End of training
Browse files- all_results.json +13 -0
- eval_results.json +8 -0
- runs/May06_12-49-49_ca6400a0cf7a/events.out.tfevents.1715003319.ca6400a0cf7a.925.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +1117 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9998384230085636,
|
| 3 |
+
"eval_accuracy": 0.9605053856292324,
|
| 4 |
+
"eval_loss": 0.11349575221538544,
|
| 5 |
+
"eval_runtime": 151.8333,
|
| 6 |
+
"eval_samples_per_second": 144.915,
|
| 7 |
+
"eval_steps_per_second": 4.531,
|
| 8 |
+
"total_flos": 4.921883515023262e+18,
|
| 9 |
+
"train_loss": 0.25020766805354133,
|
| 10 |
+
"train_runtime": 3368.4805,
|
| 11 |
+
"train_samples_per_second": 58.787,
|
| 12 |
+
"train_steps_per_second": 0.459
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9998384230085636,
|
| 3 |
+
"eval_accuracy": 0.9605053856292324,
|
| 4 |
+
"eval_loss": 0.11349575221538544,
|
| 5 |
+
"eval_runtime": 151.8333,
|
| 6 |
+
"eval_samples_per_second": 144.915,
|
| 7 |
+
"eval_steps_per_second": 4.531
|
| 8 |
+
}
|
runs/May06_12-49-49_ca6400a0cf7a/events.out.tfevents.1715003319.ca6400a0cf7a.925.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f109431f070ab2af3d841865c67ed313156e323778b909a611991da77fd8d3fb
|
| 3 |
+
size 411
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9998384230085636,
|
| 3 |
+
"total_flos": 4.921883515023262e+18,
|
| 4 |
+
"train_loss": 0.25020766805354133,
|
| 5 |
+
"train_runtime": 3368.4805,
|
| 6 |
+
"train_samples_per_second": 58.787,
|
| 7 |
+
"train_steps_per_second": 0.459
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1117 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9605053856292324,
|
| 3 |
+
"best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-cancer/checkpoint-1547",
|
| 4 |
+
"epoch": 0.9998384230085636,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1547,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.006463079657456778,
|
| 13 |
+
"grad_norm": 7.331174850463867,
|
| 14 |
+
"learning_rate": 3.225806451612903e-06,
|
| 15 |
+
"loss": 0.7162,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.012926159314913557,
|
| 20 |
+
"grad_norm": 4.575641632080078,
|
| 21 |
+
"learning_rate": 6.451612903225806e-06,
|
| 22 |
+
"loss": 0.6444,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.019389238972370333,
|
| 27 |
+
"grad_norm": 3.2741119861602783,
|
| 28 |
+
"learning_rate": 9.67741935483871e-06,
|
| 29 |
+
"loss": 0.5618,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.025852318629827113,
|
| 34 |
+
"grad_norm": 4.564846515655518,
|
| 35 |
+
"learning_rate": 1.2903225806451613e-05,
|
| 36 |
+
"loss": 0.4959,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.03231539828728389,
|
| 41 |
+
"grad_norm": 4.165364742279053,
|
| 42 |
+
"learning_rate": 1.6129032258064517e-05,
|
| 43 |
+
"loss": 0.447,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.038778477944740666,
|
| 48 |
+
"grad_norm": 4.854116439819336,
|
| 49 |
+
"learning_rate": 1.935483870967742e-05,
|
| 50 |
+
"loss": 0.409,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.045241557602197446,
|
| 55 |
+
"grad_norm": 4.5222907066345215,
|
| 56 |
+
"learning_rate": 2.258064516129032e-05,
|
| 57 |
+
"loss": 0.3992,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.051704637259654226,
|
| 62 |
+
"grad_norm": 5.930300235748291,
|
| 63 |
+
"learning_rate": 2.5806451612903226e-05,
|
| 64 |
+
"loss": 0.3832,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.058167716917111006,
|
| 69 |
+
"grad_norm": 6.286398410797119,
|
| 70 |
+
"learning_rate": 2.9032258064516133e-05,
|
| 71 |
+
"loss": 0.399,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.06463079657456779,
|
| 76 |
+
"grad_norm": 8.192097663879395,
|
| 77 |
+
"learning_rate": 3.2258064516129034e-05,
|
| 78 |
+
"loss": 0.3631,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.07109387623202457,
|
| 83 |
+
"grad_norm": 7.482578754425049,
|
| 84 |
+
"learning_rate": 3.548387096774194e-05,
|
| 85 |
+
"loss": 0.3702,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.07755695588948133,
|
| 90 |
+
"grad_norm": 5.62371301651001,
|
| 91 |
+
"learning_rate": 3.870967741935484e-05,
|
| 92 |
+
"loss": 0.352,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.08402003554693811,
|
| 97 |
+
"grad_norm": 5.444067001342773,
|
| 98 |
+
"learning_rate": 4.1935483870967746e-05,
|
| 99 |
+
"loss": 0.3338,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.09048311520439489,
|
| 104 |
+
