Training in progress, epoch 0
Browse files- all_results.json +8 -0
- logs/events.out.tfevents.1731795006.AZHAR-PC.2358415.0 +3 -0
- model.safetensors +1 -1
- train_results.json +8 -0
- trainer_state.json +3352 -0
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 99.24812030075188,
|
| 3 |
+
"total_flos": 2.7357484620290458e+19,
|
| 4 |
+
"train_loss": 0.32434610511768946,
|
| 5 |
+
"train_runtime": 9596.5736,
|
| 6 |
+
"train_samples_per_second": 5.544,
|
| 7 |
+
"train_steps_per_second": 0.344
|
| 8 |
+
}
|
logs/events.out.tfevents.1731795006.AZHAR-PC.2358415.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ab7546bccfddb2b52e7a26509c7a5ebf182797e6f48e8df7a4dd4074adc02414
|
| 3 |
+
size 6664
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 784887780
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0ddf4b63488c6d75e4fbbe0dddceb8a69a1a73dd54b5568a5dd71761aa2321b8
|
| 3 |
size 784887780
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 99.24812030075188,
|
| 3 |
+
"total_flos": 2.7357484620290458e+19,
|
| 4 |
+
"train_loss": 0.32434610511768946,
|
| 5 |
+
"train_runtime": 9596.5736,
|
| 6 |
+
"train_samples_per_second": 5.544,
|
| 7 |
+
"train_steps_per_second": 0.344
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3352 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.6666666666666666,
|
| 3 |
+
"best_model_checkpoint": "/data/data2/azhar/mine/testing_chamber/bpn-mining-model/swin-transformer3/checkpoint-532",
|
| 4 |
+
"epoch": 99.24812030075188,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 3300,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.3007518796992481,
|
| 13 |
+
"grad_norm": 14.913878440856934,
|
| 14 |
+
"learning_rate": 1.5151515151515152e-06,
|
| 15 |
+
"loss": 2.3582,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.6015037593984962,
|
| 20 |
+
"grad_norm": 15.318342208862305,
|
| 21 |
+
"learning_rate": 3.0303030303030305e-06,
|
| 22 |
+
"loss": 2.1222,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.9022556390977443,
|
| 27 |
+
"grad_norm": 20.099170684814453,
|
| 28 |
+
"learning_rate": 4.5454545454545455e-06,
|
| 29 |
+
"loss": 1.6659,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.9924812030075187,
|
| 34 |
+
"eval_accuracy": 0.6333333333333333,
|
| 35 |
+
"eval_f1": 0.624952380952381,
|
| 36 |
+
"eval_loss": 1.0638588666915894,
|
| 37 |
+
"eval_runtime": 6.7375,
|
| 38 |
+
"eval_samples_per_second": 8.905,
|
| 39 |
+
"eval_steps_per_second": 2.226,
|
| 40 |
+
"step": 33
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"epoch": 1.2030075187969924,
|
| 44 |
+
"grad_norm": 18.861299514770508,
|
| 45 |
+
"learning_rate": 6.060606060606061e-06,
|
| 46 |
+
"loss": 1.0907,
|
| 47 |
+
"step": 40
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 1.5037593984962405,
|
| 51 |
+
"grad_norm": 14.466804504394531,
|
| 52 |
+
"learning_rate": 7.5757575757575764e-06,
|
| 53 |
+
"loss": 0.8232,
|
| 54 |
+
"step": 50
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 1.8045112781954886,
|
| 58 |
+
"grad_norm": 21.599777221679688,
|
| 59 |
+
"learning_rate": 9.090909090909091e-06,
|
| 60 |
+
"loss": 0.7561,
|
| 61 |
+
"step": 60
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 1.9849624060150375,
|
| 65 |
+
"eval_accuracy": 0.5166666666666667,
|
| 66 |
+
"eval_f1": 0.352014652014652,
|
| 67 |
+
"eval_loss": 0.7257525324821472,
|
| 68 |
+
"eval_runtime": 6.3119,
|
| 69 |
+
"eval_samples_per_second": 9.506,
|
| 70 |
+
"eval_steps_per_second": 2.376,
|
| 71 |
+
"step": 66
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 2.1052631578947367,
|
| 75 |
+
