
"learning_rate": 6.554758800521513e-06,
|
| 811 |
+
"loss": 0.1968359375,
|
| 812 |
+
"step": 66000
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 2.709419817470665,
|
| 816 |
+
"learning_rate": 6.452900912646675e-06,
|
| 817 |
+
"loss": 0.19569140625,
|
| 818 |
+
"step": 66500
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 2.7297913950456323,
|
| 822 |
+
"learning_rate": 6.351043024771839e-06,
|
| 823 |
+
"loss": 0.21024609375,
|
| 824 |
+
"step": 67000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 2.7501629726205996,
|
| 828 |
+
"learning_rate": 6.249185136897002e-06,
|
| 829 |
+
"loss": 0.20680859375,
|
| 830 |
+
"step": 67500
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 2.770534550195567,
|
| 834 |
+
"learning_rate": 6.147327249022165e-06,
|
| 835 |
+
"loss": 0.206203125,
|
| 836 |
+
"step": 68000
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 2.7909061277705347,
|
| 840 |
+
"learning_rate": 6.045469361147328e-06,
|
| 841 |
+
"loss": 0.1981015625,
|
| 842 |
+
"step": 68500
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 2.811277705345502,
|
| 846 |
+
"learning_rate": 5.94361147327249e-06,
|
| 847 |
+
"loss": 0.20408984375,
|
| 848 |
+
"step": 69000
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 2.8316492829204694,
|
| 852 |
+
"learning_rate": 5.841753585397653e-06,
|
| 853 |
+
"loss": 0.20057421875,
|
| 854 |
+
"step": 69500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 2.8520208604954367,
|
| 858 |
+
"learning_rate": 5.739895697522817e-06,
|
| 859 |
+
"loss": 0.19898046875,
|
| 860 |
+
"step": 70000
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 2.872392438070404,
|
| 864 |
+
"learning_rate": 5.63803780964798e-06,
|
| 865 |
+
"loss": 0.19495703125,
|
| 866 |
+
"step": 70500
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 2.8927640156453718,
|
| 870 |
+
"learning_rate": 5.536179921773143e-06,
|
| 871 |
+
"loss": 0.1974296875,
|
| 872 |
+
"step": 71000
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 2.913135593220339,
|
| 876 |
+
"learning_rate": 5.434322033898306e-06,
|
| 877 |
+
"loss": 0.20246484375,
|
| 878 |
+
"step": 71500
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 2.9335071707953064,
|
| 882 |
+
"learning_rate": 5.332464146023468e-06,
|
| 883 |
+
"loss": 0.19712109375,
|
| 884 |
+
"step": 72000
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 2.9538787483702738,
|
| 888 |
+
"learning_rate": 5.230606258148631e-06,
|
| 889 |
+
"loss": 0.202109375,
|
| 890 |
+
"step": 72500
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 2.974250325945241,
|
| 894 |
+
"learning_rate": 5.128748370273794e-06,
|
| 895 |
+
"loss": 0.2054140625,
|
| 896 |
+
"step": 73000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 2.9946219035202084,
|
| 900 |
+
"learning_rate": 5.026890482398958e-06,
|
| 901 |
+
"loss": 0.202515625,
|
| 902 |
+
"step": 73500
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 3.0,
|
| 906 |
+
"eval_accuracy": 0.884666327050433,
|
| 907 |
+
"eval_loss": 0.4447501301765442,
|
| 908 |
+
"step": 73632
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 3.014993481095176,
|
| 912 |
+
"learning_rate": 4.92503259452412e-06,
|
| 913 |
+
"loss": 0.13929296875,
|
| 914 |
+
"step": 74000
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 3.0353650586701435,
|
| 918 |
+
"learning_rate": 4.823174706649283e-06,
|
| 919 |
+
"loss": 0.14209765625,
|
| 920 |
+
"step": 74500
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 3.055736636245111,
|
| 924 |
+
"learning_rate": 4.721316818774446e-06,
|
| 925 |
+
"loss": 0.14396875,
|
| 926 |
+
"step": 75000
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 3.076108213820078,
|
| 930 |
+
"learning_rate": 4.6194589308996094e-06,
|
| 931 |
+
"loss": 0.13436328125,
|
| 932 |
+
"step": 75500
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 3.0964797913950455,
|
| 936 |
+
"learning_rate": 4.5176010430247726e-06,
|
| 937 |
+
"loss": 0.13274609375,
|
| 938 |
+
"step": 76000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 3.1168513689700132,
|
| 942 |
+
"learning_rate": 4.415743155149935e-06,
|
| 943 |
+
"loss": 0.1425546875,
|
| 944 |
+
"step": 76500
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 3.1372229465449806,