"grad_norm": 4.892695426940918,
|
| 105 |
+
"learning_rate": 4.516129032258064e-05,
|
| 106 |
+
"loss": 0.3463,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.09694619486185167,
|
| 111 |
+
"grad_norm": 5.656785488128662,
|
| 112 |
+
"learning_rate": 4.8387096774193554e-05,
|
| 113 |
+
"loss": 0.3718,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.10340927451930845,
|
| 118 |
+
"grad_norm": 7.789802074432373,
|
| 119 |
+
"learning_rate": 4.982040229885058e-05,
|
| 120 |
+
"loss": 0.3412,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.10987235417676523,
|
| 125 |
+
"grad_norm": 7.198510646820068,
|
| 126 |
+
"learning_rate": 4.946120689655172e-05,
|
| 127 |
+
"loss": 0.3577,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.11633543383422201,
|
| 132 |
+
"grad_norm": 7.838411808013916,
|
| 133 |
+
"learning_rate": 4.9102011494252875e-05,
|
| 134 |
+
"loss": 0.3522,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.12279851349167878,
|
| 139 |
+
"grad_norm": 7.711742401123047,
|
| 140 |
+
"learning_rate": 4.8742816091954026e-05,
|
| 141 |
+
"loss": 0.3381,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.12926159314913557,
|
| 146 |
+
"grad_norm": 5.089247703552246,
|
| 147 |
+
"learning_rate": 4.838362068965517e-05,
|
| 148 |
+
"loss": 0.3507,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.13572467280659234,
|
| 153 |
+
"grad_norm": 4.4015913009643555,
|
| 154 |
+
"learning_rate": 4.802442528735632e-05,
|
| 155 |
+
"loss": 0.3162,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.14218775246404913,
|
| 160 |
+
"grad_norm": 4.59859561920166,
|
| 161 |
+
"learning_rate": 4.7665229885057474e-05,
|
| 162 |
+
"loss": 0.274,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.1486508321215059,
|
| 167 |
+
"grad_norm": 3.7790114879608154,
|
| 168 |
+
"learning_rate": 4.730603448275862e-05,
|
| 169 |
+
"loss": 0.3161,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.15511391177896267,
|
| 174 |
+
"grad_norm": 5.165287017822266,
|
| 175 |
+
"learning_rate": 4.694683908045977e-05,
|
| 176 |
+
"loss": 0.2878,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.16157699143641946,
|
| 181 |
+
"grad_norm": 6.651523590087891,
|
| 182 |
+
"learning_rate": 4.658764367816092e-05,
|
| 183 |
+
"loss": 0.3133,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.16804007109387623,
|
| 188 |
+
"grad_norm": 6.430370807647705,
|
| 189 |
+
"learning_rate": 4.622844827586207e-05,
|
| 190 |
+
"loss": 0.2744,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.17450315075133302,
|
| 195 |
+
"grad_norm": 3.7495827674865723,
|
| 196 |
+
"learning_rate": 4.5869252873563225e-05,
|
| 197 |
+
"loss": 0.2638,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.18096623040878979,
|
| 202 |
+
"grad_norm": 4.071202278137207,
|
| 203 |
+
"learning_rate": 4.551005747126437e-05,
|
| 204 |
+
"loss": 0.2698,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.18742931006624658,
|
| 209 |
+
"grad_norm": 5.824757099151611,
|
| 210 |
+
"learning_rate": 4.515086206896552e-05,
|
| 211 |
+
"loss": 0.2916,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.19389238972370335,
|
| 216 |
+
"grad_norm": 4.58206033706665,
|
| 217 |
+
"learning_rate": 4.4791666666666673e-05,
|
| 218 |
+
"loss": 0.2674,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.2003554693811601,
|
| 223 |
+
"grad_norm": 7.343913555145264,
|
| 224 |
+
"learning_rate": 4.443247126436782e-05,
|
| 225 |
+
"loss": 0.3069,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.2068185490386169,
|
| 230 |
+
"grad_norm": 4.5890302658081055,
|
| 231 |
+
"learning_rate": 4.407327586206897e-05,
|
| 232 |
+
"loss": 0.2889,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.21328162869607367,
|
| 237 |
+
"grad_norm": 6.601052761077881,
|
| 238 |
+
"learning_rate": 4.371408045977012e-05,
|
| 239 |
+
"loss": 0.2509,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.21974470835353047,
|
| 244 |
+
"grad_norm": 2.973667621612549,
|
| 245 |
+
"learning_rate": 4.3354885057471266e-05,
|
| 246 |
+
"loss": 0.2667,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.22620778801098723,
|
| 251 |
+
"grad_norm": 7.966582298278809,
|
| 252 |
+
"learning_rate": 4.299568965517242e-05,
|
| 253 |
+
"loss": 0.2888,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.23267086766844403,
|
| 258 |
+
"grad_norm": 2.4454495906829834,
|
| 259 |
+
"learning_rate": 4.263649425287356e-05,
|
| 260 |
+
"loss": 0.2638,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.2391339473259008,
