Commit
•
0c4aff2
1
Parent(s):
bb34f51
End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.25,
|
| 3 |
+
"eval_accuracy": 0.7368421052631579,
|
| 4 |
+
"eval_loss": 0.4468163251876831,
|
| 5 |
+
"eval_runtime": 87.805,
|
| 6 |
+
"eval_samples_per_second": 1.082,
|
| 7 |
+
"eval_steps_per_second": 0.547
|
| 8 |
+
}
|
runs/May19_00-20-31_str-3090-2/events.out.tfevents.1684439830.str-3090-2
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 sha256:413063506849578ccc2ad6f718273dea08c7ce5bf586c7c1811479aadea660b4
|
| 3 |
+
size 363
|
test_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.25,
|
| 3 |
+
"eval_accuracy": 0.7368421052631579,
|
| 4 |
+
"eval_loss": 0.4468163251876831,
|
| 5 |
+
"eval_runtime": 87.805,
|
| 6 |
+
"eval_samples_per_second": 1.082,
|
| 7 |
+
"eval_steps_per_second": 0.547
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1081 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.7115384615384616,
|
| 3 |
+
"best_model_checkpoint": "videomae-large-finetuned-kinetics-mopping/checkpoint-1672",
|
| 4 |
+
"epoch": 3.25,
|
| 5 |
+
"global_step": 1672,
|
| 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": 2.9761904761904763e-06,
|
| 13 |
+
"loss": 0.7257,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.01,
|
| 18 |
+
"learning_rate": 5.9523809523809525e-06,
|
| 19 |
+
"loss": 0.7446,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.02,
|
| 24 |
+
"learning_rate": 8.92857142857143e-06,
|
| 25 |
+
"loss": 0.7318,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.02,
|
| 30 |
+
"learning_rate": 1.1904761904761905e-05,
|
| 31 |
+
"loss": 0.6849,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.03,
|
| 36 |
+
"learning_rate": 1.4880952380952381e-05,
|
| 37 |
+
"loss": 0.66,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.04,
|
| 42 |
+
"learning_rate": 1.785714285714286e-05,
|
| 43 |
+
"loss": 0.5942,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.04,
|
| 48 |
+
"learning_rate": 2.0833333333333336e-05,
|
| 49 |
+
"loss": 1.0265,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.05,
|
| 54 |
+
"learning_rate": 2.380952380952381e-05,
|
| 55 |
+
"loss": 0.6231,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.05,
|
| 60 |
+
"learning_rate": 2.6785714285714288e-05,
|
| 61 |
+
"loss": 0.6112,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.06,
|
| 66 |
+
"learning_rate": 2.9761904761904762e-05,
|
| 67 |
+
"loss": 0.7304,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.07,
|
| 72 |
+
"learning_rate": 3.273809523809524e-05,
|
| 73 |
+
"loss": 0.5416,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.07,
|
| 78 |
+
"learning_rate": 3.571428571428572e-05,
|
| 79 |
+
"loss": 0.6291,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.08,
|
| 84 |
+
"learning_rate": 3.8690476190476195e-05,
|
| 85 |
+
"loss": 0.6386,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.08,
|
| 90 |
+
"learning_rate": 4.166666666666667e-05,
|
| 91 |
+
"loss": 0.5992,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.09,
|
| 96 |
+
"learning_rate": 4.464285714285715e-05,
|
| 97 |
+
"loss": 0.7842,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.1,
|
| 102 |
+
"learning_rate": 4.761904761904762e-05,
|
| 103 |
+
"loss": 0.6465,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.1,
|
| 108 |
+
"learning_rate": 4.993351063829788e-05,
|
| 109 |
+
"loss": 1.8611,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.11,
|
| 114 |
+
"learning_rate": 4.9601063829787235e-05,
|
| 115 |
+
"loss": 0.8756,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.11,
|
| 120 |
+
"learning_rate": 4.92686170212766e-05,
|
| 121 |
+
"loss": 0.6679,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.12,
|
| 126 |
+
"learning_rate": 4.893617021276596e-05,
|
| 127 |
+