|
| 948 |
+
"learning_rate": 4.313885267275098e-06,
|
| 949 |
+
"loss": 0.1465,
|
| 950 |
+
"step": 77000
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 3.157594524119948,
|
| 954 |
+
"learning_rate": 4.212027379400261e-06,
|
| 955 |
+
"loss": 0.13299609375,
|
| 956 |
+
"step": 77500
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 3.1779661016949152,
|
| 960 |
+
"learning_rate": 4.110169491525424e-06,
|
| 961 |
+
"loss": 0.146578125,
|
| 962 |
+
"step": 78000
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 3.1983376792698825,
|
| 966 |
+
"learning_rate": 4.0083116036505874e-06,
|
| 967 |
+
"loss": 0.1382421875,
|
| 968 |
+
"step": 78500
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 3.21870925684485,
|
| 972 |
+
"learning_rate": 3.90645371577575e-06,
|
| 973 |
+
"loss": 0.14358984375,
|
| 974 |
+
"step": 79000
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 3.2390808344198176,
|
| 978 |
+
"learning_rate": 3.8045958279009133e-06,
|
| 979 |
+
"loss": 0.1374140625,
|
| 980 |
+
"step": 79500
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 3.259452411994785,
|
| 984 |
+
"learning_rate": 3.702737940026076e-06,
|
| 985 |
+
"loss": 0.14011328125,
|
| 986 |
+
"step": 80000
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 3.2798239895697523,
|
| 990 |
+
"learning_rate": 3.6008800521512388e-06,
|
| 991 |
+
"loss": 0.14218359375,
|
| 992 |
+
"step": 80500
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 3.3001955671447196,
|
| 996 |
+
"learning_rate": 3.4990221642764015e-06,
|
| 997 |
+
"loss": 0.14026953125,
|
| 998 |
+
"step": 81000
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 3.320567144719687,
|
| 1002 |
+
"learning_rate": 3.397164276401565e-06,
|
| 1003 |
+
"loss": 0.1326171875,
|
| 1004 |
+
"step": 81500
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 3.3409387222946547,
|
| 1008 |
+
"learning_rate": 3.2953063885267278e-06,
|
| 1009 |
+
"loss": 0.124484375,
|
| 1010 |
+
"step": 82000
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 3.361310299869622,
|
| 1014 |
+
"learning_rate": 3.1934485006518905e-06,
|
| 1015 |
+
"loss": 0.13862890625,
|
| 1016 |
+
"step": 82500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 3.3816818774445894,
|
| 1020 |
+
"learning_rate": 3.091590612777054e-06,
|
| 1021 |
+
"loss": 0.14094140625,
|
| 1022 |
+
"step": 83000
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 3.4020534550195567,
|
| 1026 |
+
"learning_rate": 2.9897327249022168e-06,
|
| 1027 |
+
"loss": 0.13262890625,
|
| 1028 |
+
"step": 83500
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 3.422425032594524,
|
| 1032 |
+
"learning_rate": 2.8878748370273795e-06,
|
| 1033 |
+
"loss": 0.14966015625,
|
| 1034 |
+
"step": 84000
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 3.4427966101694913,
|
| 1038 |
+
"learning_rate": 2.7860169491525422e-06,
|
| 1039 |
+
"loss": 0.136,
|
| 1040 |
+
"step": 84500
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 3.463168187744459,
|
| 1044 |
+
"learning_rate": 2.684159061277706e-06,
|
| 1045 |
+
"loss": 0.15666796875,
|
| 1046 |
+
"step": 85000
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 3.4835397653194264,
|
| 1050 |
+
"learning_rate": 2.5823011734028685e-06,
|
| 1051 |
+
"loss": 0.12829296875,
|
| 1052 |
+
"step": 85500
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 3.5039113428943938,
|
| 1056 |
+
"learning_rate": 2.4804432855280312e-06,
|
| 1057 |
+
"loss": 0.134625,
|
| 1058 |
+
"step": 86000
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 3.524282920469361,
|
| 1062 |
+
"learning_rate": 2.3785853976531944e-06,
|
| 1063 |
+
"loss": 0.1383125,
|
| 1064 |
+
"step": 86500
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 3.5446544980443284,
|
| 1068 |
+
"learning_rate": 2.2767275097783575e-06,
|
| 1069 |
+
"loss": 0.145578125,
|
| 1070 |
+
"step": 87000
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 3.565026075619296,
|
| 1074 |
+
"learning_rate": 2.1748696219035202e-06,
|
| 1075 |
+
"loss": 0.13809765625,
|
| 1076 |
+
"step": 87500
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 3.5853976531942635,
|
| 1080 |
+
"learning_rate": 2.0730117340286834e-06,
|
| 1081 |
+