|
| 265 |
+
"grad_norm": 4.780755996704102,
|
| 266 |
+
"learning_rate": 4.2277298850574714e-05,
|
| 267 |
+
"loss": 0.2771,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.24559702698335756,
|
| 272 |
+
"grad_norm": 7.205840587615967,
|
| 273 |
+
"learning_rate": 4.1918103448275866e-05,
|
| 274 |
+
"loss": 0.289,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.25206010664081435,
|
| 279 |
+
"grad_norm": 6.309414863586426,
|
| 280 |
+
"learning_rate": 4.155890804597701e-05,
|
| 281 |
+
"loss": 0.2471,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.25852318629827115,
|
| 286 |
+
"grad_norm": 4.736650466918945,
|
| 287 |
+
"learning_rate": 4.119971264367816e-05,
|
| 288 |
+
"loss": 0.2204,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.2649862659557279,
|
| 293 |
+
"grad_norm": 4.276492595672607,
|
| 294 |
+
"learning_rate": 4.0840517241379314e-05,
|
| 295 |
+
"loss": 0.2546,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.2714493456131847,
|
| 300 |
+
"grad_norm": 4.888199329376221,
|
| 301 |
+
"learning_rate": 4.048132183908046e-05,
|
| 302 |
+
"loss": 0.2343,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.27791242527064147,
|
| 307 |
+
"grad_norm": 3.0027356147766113,
|
| 308 |
+
"learning_rate": 4.012212643678161e-05,
|
| 309 |
+
"loss": 0.2644,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.28437550492809827,
|
| 314 |
+
"grad_norm": 3.5034611225128174,
|
| 315 |
+
"learning_rate": 3.976293103448276e-05,
|
| 316 |
+
"loss": 0.2318,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.290838584585555,
|
| 321 |
+
"grad_norm": 4.029373645782471,
|
| 322 |
+
"learning_rate": 3.940373563218391e-05,
|
| 323 |
+
"loss": 0.2393,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.2973016642430118,
|
| 328 |
+
"grad_norm": 2.631014585494995,
|
| 329 |
+
"learning_rate": 3.904454022988506e-05,
|
| 330 |
+
"loss": 0.2497,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.3037647439004686,
|
| 335 |
+
"grad_norm": 3.7511560916900635,
|
| 336 |
+
"learning_rate": 3.86853448275862e-05,
|
| 337 |
+
"loss": 0.2776,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.31022782355792533,
|
| 342 |
+
"grad_norm": 4.721391677856445,
|
| 343 |
+
"learning_rate": 3.8326149425287355e-05,
|
| 344 |
+
"loss": 0.2353,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.3166909032153821,
|
| 349 |
+
"grad_norm": 3.97963285446167,
|
| 350 |
+
"learning_rate": 3.796695402298851e-05,
|
| 351 |
+
"loss": 0.2549,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.3231539828728389,
|
| 356 |
+
"grad_norm": 4.646782398223877,
|
| 357 |
+
"learning_rate": 3.760775862068966e-05,
|
| 358 |
+
"loss": 0.2473,
|
| 359 |
+
"step": 500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.3296170625302957,
|
| 363 |
+
"grad_norm": 4.661077976226807,
|
| 364 |
+
"learning_rate": 3.724856321839081e-05,
|
| 365 |
+
"loss": 0.2581,
|
| 366 |
+
"step": 510
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.33608014218775245,
|
| 370 |
+
"grad_norm": 7.082584381103516,
|
| 371 |
+
"learning_rate": 3.6889367816091954e-05,
|
| 372 |
+
"loss": 0.2427,
|
| 373 |
+
"step": 520
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.34254322184520924,
|
| 377 |
+
"grad_norm": 4.724052906036377,
|
| 378 |
+
"learning_rate": 3.6530172413793106e-05,
|
| 379 |
+
"loss": 0.2683,
|
| 380 |
+
"step": 530
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.34900630150266604,
|
| 384 |
+
"grad_norm": 3.6246917247772217,
|
| 385 |
+
"learning_rate": 3.617097701149426e-05,
|
| 386 |
+
"loss": 0.2636,
|
| 387 |
+
"step": 540
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.3554693811601228,
|
| 391 |
+
"grad_norm": 9.237018585205078,
|
| 392 |
+
"learning_rate": 3.58117816091954e-05,
|
| 393 |
+
"loss": 0.2502,
|
| 394 |
+
"step": 550
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.36193246081757957,
|
| 398 |
+
"grad_norm": 3.987290382385254,
|
| 399 |
+
"learning_rate": 3.5452586206896554e-05,
|
| 400 |
+
"loss": 0.2359,
|
| 401 |
+
"step": 560
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.36839554047503636,
|
| 405 |
+
"grad_norm": 6.946052074432373,
|
| 406 |
+
"learning_rate": 3.5093390804597706e-05,
|
| 407 |
+
"loss": 0.2445,
|
| 408 |
+
"step": 570
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.37485862013249316,
|
| 412 |
+
"grad_norm": 4.0834736824035645,
|
| 413 |
+
"learning_rate": 3.473419540229885e-05,
|
| 414 |
+
"loss": 0.2515,