"loss": 0.6566,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.13,
|
| 132 |
+
"learning_rate": 4.860372340425532e-05,
|
| 133 |
+
"loss": 0.543,
|
| 134 |
+
"step": 210
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.13,
|
| 138 |
+
"learning_rate": 4.8271276595744686e-05,
|
| 139 |
+
"loss": 0.547,
|
| 140 |
+
"step": 220
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.14,
|
| 144 |
+
"learning_rate": 4.793882978723405e-05,
|
| 145 |
+
"loss": 0.6893,
|
| 146 |
+
"step": 230
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.14,
|
| 150 |
+
"learning_rate": 4.7606382978723405e-05,
|
| 151 |
+
"loss": 0.9666,
|
| 152 |
+
"step": 240
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.15,
|
| 156 |
+
"learning_rate": 4.727393617021277e-05,
|
| 157 |
+
"loss": 0.7574,
|
| 158 |
+
"step": 250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.16,
|
| 162 |
+
"learning_rate": 4.694148936170213e-05,
|
| 163 |
+
"loss": 0.6599,
|
| 164 |
+
"step": 260
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.16,
|
| 168 |
+
"learning_rate": 4.660904255319149e-05,
|
| 169 |
+
"loss": 0.682,
|
| 170 |
+
"step": 270
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.17,
|
| 174 |
+
"learning_rate": 4.627659574468085e-05,
|
| 175 |
+
"loss": 0.6714,
|
| 176 |
+
"step": 280
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.17,
|
| 180 |
+
"learning_rate": 4.594414893617022e-05,
|
| 181 |
+
"loss": 0.7409,
|
| 182 |
+
"step": 290
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.18,
|
| 186 |
+
"learning_rate": 4.5611702127659576e-05,
|
| 187 |
+
"loss": 0.5337,
|
| 188 |
+
"step": 300
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.19,
|
| 192 |
+
"learning_rate": 4.527925531914894e-05,
|
| 193 |
+
"loss": 0.9172,
|
| 194 |
+
"step": 310
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.19,
|
| 198 |
+
"learning_rate": 4.49468085106383e-05,
|
| 199 |
+
"loss": 0.6787,
|
| 200 |
+
"step": 320
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.2,
|
| 204 |
+
"learning_rate": 4.461436170212766e-05,
|
| 205 |
+
"loss": 0.6626,
|
| 206 |
+
"step": 330
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.2,
|
| 210 |
+
"learning_rate": 4.428191489361702e-05,
|
| 211 |
+
"loss": 0.6113,
|
| 212 |
+
"step": 340
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.21,
|
| 216 |
+
"learning_rate": 4.394946808510639e-05,
|
| 217 |
+
"loss": 0.703,
|
| 218 |
+
"step": 350
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.22,
|
| 222 |
+
"learning_rate": 4.3617021276595746e-05,
|
| 223 |
+
"loss": 0.6532,
|
| 224 |
+
"step": 360
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.22,
|
| 228 |
+
"learning_rate": 4.328457446808511e-05,
|
| 229 |
+
"loss": 0.8501,
|
| 230 |
+
"step": 370
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.23,
|
| 234 |
+
"learning_rate": 4.295212765957447e-05,
|
| 235 |
+
"loss": 0.5234,
|
| 236 |
+
"step": 380
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.23,
|
| 240 |
+
"learning_rate": 4.2619680851063835e-05,
|
| 241 |
+
"loss": 0.8803,
|
| 242 |
+
"step": 390
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.24,
|
| 246 |
+
"learning_rate": 4.228723404255319e-05,
|
| 247 |
+
"loss": 0.6746,
|
| 248 |
+
"step": 400
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.25,
|
| 252 |
+
"learning_rate": 4.1954787234042554e-05,
|
| 253 |
+
"loss": 0.6868,
|
| 254 |
+
"step": 410
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.25,
|
| 258 |
+
"eval_accuracy": 0.21153846153846154,
|
| 259 |
+
"eval_loss": 0.7473957538604736,
|
| 260 |
+
"eval_runtime": 184.1286,
|
| 261 |
+
"eval_samples_per_second": 1.694,
|
| 262 |
+
"eval_steps_per_second": 0.847,
|
| 263 |
+
"step": 418
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.0,
|
| 267 |
+
"learning_rate": 4.1622340425531916e-05,