"grad_norm": 8.610815048217773,
|
| 76 |
+
"learning_rate": 1.0606060606060607e-05,
|
| 77 |
+
"loss": 0.7409,
|
| 78 |
+
"step": 70
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 2.406015037593985,
|
| 82 |
+
"grad_norm": 7.523438930511475,
|
| 83 |
+
"learning_rate": 1.2121212121212122e-05,
|
| 84 |
+
"loss": 0.7021,
|
| 85 |
+
"step": 80
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 2.706766917293233,
|
| 89 |
+
"grad_norm": 17.187030792236328,
|
| 90 |
+
"learning_rate": 1.3636363636363637e-05,
|
| 91 |
+
"loss": 0.7106,
|
| 92 |
+
"step": 90
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 2.9774436090225564,
|
| 96 |
+
"eval_accuracy": 0.5,
|
| 97 |
+
"eval_f1": 0.3755129958960329,
|
| 98 |
+
"eval_loss": 0.7334391474723816,
|
| 99 |
+
"eval_runtime": 6.4801,
|
| 100 |
+
"eval_samples_per_second": 9.259,
|
| 101 |
+
"eval_steps_per_second": 2.315,
|
| 102 |
+
"step": 99
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 3.007518796992481,
|
| 106 |
+
"grad_norm": 20.541980743408203,
|
| 107 |
+
"learning_rate": 1.5151515151515153e-05,
|
| 108 |
+
"loss": 0.6882,
|
| 109 |
+
"step": 100
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 3.308270676691729,
|
| 113 |
+
"grad_norm": 18.346364974975586,
|
| 114 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 115 |
+
"loss": 0.7986,
|
| 116 |
+
"step": 110
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 3.6090225563909772,
|
| 120 |
+
"grad_norm": 14.835421562194824,
|
| 121 |
+
"learning_rate": 1.8181818181818182e-05,
|
| 122 |
+
"loss": 0.721,
|
| 123 |
+
"step": 120
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 3.909774436090226,
|
| 127 |
+
"grad_norm": 5.117762565612793,
|
| 128 |
+
"learning_rate": 1.9696969696969697e-05,
|
| 129 |
+
"loss": 0.6749,
|
| 130 |
+
"step": 130
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 4.0,
|
| 134 |
+
"eval_accuracy": 0.48333333333333334,
|
| 135 |
+
"eval_f1": 0.3661064425770308,
|
| 136 |
+
"eval_loss": 0.7087975740432739,
|
| 137 |
+
"eval_runtime": 6.5607,
|
| 138 |
+
"eval_samples_per_second": 9.145,
|
| 139 |
+
"eval_steps_per_second": 2.286,
|
| 140 |
+
"step": 133
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 4.2105263157894735,
|
| 144 |
+
"grad_norm": 5.199806213378906,
|
| 145 |
+
"learning_rate": 2.1212121212121215e-05,
|
| 146 |
+
"loss": 0.6836,
|
| 147 |
+
"step": 140
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 4.511278195488722,
|
| 151 |
+
"grad_norm": 5.881252288818359,
|
| 152 |
+
"learning_rate": 2.272727272727273e-05,
|
| 153 |
+
"loss": 0.7035,
|
| 154 |
+
"step": 150
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 4.81203007518797,
|
| 158 |
+
"grad_norm": 19.35033416748047,
|
| 159 |
+
"learning_rate": 2.4242424242424244e-05,
|
| 160 |
+
"loss": 0.751,
|
| 161 |
+
"step": 160
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 4.992481203007519,
|
| 165 |
+
"eval_accuracy": 0.48333333333333334,
|
| 166 |
+
"eval_f1": 0.3661064425770308,
|
| 167 |
+
"eval_loss": 0.735622763633728,
|
| 168 |
+
"eval_runtime": 6.9036,
|
| 169 |
+
"eval_samples_per_second": 8.691,
|
| 170 |
+
"eval_steps_per_second": 2.173,
|
| 171 |
+
"step": 166
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 5.112781954887218,
|
| 175 |
+
"grad_norm": 9.151863098144531,
|
| 176 |
+
"learning_rate": 2.575757575757576e-05,
|
| 177 |
+
"loss": 0.6634,
|
| 178 |
+
"step": 170
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 5.413533834586466,
|
| 182 |
+
"grad_norm": 23.333147048950195,
|
| 183 |
+
"learning_rate": 2.7272727272727273e-05,
|
| 184 |
+
"loss": 0.7291,
|