|
| 415 |
+
"step": 580
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.3813216997899499,
|
| 419 |
+
"grad_norm": 3.3558614253997803,
|
| 420 |
+
"learning_rate": 3.4375e-05,
|
| 421 |
+
"loss": 0.2511,
|
| 422 |
+
"step": 590
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.3877847794474067,
|
| 426 |
+
"grad_norm": 4.314471244812012,
|
| 427 |
+
"learning_rate": 3.4015804597701154e-05,
|
| 428 |
+
"loss": 0.232,
|
| 429 |
+
"step": 600
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.3942478591048635,
|
| 433 |
+
"grad_norm": 3.44812273979187,
|
| 434 |
+
"learning_rate": 3.36566091954023e-05,
|
| 435 |
+
"loss": 0.1993,
|
| 436 |
+
"step": 610
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.4007109387623202,
|
| 440 |
+
"grad_norm": 4.297704219818115,
|
| 441 |
+
"learning_rate": 3.329741379310345e-05,
|
| 442 |
+
"loss": 0.2556,
|
| 443 |
+
"step": 620
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.407174018419777,
|
| 447 |
+
"grad_norm": 2.1731884479522705,
|
| 448 |
+
"learning_rate": 3.2938218390804595e-05,
|
| 449 |
+
"loss": 0.2168,
|
| 450 |
+
"step": 630
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.4136370980772338,
|
| 454 |
+
"grad_norm": 3.119593620300293,
|
| 455 |
+
"learning_rate": 3.2579022988505747e-05,
|
| 456 |
+
"loss": 0.2089,
|
| 457 |
+
"step": 640
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.4201001777346906,
|
| 461 |
+
"grad_norm": 3.971031665802002,
|
| 462 |
+
"learning_rate": 3.22198275862069e-05,
|
| 463 |
+
"loss": 0.2373,
|
| 464 |
+
"step": 650
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.42656325739214734,
|
| 468 |
+
"grad_norm": 3.3221435546875,
|
| 469 |
+
"learning_rate": 3.186063218390804e-05,
|
| 470 |
+
"loss": 0.243,
|
| 471 |
+
"step": 660
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.43302633704960414,
|
| 475 |
+
"grad_norm": 3.1322009563446045,
|
| 476 |
+
"learning_rate": 3.1501436781609195e-05,
|
| 477 |
+
"loss": 0.2239,
|
| 478 |
+
"step": 670
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.43948941670706093,
|
| 482 |
+
"grad_norm": 3.9616572856903076,
|
| 483 |
+
"learning_rate": 3.1142241379310346e-05,
|
| 484 |
+
"loss": 0.2316,
|
| 485 |
+
"step": 680
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.44595249636451767,
|
| 489 |
+
"grad_norm": 2.667262077331543,
|
| 490 |
+
"learning_rate": 3.078304597701149e-05,
|
| 491 |
+
"loss": 0.204,
|
| 492 |
+
"step": 690
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.45241557602197446,
|
| 496 |
+
"grad_norm": 5.354103088378906,
|
| 497 |
+
"learning_rate": 3.042385057471265e-05,
|
| 498 |
+
"loss": 0.2035,
|
| 499 |
+
"step": 700
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.45887865567943126,
|
| 503 |
+
"grad_norm": 3.422032356262207,
|
| 504 |
+
"learning_rate": 3.0064655172413798e-05,
|
| 505 |
+
"loss": 0.2518,
|
| 506 |
+
"step": 710
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.46534173533688805,
|
| 510 |
+
"grad_norm": 6.190507411956787,
|
| 511 |
+
"learning_rate": 2.9705459770114946e-05,
|
| 512 |
+
"loss": 0.2148,
|
| 513 |
+
"step": 720
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.4718048149943448,
|
| 517 |
+
"grad_norm": 4.419785976409912,
|
| 518 |
+
"learning_rate": 2.9346264367816094e-05,
|
| 519 |
+
"loss": 0.2399,
|
| 520 |
+
"step": 730
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.4782678946518016,
|
| 524 |
+
"grad_norm": 6.475925922393799,
|
| 525 |
+
"learning_rate": 2.8987068965517246e-05,
|
| 526 |
+
"loss": 0.2308,
|
| 527 |
+
"step": 740
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.4847309743092584,
|
| 531 |
+
"grad_norm": 2.907485246658325,
|
| 532 |
+
"learning_rate": 2.8627873563218394e-05,
|
| 533 |
+
"loss": 0.2071,
|
| 534 |
+
"step": 750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.4911940539667151,
|
| 538 |
+
"grad_norm": 3.476051092147827,
|
| 539 |
+
"learning_rate": 2.8268678160919542e-05,
|
| 540 |
+
"loss": 0.2374,
|
| 541 |
+
"step": 760
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.4976571336241719,
|
| 545 |
+
"grad_norm": 6.298080921173096,
|
| 546 |
+
"learning_rate": 2.7909482758620694e-05,
|
| 547 |
+
"loss": 0.2277,
|
| 548 |
+
"step": 770
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.5041202132816287,
|
| 552 |
+
"grad_norm": 4.822377681732178,
|
| 553 |
+
"learning_rate": 2.7550287356321842e-05,
|
| 554 |
+
"loss": 0.2259,
|
| 555 |
+
"step": 780