|
| 268 |
+
"loss": 0.6966,
|
| 269 |
+
"step": 420
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.01,
|
| 273 |
+
"learning_rate": 4.128989361702128e-05,
|
| 274 |
+
"loss": 0.615,
|
| 275 |
+
"step": 430
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.01,
|
| 279 |
+
"learning_rate": 4.095744680851064e-05,
|
| 280 |
+
"loss": 0.6828,
|
| 281 |
+
"step": 440
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 1.02,
|
| 285 |
+
"learning_rate": 4.0625000000000005e-05,
|
| 286 |
+
"loss": 0.627,
|
| 287 |
+
"step": 450
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 1.03,
|
| 291 |
+
"learning_rate": 4.029255319148936e-05,
|
| 292 |
+
"loss": 0.6394,
|
| 293 |
+
"step": 460
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 1.03,
|
| 297 |
+
"learning_rate": 3.9960106382978724e-05,
|
| 298 |
+
"loss": 0.4705,
|
| 299 |
+
"step": 470
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 1.04,
|
| 303 |
+
"learning_rate": 3.962765957446809e-05,
|
| 304 |
+
"loss": 0.8386,
|
| 305 |
+
"step": 480
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 1.04,
|
| 309 |
+
"learning_rate": 3.929521276595745e-05,
|
| 310 |
+
"loss": 0.6489,
|
| 311 |
+
"step": 490
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 1.05,
|
| 315 |
+
"learning_rate": 3.896276595744681e-05,
|
| 316 |
+
"loss": 0.4634,
|
| 317 |
+
"step": 500
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 1.06,
|
| 321 |
+
"learning_rate": 3.8630319148936175e-05,
|
| 322 |
+
"loss": 0.4205,
|
| 323 |
+
"step": 510
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 1.06,
|
| 327 |
+
"learning_rate": 3.829787234042553e-05,
|
| 328 |
+
"loss": 1.1553,
|
| 329 |
+
"step": 520
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 1.07,
|
| 333 |
+
"learning_rate": 3.7965425531914894e-05,
|
| 334 |
+
"loss": 0.968,
|
| 335 |
+
"step": 530
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 1.07,
|
| 339 |
+
"learning_rate": 3.763297872340426e-05,
|
| 340 |
+
"loss": 0.662,
|
| 341 |
+
"step": 540
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 1.08,
|
| 345 |
+
"learning_rate": 3.730053191489361e-05,
|
| 346 |
+
"loss": 0.8165,
|
| 347 |
+
"step": 550
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 1.08,
|
| 351 |
+
"learning_rate": 3.696808510638298e-05,
|
| 352 |
+
"loss": 0.8334,
|
| 353 |
+
"step": 560
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.09,
|
| 357 |
+
"learning_rate": 3.6635638297872346e-05,
|
| 358 |
+
"loss": 0.592,
|
| 359 |
+
"step": 570
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 1.1,
|
| 363 |
+
"learning_rate": 3.63031914893617e-05,
|
| 364 |
+
"loss": 0.6842,
|
| 365 |
+
"step": 580
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 1.1,
|
| 369 |
+
"learning_rate": 3.5970744680851064e-05,
|
| 370 |
+
"loss": 0.692,
|
| 371 |
+
"step": 590
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 1.11,
|
| 375 |
+
"learning_rate": 3.563829787234043e-05,
|
| 376 |
+
"loss": 0.7469,
|
| 377 |
+
"step": 600
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 1.11,
|
| 381 |
+
"learning_rate": 3.530585106382979e-05,
|
| 382 |
+
"loss": 0.7048,
|
| 383 |
+
"step": 610
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 1.12,
|
| 387 |
+
"learning_rate": 3.497340425531915e-05,
|
| 388 |
+
"loss": 0.7271,
|
| 389 |
+
"step": 620
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 1.13,
|
| 393 |
+
"learning_rate": 3.4640957446808516e-05,
|
| 394 |
+
"loss": 0.4924,
|
| 395 |
+
"step": 630
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 1.13,
|
| 399 |
+
"learning_rate": 3.430851063829787e-05,
|
| 400 |
+
"loss": 0.6303,
|
| 401 |
+
"step": 640
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.14,
|
| 405 |
+
"learning_rate": 3.3976063829787235e-05,
|
| 406 |
+
"loss": 0.8736,