| 185 |
+
"step": 180
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 5.714285714285714,
|
| 189 |
+
"grad_norm": 5.380008220672607,
|
| 190 |
+
"learning_rate": 2.878787878787879e-05,
|
| 191 |
+
"loss": 0.7146,
|
| 192 |
+
"step": 190
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 5.984962406015038,
|
| 196 |
+
"eval_accuracy": 0.48333333333333334,
|
| 197 |
+
"eval_f1": 0.3149812734082397,
|
| 198 |
+
"eval_loss": 0.7836605310440063,
|
| 199 |
+
"eval_runtime": 6.6834,
|
| 200 |
+
"eval_samples_per_second": 8.977,
|
| 201 |
+
"eval_steps_per_second": 2.244,
|
| 202 |
+
"step": 199
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 6.015037593984962,
|
| 206 |
+
"grad_norm": 5.286394119262695,
|
| 207 |
+
"learning_rate": 3.0303030303030306e-05,
|
| 208 |
+
"loss": 0.7468,
|
| 209 |
+
"step": 200
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 6.315789473684211,
|
| 213 |
+
"grad_norm": 6.3437323570251465,
|
| 214 |
+
"learning_rate": 3.181818181818182e-05,
|
| 215 |
+
"loss": 0.6837,
|
| 216 |
+
"step": 210
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 6.616541353383458,
|
| 220 |
+
"grad_norm": 22.48058319091797,
|
| 221 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 222 |
+
"loss": 0.7065,
|
| 223 |
+
"step": 220
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 6.917293233082707,
|
| 227 |
+
"grad_norm": 10.246678352355957,
|
| 228 |
+
"learning_rate": 3.484848484848485e-05,
|
| 229 |
+
"loss": 0.6699,
|
| 230 |
+
"step": 230
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 6.977443609022556,
|
| 234 |
+
"eval_accuracy": 0.48333333333333334,
|
| 235 |
+
"eval_f1": 0.34242424242424246,
|
| 236 |
+
"eval_loss": 0.75690758228302,
|
| 237 |
+
"eval_runtime": 6.5095,
|
| 238 |
+
"eval_samples_per_second": 9.217,
|
| 239 |
+
"eval_steps_per_second": 2.304,
|
| 240 |
+
"step": 232
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 7.2180451127819545,
|
| 244 |
+
"grad_norm": 6.964570045471191,
|
| 245 |
+
"learning_rate": 3.6363636363636364e-05,
|
| 246 |
+
"loss": 0.6882,
|
| 247 |
+
"step": 240
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 7.518796992481203,
|
| 251 |
+
"grad_norm": 11.94023609161377,
|
| 252 |
+
"learning_rate": 3.787878787878788e-05,
|
| 253 |
+
"loss": 0.6776,
|
| 254 |
+
"step": 250
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 7.819548872180452,
|
| 258 |
+
"grad_norm": 10.98417854309082,
|
| 259 |
+
"learning_rate": 3.939393939393939e-05,
|
| 260 |
+
"loss": 0.6521,
|
| 261 |
+
"step": 260
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 8.0,
|
| 265 |
+
"eval_accuracy": 0.5333333333333333,
|
| 266 |
+
"eval_f1": 0.4674368848951647,
|
| 267 |
+
"eval_loss": 0.7255270481109619,
|
| 268 |
+
"eval_runtime": 6.3154,
|
| 269 |
+
"eval_samples_per_second": 9.501,
|
| 270 |
+
"eval_steps_per_second": 2.375,
|
| 271 |
+
"step": 266
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 8.1203007518797,
|
| 275 |
+
"grad_norm": 22.907865524291992,
|
| 276 |
+
"learning_rate": 4.0909090909090915e-05,
|
| 277 |
+
"loss": 0.7176,
|
| 278 |
+
"step": 270
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 8.421052631578947,
|
| 282 |
+
"grad_norm": 10.256561279296875,
|
| 283 |
+
"learning_rate": 4.242424242424243e-05,
|
| 284 |
+
"loss": 0.5968,
|
| 285 |
+
"step": 280
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 8.721804511278195,
|
| 289 |
+
"grad_norm": 18.539447784423828,
|
| 290 |
+
"learning_rate": 4.3939393939393944e-05,
|
| 291 |
+
"loss": 0.6885,
|
| 292 |
+
"step": 290
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 8.992481203007518,