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.5105832929390854,
|
| 559 |
+
"grad_norm": 3.966501474380493,
|
| 560 |
+
"learning_rate": 2.719109195402299e-05,
|
| 561 |
+
"loss": 0.2212,
|
| 562 |
+
"step": 790
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.5170463725965423,
|
| 566 |
+
"grad_norm": 2.599501848220825,
|
| 567 |
+
"learning_rate": 2.6831896551724138e-05,
|
| 568 |
+
"loss": 0.2202,
|
| 569 |
+
"step": 800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.523509452253999,
|
| 573 |
+
"grad_norm": 4.360517501831055,
|
| 574 |
+
"learning_rate": 2.647270114942529e-05,
|
| 575 |
+
"loss": 0.2482,
|
| 576 |
+
"step": 810
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.5299725319114558,
|
| 580 |
+
"grad_norm": 4.807380199432373,
|
| 581 |
+
"learning_rate": 2.6113505747126438e-05,
|
| 582 |
+
"loss": 0.2222,
|
| 583 |
+
"step": 820
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.5364356115689126,
|
| 587 |
+
"grad_norm": 3.3132829666137695,
|
| 588 |
+
"learning_rate": 2.5754310344827586e-05,
|
| 589 |
+
"loss": 0.2421,
|
| 590 |
+
"step": 830
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.5428986912263694,
|
| 594 |
+
"grad_norm": 6.560469150543213,
|
| 595 |
+
"learning_rate": 2.5395114942528734e-05,
|
| 596 |
+
"loss": 0.196,
|
| 597 |
+
"step": 840
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.5493617708838261,
|
| 601 |
+
"grad_norm": 7.168336391448975,
|
| 602 |
+
"learning_rate": 2.5035919540229886e-05,
|
| 603 |
+
"loss": 0.244,
|
| 604 |
+
"step": 850
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.5558248505412829,
|
| 608 |
+
"grad_norm": 3.6738462448120117,
|
| 609 |
+
"learning_rate": 2.4676724137931034e-05,
|
| 610 |
+
"loss": 0.2361,
|
| 611 |
+
"step": 860
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.5622879301987397,
|
| 615 |
+
"grad_norm": 4.0579833984375,
|
| 616 |
+
"learning_rate": 2.4317528735632186e-05,
|
| 617 |
+
"loss": 0.2195,
|
| 618 |
+
"step": 870
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.5687510098561965,
|
| 622 |
+
"grad_norm": 5.091606616973877,
|
| 623 |
+
"learning_rate": 2.3958333333333334e-05,
|
| 624 |
+
"loss": 0.2128,
|
| 625 |
+
"step": 880
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.5752140895136533,
|
| 629 |
+
"grad_norm": 7.0507354736328125,
|
| 630 |
+
"learning_rate": 2.3599137931034486e-05,
|
| 631 |
+
"loss": 0.2259,
|
| 632 |
+
"step": 890
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.58167716917111,
|
| 636 |
+
"grad_norm": 4.719695568084717,
|
| 637 |
+
"learning_rate": 2.3239942528735634e-05,
|
| 638 |
+
"loss": 0.2119,
|
| 639 |
+
"step": 900
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.5881402488285669,
|
| 643 |
+
"grad_norm": 4.4290289878845215,
|
| 644 |
+
"learning_rate": 2.2880747126436782e-05,
|
| 645 |
+
"loss": 0.2138,
|
| 646 |
+
"step": 910
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.5946033284860236,
|
| 650 |
+
"grad_norm": 4.132259368896484,
|
| 651 |
+
"learning_rate": 2.2521551724137934e-05,
|
| 652 |
+
"loss": 0.2111,
|
| 653 |
+
"step": 920
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.6010664081434803,
|
| 657 |
+
"grad_norm": 6.010085105895996,
|
| 658 |
+
"learning_rate": 2.2162356321839082e-05,
|
| 659 |
+
"loss": 0.2106,
|
| 660 |
+
"step": 930
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.6075294878009372,
|
| 664 |
+
"grad_norm": 3.847933769226074,
|
| 665 |
+
"learning_rate": 2.180316091954023e-05,
|
| 666 |
+
"loss": 0.2146,
|
| 667 |
+
"step": 940
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.6139925674583939,
|
| 671 |
+
"grad_norm": 4.39457893371582,
|
| 672 |
+
"learning_rate": 2.144396551724138e-05,
|
| 673 |
+
"loss": 0.2279,
|
| 674 |
+
"step": 950
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.6204556471158507,
|
| 678 |
+
"grad_norm": 3.269622564315796,
|
| 679 |
+
"learning_rate": 2.108477011494253e-05,
|
| 680 |
+
"loss": 0.2259,
|
| 681 |
+
"step": 960
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.6269187267733075,
|
| 685 |
+
"grad_norm": 3.5994322299957275,
|
| 686 |
+
"learning_rate": 2.0725574712643678e-05,
|
| 687 |
+
"loss": 0.1996,
|
| 688 |
+
"step": 970
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.6333818064307642,
|
| 692 |
+
"grad_norm": 4.982938289642334,
|
| 693 |
+
"learning_rate": 2.036637931034483e-05,
|
| 694 |
+
"loss": 0.2311,
|
| 695 |
+
"step": 980
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.639844886088221,
|
| 699 |
+