|
| 407 |
+
"step": 650
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 1.14,
|
| 411 |
+
"learning_rate": 3.36436170212766e-05,
|
| 412 |
+
"loss": 0.5803,
|
| 413 |
+
"step": 660
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.15,
|
| 417 |
+
"learning_rate": 3.331117021276596e-05,
|
| 418 |
+
"loss": 0.7117,
|
| 419 |
+
"step": 670
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 1.16,
|
| 423 |
+
"learning_rate": 3.2978723404255317e-05,
|
| 424 |
+
"loss": 0.7229,
|
| 425 |
+
"step": 680
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 1.16,
|
| 429 |
+
"learning_rate": 3.2646276595744686e-05,
|
| 430 |
+
"loss": 0.5549,
|
| 431 |
+
"step": 690
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 1.17,
|
| 435 |
+
"learning_rate": 3.231382978723405e-05,
|
| 436 |
+
"loss": 0.596,
|
| 437 |
+
"step": 700
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.17,
|
| 441 |
+
"learning_rate": 3.1981382978723405e-05,
|
| 442 |
+
"loss": 0.6088,
|
| 443 |
+
"step": 710
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 1.18,
|
| 447 |
+
"learning_rate": 3.164893617021277e-05,
|
| 448 |
+
"loss": 1.2256,
|
| 449 |
+
"step": 720
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.19,
|
| 453 |
+
"learning_rate": 3.131648936170213e-05,
|
| 454 |
+
"loss": 0.9902,
|
| 455 |
+
"step": 730
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 1.19,
|
| 459 |
+
"learning_rate": 3.098404255319149e-05,
|
| 460 |
+
"loss": 0.4703,
|
| 461 |
+
"step": 740
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.2,
|
| 465 |
+
"learning_rate": 3.065159574468085e-05,
|
| 466 |
+
"loss": 0.7934,
|
| 467 |
+
"step": 750
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 1.2,
|
| 471 |
+
"learning_rate": 3.0319148936170216e-05,
|
| 472 |
+
"loss": 0.6965,
|
| 473 |
+
"step": 760
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.21,
|
| 477 |
+
"learning_rate": 2.998670212765958e-05,
|
| 478 |
+
"loss": 0.5744,
|
| 479 |
+
"step": 770
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.22,
|
| 483 |
+
"learning_rate": 2.9654255319148938e-05,
|
| 484 |
+
"loss": 0.4907,
|
| 485 |
+
"step": 780
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.22,
|
| 489 |
+
"learning_rate": 2.9321808510638298e-05,
|
| 490 |
+
"loss": 1.0854,
|
| 491 |
+
"step": 790
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 1.23,
|
| 495 |
+
"learning_rate": 2.898936170212766e-05,
|
| 496 |
+
"loss": 0.7813,
|
| 497 |
+
"step": 800
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 1.23,
|
| 501 |
+
"learning_rate": 2.865691489361702e-05,
|
| 502 |
+
"loss": 0.6902,
|
| 503 |
+
"step": 810
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 1.24,
|
| 507 |
+
"learning_rate": 2.8324468085106386e-05,
|
| 508 |
+
"loss": 0.8116,
|
| 509 |
+
"step": 820
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 1.25,
|
| 513 |
+
"learning_rate": 2.799202127659575e-05,
|
| 514 |
+
"loss": 0.7265,
|
| 515 |
+
"step": 830
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.25,
|
| 519 |
+
"eval_accuracy": 0.6025641025641025,
|
| 520 |
+
"eval_loss": 0.6471951007843018,
|
| 521 |
+
"eval_runtime": 175.8702,
|
| 522 |
+
"eval_samples_per_second": 1.774,
|
| 523 |
+
"eval_steps_per_second": 0.887,
|
| 524 |
+
"step": 836
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 2.0,
|
| 528 |
+
"learning_rate": 2.765957446808511e-05,
|
| 529 |
+
"loss": 0.6714,
|
| 530 |
+
"step": 840
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 2.01,
|
| 534 |
+
"learning_rate": 2.7327127659574468e-05,
|
| 535 |
+
"loss": 0.7021,
|
| 536 |
+
"step": 850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 2.01,
|
| 540 |
+
"learning_rate": 2.699468085106383e-05,
|
| 541 |
+
"loss": 0.5664,
|
| 542 |
+
"step": 860