|
| 296 |
+
"eval_accuracy": 0.5166666666666667,
|
| 297 |
+
"eval_f1": 0.4070028011204482,
|
| 298 |
+
"eval_loss": 0.7252522706985474,
|
| 299 |
+
"eval_runtime": 6.719,
|
| 300 |
+
"eval_samples_per_second": 8.93,
|
| 301 |
+
"eval_steps_per_second": 2.232,
|
| 302 |
+
"step": 299
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 9.022556390977444,
|
| 306 |
+
"grad_norm": 4.7263336181640625,
|
| 307 |
+
"learning_rate": 4.545454545454546e-05,
|
| 308 |
+
"loss": 0.7568,
|
| 309 |
+
"step": 300
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 9.323308270676693,
|
| 313 |
+
"grad_norm": 11.226090431213379,
|
| 314 |
+
"learning_rate": 4.696969696969697e-05,
|
| 315 |
+
"loss": 0.6667,
|
| 316 |
+
"step": 310
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 9.62406015037594,
|
| 320 |
+
"grad_norm": 20.75147819519043,
|
| 321 |
+
"learning_rate": 4.848484848484849e-05,
|
| 322 |
+
"loss": 0.6658,
|
| 323 |
+
"step": 320
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 9.924812030075188,
|
| 327 |
+
"grad_norm": 23.47601318359375,
|
| 328 |
+
"learning_rate": 5e-05,
|
| 329 |
+
"loss": 0.6407,
|
| 330 |
+
"step": 330
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 9.984962406015038,
|
| 334 |
+
"eval_accuracy": 0.6,
|
| 335 |
+
"eval_f1": 0.5908571428571429,
|
| 336 |
+
"eval_loss": 0.6506010890007019,
|
| 337 |
+
"eval_runtime": 6.6092,
|
| 338 |
+
"eval_samples_per_second": 9.078,
|
| 339 |
+
"eval_steps_per_second": 2.27,
|
| 340 |
+
"step": 332
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 10.225563909774436,
|
| 344 |
+
"grad_norm": 14.811893463134766,
|
| 345 |
+
"learning_rate": 4.983164983164983e-05,
|
| 346 |
+
"loss": 0.654,
|
| 347 |
+
"step": 340
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 10.526315789473685,
|
| 351 |
+
"grad_norm": 10.347320556640625,
|
| 352 |
+
"learning_rate": 4.966329966329967e-05,
|
| 353 |
+
"loss": 0.6607,
|
| 354 |
+
"step": 350
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 10.827067669172932,
|
| 358 |
+
"grad_norm": 10.15239143371582,
|
| 359 |
+
"learning_rate": 4.94949494949495e-05,
|
| 360 |
+
"loss": 0.6436,
|
| 361 |
+
"step": 360
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 10.977443609022556,
|
| 365 |
+
"eval_accuracy": 0.55,
|
| 366 |
+
"eval_f1": 0.44420062695924767,
|
| 367 |
+
"eval_loss": 0.6720244288444519,
|
| 368 |
+
"eval_runtime": 6.5965,
|
| 369 |
+
"eval_samples_per_second": 9.096,
|
| 370 |
+
"eval_steps_per_second": 2.274,
|
| 371 |
+
"step": 365
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 11.12781954887218,
|
| 375 |
+
"grad_norm": 10.023416519165039,
|
| 376 |
+
"learning_rate": 4.932659932659932e-05,
|
| 377 |
+
"loss": 0.7135,
|
| 378 |
+
"step": 370
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 11.428571428571429,
|
| 382 |
+
"grad_norm": 13.441643714904785,
|
| 383 |
+
"learning_rate": 4.915824915824916e-05,
|
| 384 |
+
"loss": 0.6766,
|
| 385 |
+
"step": 380
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 11.729323308270677,
|
| 389 |
+
"grad_norm": 12.908476829528809,
|
| 390 |
+
"learning_rate": 4.898989898989899e-05,
|
| 391 |
+
"loss": 0.7865,
|
| 392 |
+
"step": 390
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 12.0,
|
| 396 |
+
"eval_accuracy": 0.55,
|
| 397 |
+
"eval_f1": 0.47922500814067076,
|
| 398 |
+
"eval_loss": 0.6605614423751831,
|
| 399 |
+
"eval_runtime": 6.4866,
|
| 400 |
+
"eval_samples_per_second": 9.25,
|
| 401 |
+
"eval_steps_per_second": 2.312,
|
| 402 |
+
"step": 399
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 12.030075187969924,
|