"grad_norm": 5.611133098602295,
|
| 700 |
+
"learning_rate": 2.0007183908045978e-05,
|
| 701 |
+
"loss": 0.2073,
|
| 702 |
+
"step": 990
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.6463079657456778,
|
| 706 |
+
"grad_norm": 4.818994522094727,
|
| 707 |
+
"learning_rate": 1.964798850574713e-05,
|
| 708 |
+
"loss": 0.2233,
|
| 709 |
+
"step": 1000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.6527710454031346,
|
| 713 |
+
"grad_norm": 2.8192803859710693,
|
| 714 |
+
"learning_rate": 1.9288793103448278e-05,
|
| 715 |
+
"loss": 0.1841,
|
| 716 |
+
"step": 1010
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.6592341250605914,
|
| 720 |
+
"grad_norm": 3.74814510345459,
|
| 721 |
+
"learning_rate": 1.8929597701149426e-05,
|
| 722 |
+
"loss": 0.1908,
|
| 723 |
+
"step": 1020
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.6656972047180482,
|
| 727 |
+
"grad_norm": 5.619290351867676,
|
| 728 |
+
"learning_rate": 1.8570402298850574e-05,
|
| 729 |
+
"loss": 0.2135,
|
| 730 |
+
"step": 1030
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.6721602843755049,
|
| 734 |
+
"grad_norm": 3.431131362915039,
|
| 735 |
+
"learning_rate": 1.8211206896551726e-05,
|
| 736 |
+
"loss": 0.2036,
|
| 737 |
+
"step": 1040
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.6786233640329618,
|
| 741 |
+
"grad_norm": 4.3107781410217285,
|
| 742 |
+
"learning_rate": 1.7852011494252874e-05,
|
| 743 |
+
"loss": 0.2013,
|
| 744 |
+
"step": 1050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.6850864436904185,
|
| 748 |
+
"grad_norm": 2.6932215690612793,
|
| 749 |
+
"learning_rate": 1.7492816091954022e-05,
|
| 750 |
+
"loss": 0.1654,
|
| 751 |
+
"step": 1060
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.6915495233478752,
|
| 755 |
+
"grad_norm": 5.07429313659668,
|
| 756 |
+
"learning_rate": 1.7133620689655174e-05,
|
| 757 |
+
"loss": 0.2128,
|
| 758 |
+
"step": 1070
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.6980126030053321,
|
| 762 |
+
"grad_norm": 6.4850873947143555,
|
| 763 |
+
"learning_rate": 1.6774425287356325e-05,
|
| 764 |
+
"loss": 0.1926,
|
| 765 |
+
"step": 1080
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.7044756826627888,
|
| 769 |
+
"grad_norm": 4.696744441986084,
|
| 770 |
+
"learning_rate": 1.6415229885057474e-05,
|
| 771 |
+
"loss": 0.2329,
|
| 772 |
+
"step": 1090
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.7109387623202456,
|
| 776 |
+
"grad_norm": 7.928685188293457,
|
| 777 |
+
"learning_rate": 1.6056034482758622e-05,
|
| 778 |
+
"loss": 0.2494,
|
| 779 |
+
"step": 1100
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.7174018419777024,
|
| 783 |
+
"grad_norm": 6.6810455322265625,
|
| 784 |
+
"learning_rate": 1.569683908045977e-05,
|
| 785 |
+
"loss": 0.2082,
|
| 786 |
+
"step": 1110
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.7238649216351591,
|
| 790 |
+
"grad_norm": 6.868622303009033,
|
| 791 |
+
"learning_rate": 1.533764367816092e-05,
|
| 792 |
+
"loss": 0.2153,
|
| 793 |
+
"step": 1120
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.7303280012926159,
|
| 797 |
+
"grad_norm": 4.763729095458984,
|
| 798 |
+
"learning_rate": 1.497844827586207e-05,
|
| 799 |
+
"loss": 0.1972,
|
| 800 |
+
"step": 1130
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.7367910809500727,
|
| 804 |
+
"grad_norm": 3.208186149597168,
|
| 805 |
+
"learning_rate": 1.461925287356322e-05,
|
| 806 |
+
"loss": 0.201,
|
| 807 |
+
"step": 1140
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.7432541606075295,
|
| 811 |
+
"grad_norm": 4.111704349517822,
|
| 812 |
+
"learning_rate": 1.4260057471264368e-05,
|
| 813 |
+
"loss": 0.2154,
|
| 814 |
+
"step": 1150
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.7497172402649863,
|
| 818 |
+
"grad_norm": 6.802502632141113,
|
| 819 |
+
"learning_rate": 1.3900862068965518e-05,
|
| 820 |
+
"loss": 0.1819,
|
| 821 |
+
"step": 1160
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.7561803199224431,
|
| 825 |
+
"grad_norm": 4.15958833694458,
|
| 826 |
+
"learning_rate": 1.3541666666666666e-05,
|
| 827 |
+
"loss": 0.1823,
|
| 828 |
+
"step": 1170
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.7626433995798998,
|
| 832 |
+
"grad_norm": 2.806739568710327,
|
| 833 |
+
"learning_rate": 1.3182471264367816e-05,
|
| 834 |
+
"loss": 0.1963,
|
| 835 |
+
"step": 1180
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.7691064792373566,
|
| 839 |
+
"grad_norm": 5.593130111694336,