"loss": 0.131578125,
|
| 1082 |
+
"step": 88000
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 3.605769230769231,
|
| 1086 |
+
"learning_rate": 1.971153846153846e-06,
|
| 1087 |
+
"loss": 0.128625,
|
| 1088 |
+
"step": 88500
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 3.626140808344198,
|
| 1092 |
+
"learning_rate": 1.8692959582790093e-06,
|
| 1093 |
+
"loss": 0.1566328125,
|
| 1094 |
+
"step": 89000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 3.6465123859191655,
|
| 1098 |
+
"learning_rate": 1.7674380704041722e-06,
|
| 1099 |
+
"loss": 0.13289453125,
|
| 1100 |
+
"step": 89500
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 3.666883963494133,
|
| 1104 |
+
"learning_rate": 1.6655801825293353e-06,
|
| 1105 |
+
"loss": 0.13489453125,
|
| 1106 |
+
"step": 90000
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 3.6872555410691,
|
| 1110 |
+
"learning_rate": 1.563722294654498e-06,
|
| 1111 |
+
"loss": 0.1342890625,
|
| 1112 |
+
"step": 90500
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 3.707627118644068,
|
| 1116 |
+
"learning_rate": 1.4618644067796612e-06,
|
| 1117 |
+
"loss": 0.13596875,
|
| 1118 |
+
"step": 91000
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 3.7279986962190352,
|
| 1122 |
+
"learning_rate": 1.3600065189048241e-06,
|
| 1123 |
+
"loss": 0.12205859375,
|
| 1124 |
+
"step": 91500
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 3.7483702737940026,
|
| 1128 |
+
"learning_rate": 1.258148631029987e-06,
|
| 1129 |
+
"loss": 0.136265625,
|
| 1130 |
+
"step": 92000
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 3.76874185136897,
|
| 1134 |
+
"learning_rate": 1.15629074315515e-06,
|
| 1135 |
+
"loss": 0.1251875,
|
| 1136 |
+
"step": 92500
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 3.7891134289439377,
|
| 1140 |
+
"learning_rate": 1.0544328552803131e-06,
|
| 1141 |
+
"loss": 0.1440859375,
|
| 1142 |
+
"step": 93000
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 3.809485006518905,
|
| 1146 |
+
"learning_rate": 9.52574967405476e-07,
|
| 1147 |
+
"loss": 0.14747265625,
|
| 1148 |
+
"step": 93500
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 3.8298565840938723,
|
| 1152 |
+
"learning_rate": 8.507170795306389e-07,
|
| 1153 |
+
"loss": 0.12144921875,
|
| 1154 |
+
"step": 94000
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 3.8502281616688396,
|
| 1158 |
+
"learning_rate": 7.488591916558019e-07,
|
| 1159 |
+
"loss": 0.138546875,
|
| 1160 |
+
"step": 94500
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 3.870599739243807,
|
| 1164 |
+
"learning_rate": 6.470013037809649e-07,
|
| 1165 |
+
"loss": 0.1290390625,
|
| 1166 |
+
"step": 95000
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 3.8909713168187743,
|
| 1170 |
+
"learning_rate": 5.451434159061278e-07,
|
| 1171 |
+
"loss": 0.13651953125,
|
| 1172 |
+
"step": 95500
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 3.9113428943937416,
|
| 1176 |
+
"learning_rate": 4.432855280312908e-07,
|
| 1177 |
+
"loss": 0.14065234375,
|
| 1178 |
+
"step": 96000
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 3.9317144719687094,
|
| 1182 |
+
"learning_rate": 3.4142764015645373e-07,
|
| 1183 |
+
"loss": 0.129953125,
|
| 1184 |
+
"step": 96500
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 3.9520860495436767,
|
| 1188 |
+
"learning_rate": 2.395697522816167e-07,
|
| 1189 |
+
"loss": 0.127765625,
|
| 1190 |
+
"step": 97000
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 3.972457627118644,
|
| 1194 |
+
"learning_rate": 1.3771186440677968e-07,
|
| 1195 |
+
"loss": 0.12146875,
|
| 1196 |
+
"step": 97500
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 3.9928292046936114,
|
| 1200 |
+
"learning_rate": 3.585397653194264e-08,
|
| 1201 |
+
"loss": 0.132578125,
|
| 1202 |
+
"step": 98000
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 4.0,
|
| 1206 |
+
"eval_accuracy": 0.88762098828324,
|
| 1207 |
+
"eval_loss": 0.5551679134368896,
|
| 1208 |
+
"step": 98176
|
| 1209 |
+
}
|
| 1210 |
+
],
|
| 1211 |
+
"max_steps": 98176,
|
| 1212 |
+
"num_train_epochs": 4,
|
| 1213 |
+
"total_flos": 124136967570323400,
|
| 1214 |
+
"trial_name": null,
|
| 1215 |
+
"trial_params": null
|
| 1216 |
+
}
|