| 406 |
+
"grad_norm": 8.809733390808105,
|
| 407 |
+
"learning_rate": 4.882154882154882e-05,
|
| 408 |
+
"loss": 0.7082,
|
| 409 |
+
"step": 400
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 12.330827067669173,
|
| 413 |
+
"grad_norm": 8.784128189086914,
|
| 414 |
+
"learning_rate": 4.865319865319866e-05,
|
| 415 |
+
"loss": 0.656,
|
| 416 |
+
"step": 410
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 12.631578947368421,
|
| 420 |
+
"grad_norm": 5.786237716674805,
|
| 421 |
+
"learning_rate": 4.848484848484849e-05,
|
| 422 |
+
"loss": 0.6632,
|
| 423 |
+
"step": 420
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 12.93233082706767,
|
| 427 |
+
"grad_norm": 35.784297943115234,
|
| 428 |
+
"learning_rate": 4.831649831649832e-05,
|
| 429 |
+
"loss": 0.7191,
|
| 430 |
+
"step": 430
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 12.992481203007518,
|
| 434 |
+
"eval_accuracy": 0.65,
|
| 435 |
+
"eval_f1": 0.6465502675302731,
|
| 436 |
+
"eval_loss": 0.6406596899032593,
|
| 437 |
+
"eval_runtime": 6.6562,
|
| 438 |
+
"eval_samples_per_second": 9.014,
|
| 439 |
+
"eval_steps_per_second": 2.254,
|
| 440 |
+
"step": 432
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 13.233082706766917,
|
| 444 |
+
"grad_norm": 16.550561904907227,
|
| 445 |
+
"learning_rate": 4.814814814814815e-05,
|
| 446 |
+
"loss": 0.6126,
|
| 447 |
+
"step": 440
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 13.533834586466165,
|
| 451 |
+
"grad_norm": 6.941432476043701,
|
| 452 |
+
"learning_rate": 4.797979797979798e-05,
|
| 453 |
+
"loss": 0.6654,
|
| 454 |
+
"step": 450
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 13.834586466165414,
|
| 458 |
+
"grad_norm": 16.613155364990234,
|
| 459 |
+
"learning_rate": 4.7811447811447815e-05,
|
| 460 |
+
"loss": 0.5889,
|
| 461 |
+
"step": 460
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 13.984962406015038,
|
| 465 |
+
"eval_accuracy": 0.48333333333333334,
|
| 466 |
+
"eval_f1": 0.3618599791013584,
|
| 467 |
+
"eval_loss": 0.8008177876472473,
|
| 468 |
+
"eval_runtime": 6.5666,
|
| 469 |
+
"eval_samples_per_second": 9.137,
|
| 470 |
+
"eval_steps_per_second": 2.284,
|
| 471 |
+
"step": 465
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 14.135338345864662,
|
| 475 |
+
"grad_norm": 5.850649833679199,
|
| 476 |
+
"learning_rate": 4.7643097643097646e-05,
|
| 477 |
+
"loss": 0.7016,
|
| 478 |
+
"step": 470
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 14.436090225563909,
|
| 482 |
+
"grad_norm": 3.8176016807556152,
|
| 483 |
+
"learning_rate": 4.7474747474747476e-05,
|
| 484 |
+
"loss": 0.6036,
|
| 485 |
+
"step": 480
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 14.736842105263158,
|
| 489 |
+
"grad_norm": 14.288350105285645,
|
| 490 |
+
"learning_rate": 4.730639730639731e-05,
|
| 491 |
+
"loss": 0.5489,
|
| 492 |
+
"step": 490
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 14.977443609022556,
|
| 496 |
+
"eval_accuracy": 0.5333333333333333,
|
| 497 |
+
"eval_f1": 0.4674368848951647,
|
| 498 |
+
"eval_loss": 0.729806661605835,
|
| 499 |
+
"eval_runtime": 8.5173,
|
| 500 |
+
"eval_samples_per_second": 7.044,
|
| 501 |
+
"eval_steps_per_second": 1.761,
|
| 502 |
+
"step": 498
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 15.037593984962406,
|
| 506 |
+
"grad_norm": 17.23583221435547,
|
| 507 |
+
"learning_rate": 4.713804713804714e-05,
|
| 508 |
+
"loss": 0.6795,
|
| 509 |
+
"step": 500
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 15.338345864661655,
|
| 513 |
+
"grad_norm": 3.6928858757019043,
|
| 514 |
+
"learning_rate": 4.696969696969697e-05,