|
| 840 |
+
"learning_rate": 1.2823275862068968e-05,
|
| 841 |
+
"loss": 0.1948,
|
| 842 |
+
"step": 1190
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.7755695588948134,
|
| 846 |
+
"grad_norm": 7.160956859588623,
|
| 847 |
+
"learning_rate": 1.2464080459770116e-05,
|
| 848 |
+
"loss": 0.2249,
|
| 849 |
+
"step": 1200
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.7820326385522701,
|
| 853 |
+
"grad_norm": 5.844871520996094,
|
| 854 |
+
"learning_rate": 1.2104885057471264e-05,
|
| 855 |
+
"loss": 0.2109,
|
| 856 |
+
"step": 1210
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.788495718209727,
|
| 860 |
+
"grad_norm": 4.273647785186768,
|
| 861 |
+
"learning_rate": 1.1745689655172416e-05,
|
| 862 |
+
"loss": 0.1976,
|
| 863 |
+
"step": 1220
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.7949587978671837,
|
| 867 |
+
"grad_norm": 5.624157428741455,
|
| 868 |
+
"learning_rate": 1.1386494252873564e-05,
|
| 869 |
+
"loss": 0.2017,
|
| 870 |
+
"step": 1230
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.8014218775246404,
|
| 874 |
+
"grad_norm": 5.627654552459717,
|
| 875 |
+
"learning_rate": 1.1027298850574714e-05,
|
| 876 |
+
"loss": 0.2153,
|
| 877 |
+
"step": 1240
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.8078849571820973,
|
| 881 |
+
"grad_norm": 4.900112628936768,
|
| 882 |
+
"learning_rate": 1.0668103448275862e-05,
|
| 883 |
+
"loss": 0.2436,
|
| 884 |
+
"step": 1250
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.814348036839554,
|
| 888 |
+
"grad_norm": 4.697551727294922,
|
| 889 |
+
"learning_rate": 1.0308908045977012e-05,
|
| 890 |
+
"loss": 0.1912,
|
| 891 |
+
"step": 1260
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.8208111164970108,
|
| 895 |
+
"grad_norm": 7.477961540222168,
|
| 896 |
+
"learning_rate": 9.949712643678162e-06,
|
| 897 |
+
"loss": 0.2109,
|
| 898 |
+
"step": 1270
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.8272741961544676,
|
| 902 |
+
"grad_norm": 4.095799922943115,
|
| 903 |
+
"learning_rate": 9.590517241379312e-06,
|
| 904 |
+
"loss": 0.2129,
|
| 905 |
+
"step": 1280
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.8337372758119244,
|
| 909 |
+
"grad_norm": 6.4529924392700195,
|
| 910 |
+
"learning_rate": 9.23132183908046e-06,
|
| 911 |
+
"loss": 0.2272,
|
| 912 |
+
"step": 1290
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.8402003554693812,
|
| 916 |
+
"grad_norm": 5.603827476501465,
|
| 917 |
+
"learning_rate": 8.87212643678161e-06,
|
| 918 |
+
"loss": 0.1771,
|
| 919 |
+
"step": 1300
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.846663435126838,
|
| 923 |
+
"grad_norm": 9.214598655700684,
|
| 924 |
+
"learning_rate": 8.512931034482758e-06,
|
| 925 |
+
"loss": 0.1775,
|
| 926 |
+
"step": 1310
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.8531265147842947,
|
| 930 |
+
"grad_norm": 3.415295124053955,
|
| 931 |
+
"learning_rate": 8.153735632183908e-06,
|
| 932 |
+
"loss": 0.1973,
|
| 933 |
+
"step": 1320
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.8595895944417515,
|
| 937 |
+
"grad_norm": 3.975978136062622,
|
| 938 |
+
"learning_rate": 7.794540229885058e-06,
|
| 939 |
+
"loss": 0.187,
|
| 940 |
+
"step": 1330
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.8660526740992083,
|
| 944 |
+
"grad_norm": 2.804140090942383,
|
| 945 |
+
"learning_rate": 7.435344827586208e-06,
|
| 946 |
+
"loss": 0.2051,
|
| 947 |
+
"step": 1340
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.872515753756665,
|
| 951 |
+
"grad_norm": 6.0080647468566895,
|
| 952 |
+
"learning_rate": 7.076149425287357e-06,
|
| 953 |
+
"loss": 0.216,
|
| 954 |
+
"step": 1350
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.8789788334141219,
|
| 958 |
+
"grad_norm": 4.182724475860596,
|
| 959 |
+
"learning_rate": 6.716954022988506e-06,
|
| 960 |
+
"loss": 0.1982,
|
| 961 |
+
"step": 1360
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.8854419130715786,
|
| 965 |
+
"grad_norm": 7.1342854499816895,
|
| 966 |
+
"learning_rate": 6.357758620689655e-06,
|
| 967 |
+
"loss": 0.1881,
|
| 968 |
+
"step": 1370
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.8919049927290353,
|
| 972 |
+
"grad_norm": 4.865418434143066,
|
| 973 |
+
"learning_rate": 5.998563218390805e-06,
|
| 974 |
+
"loss": 0.2099,
|
| 975 |
+
"step": 1380
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.8983680723864922,
|
| 979 |
+
"grad_norm": 4.020890235900879,
|
| 980 |
+
"learning_rate": 5.639367816091954e-06,