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 2.02,
|
| 546 |
+
"learning_rate": 2.666223404255319e-05,
|
| 547 |
+
"loss": 0.6453,
|
| 548 |
+
"step": 870
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 2.03,
|
| 552 |
+
"learning_rate": 2.6329787234042553e-05,
|
| 553 |
+
"loss": 0.6841,
|
| 554 |
+
"step": 880
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 2.03,
|
| 558 |
+
"learning_rate": 2.599734042553192e-05,
|
| 559 |
+
"loss": 0.3523,
|
| 560 |
+
"step": 890
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 2.04,
|
| 564 |
+
"learning_rate": 2.566489361702128e-05,
|
| 565 |
+
"loss": 0.5916,
|
| 566 |
+
"step": 900
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 2.04,
|
| 570 |
+
"learning_rate": 2.5332446808510642e-05,
|
| 571 |
+
"loss": 0.6005,
|
| 572 |
+
"step": 910
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 2.05,
|
| 576 |
+
"learning_rate": 2.5e-05,
|
| 577 |
+
"loss": 0.9208,
|
| 578 |
+
"step": 920
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 2.06,
|
| 582 |
+
"learning_rate": 2.466755319148936e-05,
|
| 583 |
+
"loss": 1.1425,
|
| 584 |
+
"step": 930
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 2.06,
|
| 588 |
+
"learning_rate": 2.4335106382978727e-05,
|
| 589 |
+
"loss": 0.6431,
|
| 590 |
+
"step": 940
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 2.07,
|
| 594 |
+
"learning_rate": 2.4002659574468086e-05,
|
| 595 |
+
"loss": 0.843,
|
| 596 |
+
"step": 950
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 2.07,
|
| 600 |
+
"learning_rate": 2.3670212765957446e-05,
|
| 601 |
+
"loss": 0.6568,
|
| 602 |
+
"step": 960
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 2.08,
|
| 606 |
+
"learning_rate": 2.333776595744681e-05,
|
| 607 |
+
"loss": 0.5618,
|
| 608 |
+
"step": 970
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 2.09,
|
| 612 |
+
"learning_rate": 2.300531914893617e-05,
|
| 613 |
+
"loss": 0.4693,
|
| 614 |
+
"step": 980
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 2.09,
|
| 618 |
+
"learning_rate": 2.2672872340425534e-05,
|
| 619 |
+
"loss": 0.9186,
|
| 620 |
+
"step": 990
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 2.1,
|
| 624 |
+
"learning_rate": 2.2340425531914894e-05,
|
| 625 |
+
"loss": 0.6835,
|
| 626 |
+
"step": 1000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 2.1,
|
| 630 |
+
"learning_rate": 2.2007978723404257e-05,
|
| 631 |
+
"loss": 0.6809,
|
| 632 |
+
"step": 1010
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 2.11,
|
| 636 |
+
"learning_rate": 2.167553191489362e-05,
|
| 637 |
+
"loss": 0.607,
|
| 638 |
+
"step": 1020
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 2.12,
|
| 642 |
+
"learning_rate": 2.134308510638298e-05,
|
| 643 |
+
"loss": 0.6872,
|
| 644 |
+
"step": 1030
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 2.12,
|
| 648 |
+
"learning_rate": 2.1010638297872342e-05,
|
| 649 |
+
"loss": 0.6011,
|
| 650 |
+
"step": 1040
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 2.13,
|
| 654 |
+
"learning_rate": 2.0678191489361705e-05,
|
| 655 |
+
"loss": 0.6885,
|
| 656 |
+
"step": 1050
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 2.13,
|
| 660 |
+
"learning_rate": 2.0345744680851064e-05,
|
| 661 |
+
"loss": 0.5925,
|
| 662 |
+
"step": 1060
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 2.14,
|
| 666 |
+
"learning_rate": 2.0013297872340424e-05,
|
| 667 |
+
"loss": 0.5091,
|
| 668 |
+
"step": 1070
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 2.15,
|
| 672 |
+
"learning_rate": 1.968085106382979e-05,
|
| 673 |
+
"loss": 0.4648,
|
| 674 |
+
"step": 1080
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 2.15,
|
| 678 |
+
"learning_rate": 1.934840425531915e-05,
|
| 679 |
+
"loss": 0.7222,
|
| 680 |
+
"step": 1090
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 2.16,