|
| 515 |
+
"loss": 0.6264,
|
| 516 |
+
"step": 510
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 15.639097744360903,
|
| 520 |
+
"grad_norm": 22.685209274291992,
|
| 521 |
+
"learning_rate": 4.68013468013468e-05,
|
| 522 |
+
"loss": 0.593,
|
| 523 |
+
"step": 520
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 15.93984962406015,
|
| 527 |
+
"grad_norm": 7.016318321228027,
|
| 528 |
+
"learning_rate": 4.6632996632996634e-05,
|
| 529 |
+
"loss": 0.596,
|
| 530 |
+
"step": 530
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 16.0,
|
| 534 |
+
"eval_accuracy": 0.6666666666666666,
|
| 535 |
+
"eval_f1": 0.6591251885369532,
|
| 536 |
+
"eval_loss": 0.746545135974884,
|
| 537 |
+
"eval_runtime": 6.6407,
|
| 538 |
+
"eval_samples_per_second": 9.035,
|
| 539 |
+
"eval_steps_per_second": 2.259,
|
| 540 |
+
"step": 532
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 16.2406015037594,
|
| 544 |
+
"grad_norm": 13.344483375549316,
|
| 545 |
+
"learning_rate": 4.6464646464646464e-05,
|
| 546 |
+
"loss": 0.6249,
|
| 547 |
+
"step": 540
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 16.541353383458645,
|
| 551 |
+
"grad_norm": 10.184150695800781,
|
| 552 |
+
"learning_rate": 4.62962962962963e-05,
|
| 553 |
+
"loss": 0.5373,
|
| 554 |
+
"step": 550
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 16.842105263157894,
|
| 558 |
+
"grad_norm": 16.70218849182129,
|
| 559 |
+
"learning_rate": 4.612794612794613e-05,
|
| 560 |
+
"loss": 0.6136,
|
| 561 |
+
"step": 560
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 16.992481203007518,
|
| 565 |
+
"eval_accuracy": 0.5333333333333333,
|
| 566 |
+
"eval_f1": 0.4691666666666666,
|
| 567 |
+
"eval_loss": 0.9117582440376282,
|
| 568 |
+
"eval_runtime": 6.56,
|
| 569 |
+
"eval_samples_per_second": 9.146,
|
| 570 |
+
"eval_steps_per_second": 2.287,
|
| 571 |
+
"step": 565
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 17.142857142857142,
|
| 575 |
+
"grad_norm": 31.319522857666016,
|
| 576 |
+
"learning_rate": 4.595959595959596e-05,
|
| 577 |
+
"loss": 0.6073,
|
| 578 |
+
"step": 570
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 17.44360902255639,
|
| 582 |
+
"grad_norm": 18.976797103881836,
|
| 583 |
+
"learning_rate": 4.57912457912458e-05,
|
| 584 |
+
"loss": 0.5716,
|
| 585 |
+
"step": 580
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 17.74436090225564,
|
| 589 |
+
"grad_norm": 18.692249298095703,
|
| 590 |
+
"learning_rate": 4.562289562289562e-05,
|
| 591 |
+
"loss": 0.5961,
|
| 592 |
+
"step": 590
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 17.984962406015036,
|
| 596 |
+
"eval_accuracy": 0.65,
|
| 597 |
+
"eval_f1": 0.6297777777777778,
|
| 598 |
+
"eval_loss": 0.690215528011322,
|
| 599 |
+
"eval_runtime": 7.0118,
|
| 600 |
+
"eval_samples_per_second": 8.557,
|
| 601 |
+
"eval_steps_per_second": 2.139,
|
| 602 |
+
"step": 598
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 18.045112781954888,
|
| 606 |
+
"grad_norm": 56.27545928955078,
|
| 607 |
+
"learning_rate": 4.545454545454546e-05,
|
| 608 |
+
"loss": 0.5696,
|
| 609 |
+
"step": 600
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 18.345864661654137,
|
| 613 |
+
"grad_norm": 19.565650939941406,
|
| 614 |
+
"learning_rate": 4.528619528619529e-05,
|
| 615 |
+
"loss": 0.6346,
|
| 616 |
+
"step": 610
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 18.646616541353385,
|
| 620 |
+
"grad_norm": 12.217280387878418,
|
| 621 |
+
"learning_rate": 4.511784511784512e-05,
|
| 622 |
+
"loss": 0.4669,
|
| 623 |
+
"step": 620