|
| 981 |
+
"loss": 0.1819,
|
| 982 |
+
"step": 1390
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.9048311520439489,
|
| 986 |
+
"grad_norm": 3.826892852783203,
|
| 987 |
+
"learning_rate": 5.280172413793104e-06,
|
| 988 |
+
"loss": 0.1846,
|
| 989 |
+
"step": 1400
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.9112942317014057,
|
| 993 |
+
"grad_norm": 5.284408092498779,
|
| 994 |
+
"learning_rate": 4.920977011494253e-06,
|
| 995 |
+
"loss": 0.1888,
|
| 996 |
+
"step": 1410
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.9177573113588625,
|
| 1000 |
+
"grad_norm": 4.285012722015381,
|
| 1001 |
+
"learning_rate": 4.561781609195403e-06,
|
| 1002 |
+
"loss": 0.1886,
|
| 1003 |
+
"step": 1420
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.9242203910163193,
|
| 1007 |
+
"grad_norm": 3.0137135982513428,
|
| 1008 |
+
"learning_rate": 4.202586206896552e-06,
|
| 1009 |
+
"loss": 0.1976,
|
| 1010 |
+
"step": 1430
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.9306834706737761,
|
| 1014 |
+
"grad_norm": 7.759015083312988,
|
| 1015 |
+
"learning_rate": 3.843390804597701e-06,
|
| 1016 |
+
"loss": 0.1891,
|
| 1017 |
+
"step": 1440
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.9371465503312328,
|
| 1021 |
+
"grad_norm": 4.454775810241699,
|
| 1022 |
+
"learning_rate": 3.484195402298851e-06,
|
| 1023 |
+
"loss": 0.2085,
|
| 1024 |
+
"step": 1450
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.9436096299886896,
|
| 1028 |
+
"grad_norm": 4.0864949226379395,
|
| 1029 |
+
"learning_rate": 3.125e-06,
|
| 1030 |
+
"loss": 0.1859,
|
| 1031 |
+
"step": 1460
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.9500727096461464,
|
| 1035 |
+
"grad_norm": 6.020027160644531,
|
| 1036 |
+
"learning_rate": 2.7658045977011496e-06,
|
| 1037 |
+
"loss": 0.1994,
|
| 1038 |
+
"step": 1470
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.9565357893036032,
|
| 1042 |
+
"grad_norm": 5.507518768310547,
|
| 1043 |
+
"learning_rate": 2.4066091954022987e-06,
|
| 1044 |
+
"loss": 0.1908,
|
| 1045 |
+
"step": 1480
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.9629988689610599,
|
| 1049 |
+
"grad_norm": 4.609635829925537,
|
| 1050 |
+
"learning_rate": 2.047413793103448e-06,
|
| 1051 |
+
"loss": 0.2239,
|
| 1052 |
+
"step": 1490
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.9694619486185168,
|
| 1056 |
+
"grad_norm": 5.80496072769165,
|
| 1057 |
+
"learning_rate": 1.6882183908045979e-06,
|
| 1058 |
+
"loss": 0.1824,
|
| 1059 |
+
"step": 1500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.9759250282759735,
|
| 1063 |
+
"grad_norm": 4.583664417266846,
|
| 1064 |
+
"learning_rate": 1.3290229885057471e-06,
|
| 1065 |
+
"loss": 0.1866,
|
| 1066 |
+
"step": 1510
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.9823881079334302,
|
| 1070 |
+
"grad_norm": 5.140854358673096,
|
| 1071 |
+
"learning_rate": 9.698275862068966e-07,
|
| 1072 |
+
"loss": 0.1693,
|
| 1073 |
+
"step": 1520
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.9888511875908871,
|
| 1077 |
+
"grad_norm": 4.870802402496338,
|
| 1078 |
+
"learning_rate": 6.106321839080461e-07,
|
| 1079 |
+
"loss": 0.1786,
|
| 1080 |
+
"step": 1530
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.9953142672483438,
|
| 1084 |
+
"grad_norm": 4.085971832275391,
|
| 1085 |
+
"learning_rate": 2.514367816091954e-07,
|
| 1086 |
+
"loss": 0.1819,
|
| 1087 |
+
"step": 1540
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.9998384230085636,
|
| 1091 |
+
"eval_accuracy": 0.9605053856292324,
|
| 1092 |
+
"eval_loss": 0.11349575221538544,
|
| 1093 |
+
"eval_runtime": 169.4025,
|
| 1094 |
+
"eval_samples_per_second": 129.886,
|
| 1095 |
+
"eval_steps_per_second": 4.061,
|
| 1096 |
+
"step": 1547
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 0.9998384230085636,
|
| 1100 |
+
"step": 1547,
|
| 1101 |
+
"total_flos": 4.921883515023262e+18,
|
| 1102 |
+
"train_loss": 0.25020766805354133,
|
| 1103 |
+
"train_runtime": 3368.4805,
|
| 1104 |
+
"train_samples_per_second": 58.787,
|
| 1105 |
+
"train_steps_per_second": 0.459
|
| 1106 |
+
}
|
| 1107 |
+
],
|
| 1108 |
+
"logging_steps": 10,
|
| 1109 |
+
"max_steps": 1547,
|
| 1110 |
+
"num_input_tokens_seen": 0,
|
| 1111 |
+
"num_train_epochs": 1,
|
| 1112 |
+
"save_steps": 500,
|
| 1113 |
+
"total_flos": 4.921883515023262e+18,
|
| 1114 |
+
"train_batch_size": 32,
|
| 1115 |
+
"trial_name": null,
|
| 1116 |
+
"trial_params": null
|
| 1117 |
+
}
|