|
| 684 |
+
"learning_rate": 1.9015957446808512e-05,
|
| 685 |
+
"loss": 0.4034,
|
| 686 |
+
"step": 1100
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 2.16,
|
| 690 |
+
"learning_rate": 1.8683510638297875e-05,
|
| 691 |
+
"loss": 0.5973,
|
| 692 |
+
"step": 1110
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 2.17,
|
| 696 |
+
"learning_rate": 1.8351063829787234e-05,
|
| 697 |
+
"loss": 1.1715,
|
| 698 |
+
"step": 1120
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 2.18,
|
| 702 |
+
"learning_rate": 1.8018617021276597e-05,
|
| 703 |
+
"loss": 0.9911,
|
| 704 |
+
"step": 1130
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 2.18,
|
| 708 |
+
"learning_rate": 1.768617021276596e-05,
|
| 709 |
+
"loss": 0.4329,
|
| 710 |
+
"step": 1140
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 2.19,
|
| 714 |
+
"learning_rate": 1.735372340425532e-05,
|
| 715 |
+
"loss": 0.7458,
|
| 716 |
+
"step": 1150
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 2.19,
|
| 720 |
+
"learning_rate": 1.7021276595744682e-05,
|
| 721 |
+
"loss": 1.1755,
|
| 722 |
+
"step": 1160
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 2.2,
|
| 726 |
+
"learning_rate": 1.6688829787234042e-05,
|
| 727 |
+
"loss": 0.6074,
|
| 728 |
+
"step": 1170
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 2.21,
|
| 732 |
+
"learning_rate": 1.6356382978723405e-05,
|
| 733 |
+
"loss": 0.8051,
|
| 734 |
+
"step": 1180
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 2.21,
|
| 738 |
+
"learning_rate": 1.6023936170212768e-05,
|
| 739 |
+
"loss": 0.6331,
|
| 740 |
+
"step": 1190
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 2.22,
|
| 744 |
+
"learning_rate": 1.5691489361702127e-05,
|
| 745 |
+
"loss": 0.6292,
|
| 746 |
+
"step": 1200
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 2.22,
|
| 750 |
+
"learning_rate": 1.535904255319149e-05,
|
| 751 |
+
"loss": 0.6519,
|
| 752 |
+
"step": 1210
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 2.23,
|
| 756 |
+
"learning_rate": 1.5026595744680853e-05,
|
| 757 |
+
"loss": 0.6466,
|
| 758 |
+
"step": 1220
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 2.24,
|
| 762 |
+
"learning_rate": 1.4694148936170212e-05,
|
| 763 |
+
"loss": 0.6337,
|
| 764 |
+
"step": 1230
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 2.24,
|
| 768 |
+
"learning_rate": 1.4361702127659577e-05,
|
| 769 |
+
"loss": 0.6831,
|
| 770 |
+
"step": 1240
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 2.25,
|
| 774 |
+
"learning_rate": 1.4029255319148938e-05,
|
| 775 |
+
"loss": 0.6854,
|
| 776 |
+
"step": 1250
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 2.25,
|
| 780 |
+
"eval_accuracy": 0.6346153846153846,
|
| 781 |
+
"eval_loss": 0.621135950088501,
|
| 782 |
+
"eval_runtime": 177.1967,
|
| 783 |
+
"eval_samples_per_second": 1.761,
|
| 784 |
+
"eval_steps_per_second": 0.88,
|
| 785 |
+
"step": 1254
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 3.0,
|
| 789 |
+
"learning_rate": 1.3696808510638297e-05,
|
| 790 |
+
"loss": 0.4754,
|
| 791 |
+
"step": 1260
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 3.01,
|
| 795 |
+
"learning_rate": 1.3364361702127659e-05,
|
| 796 |
+
"loss": 0.6731,
|
| 797 |
+
"step": 1270
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 3.02,
|
| 801 |
+
"learning_rate": 1.3031914893617023e-05,
|
| 802 |
+
"loss": 0.5832,
|
| 803 |
+
"step": 1280
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 3.02,
|
| 807 |
+
"learning_rate": 1.2699468085106384e-05,
|
| 808 |
+
"loss": 0.6755,
|
| 809 |
+
"step": 1290
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 3.03,
|
| 813 |
+
"learning_rate": 1.2367021276595745e-05,
|
| 814 |
+
"loss": 0.7011,
|
| 815 |
+
"step": 1300
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 3.03,
|
| 819 |
+
"learning_rate": 1.2034574468085107e-05,
|
| 820 |
+
"loss": 0.5694,
|
| 821 |
+
"step": 1310
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 3.04,
|
| 825 |
+
"learning_rate": 1.170212765957447e-05,
|
| 826 |
+
"loss": 0.6381,
|
| 827 |
+
"step": 1320
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 3.05,
|
| 831 |
+
"learning_rate": 1.136968085106383e-05,
|
| 832 |
+
"loss": 0.6598,
|
| 833 |
+
"step": 1330
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 3.05,
|
| 837 |
+
"learning_rate": 1.1037234042553192e-05,
|
| 838 |
+
"loss": 0.6255,
|
| 839 |
+
"step": 1340
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 3.06,
|
| 843 |
+
"learning_rate": 1.0704787234042555e-05,
|
| 844 |
+
"loss": 0.5323,
|
| 845 |
+
"step": 1350
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 3.06,
|
| 849 |
+
"learning_rate": 1.0372340425531916e-05,
|
| 850 |
+
"loss": 0.4125,
|
| 851 |
+
"step": 1360
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 3.07,
|
| 855 |
+
"learning_rate": 1.0039893617021277e-05,
|
| 856 |
+
"loss": 0.7932,
|
| 857 |
+
"step": 1370
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 3.08,
|
| 861 |
+
"learning_rate": 9.707446808510638e-06,
|
| 862 |
+
"loss": 0.5528,
|
| 863 |
+
"step": 1380
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 3.08,
|
| 867 |
+
"learning_rate": 9.375000000000001e-06,
|
| 868 |
+
"loss": 0.6247,
|
| 869 |
+
"step": 1390
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 3.09,
|
| 873 |
+
"learning_rate": 9.042553191489362e-06,
|
| 874 |
+
"loss": 0.5198,
|
| 875 |
+
"step": 1400
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 3.09,
|
| 879 |
+
"learning_rate": 8.710106382978723e-06,
|
| 880 |
+
"loss": 0.6433,
|
| 881 |
+
"step": 1410
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 3.1,
|
| 885 |
+
"learning_rate": 8.377659574468086e-06,
|
| 886 |
+
"loss": 0.5981,
|
| 887 |
+
"step": 1420
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 3.11,
|
| 891 |
+
"learning_rate": 8.045212765957447e-06,
|
| 892 |
+
"loss": 0.7204,
|
| 893 |
+
"step": 1430
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 3.11,
|
| 897 |
+
"learning_rate": 7.712765957446808e-06,
|
| 898 |
+
"loss": 0.7437,
|
| 899 |
+
"step": 1440
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 3.12,
|
| 903 |
+
"learning_rate": 7.380319148936171e-06,
|
| 904 |
+
"loss": 0.6888,
|
| 905 |
+
"step": 1450
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 3.12,
|
| 909 |
+
"learning_rate": 7.047872340425532e-06,
|
| 910 |
+
"loss": 0.6113,
|
| 911 |
+
"step": 1460
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 3.13,
|
| 915 |
+
"learning_rate": 6.715425531914894e-06,
|
| 916 |
+
"loss": 0.6399,
|
| 917 |
+
"step": 1470
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 3.14,
|
| 921 |
+
"learning_rate": 6.3829787234042555e-06,
|
| 922 |
+
"loss": 0.5685,
|
| 923 |
+
"step": 1480
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 3.14,
|
| 927 |
+
"learning_rate": 6.0505319148936175e-06,
|
| 928 |
+
"loss": 0.7251,
|
| 929 |
+
"step": 1490
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 3.15,
|
| 933 |
+
"learning_rate": 5.718085106382979e-06,
|
| 934 |
+
"loss": 0.629,
|
| 935 |
+
"step": 1500
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 3.15,
|
| 939 |
+
"learning_rate": 5.385638297872341e-06,
|
| 940 |
+
"loss": 0.5404,
|
| 941 |
+
"step": 1510
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 3.16,
|
| 945 |
+
"learning_rate": 5.053191489361702e-06,
|
| 946 |
+
"loss": 0.6974,
|
| 947 |
+
"step": 1520
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 3.17,
|
| 951 |
+
"learning_rate": 4.720744680851065e-06,
|
| 952 |
+
"loss": 0.7332,
|
| 953 |
+
"step": 1530
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 3.17,
|
| 957 |
+
"learning_rate": 4.388297872340426e-06,
|
| 958 |
+
"loss": 0.6905,
|
| 959 |
+
"step": 1540
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 3.18,
|
| 963 |
+
"learning_rate": 4.055851063829787e-06,
|
| 964 |
+
"loss": 0.6313,
|
| 965 |
+
"step": 1550
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 3.18,
|
| 969 |
+
"learning_rate": 3.723404255319149e-06,
|
| 970 |
+
"loss": 0.6934,
|
| 971 |
+
"step": 1560
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 3.19,
|
| 975 |
+
"learning_rate": 3.3909574468085105e-06,
|
| 976 |
+
"loss": 0.6479,
|
| 977 |
+
"step": 1570
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 3.19,
|
| 981 |
+
"learning_rate": 3.0585106382978726e-06,
|
| 982 |
+
"loss": 0.6756,
|
| 983 |
+
"step": 1580
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 3.2,
|
| 987 |
+
"learning_rate": 2.726063829787234e-06,
|
| 988 |
+
"loss": 0.7393,
|
| 989 |
+
"step": 1590
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 3.21,
|
| 993 |
+
"learning_rate": 2.3936170212765957e-06,
|
| 994 |
+
"loss": 0.5521,
|
| 995 |
+
"step": 1600
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 3.21,
|
| 999 |
+
"learning_rate": 2.0611702127659573e-06,
|
| 1000 |
+
"loss": 0.5931,
|
| 1001 |
+
"step": 1610
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 3.22,
|
| 1005 |
+
"learning_rate": 1.7287234042553193e-06,
|
| 1006 |
+
"loss": 0.5324,
|
| 1007 |
+
"step": 1620
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 3.22,
|
| 1011 |
+
"learning_rate": 1.3962765957446809e-06,
|
| 1012 |
+
"loss": 0.6355,
|
| 1013 |
+
"step": 1630
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 3.23,
|
| 1017 |
+
"learning_rate": 1.0638297872340427e-06,
|
| 1018 |
+
"loss": 0.9237,
|
| 1019 |
+
"step": 1640
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 3.24,
|
| 1023 |
+
"learning_rate": 7.313829787234043e-07,
|
| 1024 |
+
"loss": 0.8162,
|
| 1025 |
+
"step": 1650
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 3.24,
|
| 1029 |
+
"learning_rate": 3.9893617021276597e-07,
|
| 1030 |
+
"loss": 0.6747,
|
| 1031 |
+
"step": 1660
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 3.25,
|
| 1035 |
+
"learning_rate": 6.648936170212767e-08,
|
| 1036 |
+
"loss": 0.7829,
|
| 1037 |
+
"step": 1670
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 3.25,
|
| 1041 |
+
"eval_accuracy": 0.7115384615384616,
|
| 1042 |
+
"eval_loss": 0.5958611965179443,
|
| 1043 |
+
"eval_runtime": 177.9778,
|
| 1044 |
+
"eval_samples_per_second": 1.753,
|
| 1045 |
+
"eval_steps_per_second": 0.877,
|
| 1046 |
+
"step": 1672
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 3.25,
|
| 1050 |
+
"step": 1672,
|
| 1051 |
+
"total_flos": 1.4683591998352392e+19,
|
| 1052 |
+
"train_loss": 0.6908089216958964,
|
| 1053 |
+
"train_runtime": 3902.1292,
|
| 1054 |
+
"train_samples_per_second": 0.857,
|
| 1055 |
+
"train_steps_per_second": 0.428
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 3.25,
|
| 1059 |
+
"eval_accuracy": 0.7368421052631579,
|
| 1060 |
+
"eval_loss": 0.4468162953853607,
|
| 1061 |
+
"eval_runtime": 90.4466,
|
| 1062 |
+
"eval_samples_per_second": 1.05,
|
| 1063 |
+
"eval_steps_per_second": 0.531,
|
| 1064 |
+
"step": 1672
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 3.25,
|
| 1068 |
+
"eval_accuracy": 0.7368421052631579,
|
| 1069 |
+
"eval_loss": 0.4468163251876831,
|
| 1070 |
+
"eval_runtime": 87.805,
|
| 1071 |
+
"eval_samples_per_second": 1.082,
|
| 1072 |
+
"eval_steps_per_second": 0.547,
|
| 1073 |
+
"step": 1672
|
| 1074 |
+
}
|
| 1075 |
+
],
|
| 1076 |
+
"max_steps": 1672,
|
| 1077 |
+
"num_train_epochs": 9223372036854775807,
|
| 1078 |
+
"total_flos": 1.4683591998352392e+19,
|
| 1079 |
+
"trial_name": null,
|
| 1080 |
+
"trial_params": null
|
| 1081 |
+
}
|