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 18.94736842105263,
|
| 627 |
+
"grad_norm": 33.36597442626953,
|
| 628 |
+
"learning_rate": 4.494949494949495e-05,
|
| 629 |
+
"loss": 0.6327,
|
| 630 |
+
"step": 630
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 18.977443609022558,
|
| 634 |
+
"eval_accuracy": 0.5666666666666667,
|
| 635 |
+
"eval_f1": 0.519047619047619,
|
| 636 |
+
"eval_loss": 0.8260101079940796,
|
| 637 |
+
"eval_runtime": 6.5494,
|
| 638 |
+
"eval_samples_per_second": 9.161,
|
| 639 |
+
"eval_steps_per_second": 2.29,
|
| 640 |
+
"step": 631
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 19.24812030075188,
|
| 644 |
+
"grad_norm": 25.23843002319336,
|
| 645 |
+
"learning_rate": 4.4781144781144786e-05,
|
| 646 |
+
"loss": 0.5808,
|
| 647 |
+
"step": 640
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 19.548872180451127,
|
| 651 |
+
"grad_norm": 14.927197456359863,
|
| 652 |
+
"learning_rate": 4.4612794612794616e-05,
|
| 653 |
+
"loss": 0.5018,
|
| 654 |
+
"step": 650
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 19.849624060150376,
|
| 658 |
+
"grad_norm": 16.2906436920166,
|
| 659 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 660 |
+
"loss": 0.6518,
|
| 661 |
+
"step": 660
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 20.0,
|
| 665 |
+
"eval_accuracy": 0.5833333333333334,
|
| 666 |
+
"eval_f1": 0.5714764779151096,
|
| 667 |
+
"eval_loss": 0.6918700933456421,
|
| 668 |
+
"eval_runtime": 6.4139,
|
| 669 |
+
"eval_samples_per_second": 9.355,
|
| 670 |
+
"eval_steps_per_second": 2.339,
|
| 671 |
+
"step": 665
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 20.150375939849624,
|
| 675 |
+
"grad_norm": 11.751864433288574,
|
| 676 |
+
"learning_rate": 4.427609427609428e-05,
|
| 677 |
+
"loss": 0.499,
|
| 678 |
+
"step": 670
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 20.451127819548873,
|
| 682 |
+
"grad_norm": 24.952850341796875,
|
| 683 |
+
"learning_rate": 4.410774410774411e-05,
|
| 684 |
+
"loss": 0.5204,
|
| 685 |
+
"step": 680
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 20.75187969924812,
|
| 689 |
+
"grad_norm": 7.636977195739746,
|
| 690 |
+
"learning_rate": 4.3939393939393944e-05,
|
| 691 |
+
"loss": 0.5551,
|
| 692 |
+
"step": 690
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 20.992481203007518,
|
| 696 |
+
"eval_accuracy": 0.55,
|
| 697 |
+
"eval_f1": 0.516,
|
| 698 |
+
"eval_loss": 1.1780269145965576,
|
| 699 |
+
"eval_runtime": 6.7584,
|
| 700 |
+
"eval_samples_per_second": 8.878,
|
| 701 |
+
"eval_steps_per_second": 2.219,
|
| 702 |
+
"step": 698
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 21.05263157894737,
|
| 706 |
+
"grad_norm": 28.386489868164062,
|
| 707 |
+
"learning_rate": 4.3771043771043774e-05,
|
| 708 |
+
"loss": 0.6306,
|
| 709 |
+
"step": 700
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 21.353383458646615,
|
| 713 |
+
"grad_norm": 9.98483657836914,
|
| 714 |
+
"learning_rate": 4.3602693602693604e-05,
|
| 715 |
+
"loss": 0.443,
|
| 716 |
+
"step": 710
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 21.654135338345863,
|
| 720 |
+
"grad_norm": 15.316218376159668,
|
| 721 |
+
"learning_rate": 4.343434343434344e-05,
|
| 722 |
+
"loss": 0.5148,
|
| 723 |
+
"step": 720
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 21.954887218045112,
|
| 727 |
+
"grad_norm": 17.753332138061523,
|
| 728 |
+
"learning_rate": 4.3265993265993265e-05,
|
| 729 |
+
"loss": 0.511,
|
| 730 |
+
"step": 730
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 21.984962406015036,
|
| 734 |
+
"eval_accuracy": 0.6,
|
| 735 |
+
"eval_f1": 0.6,
|
| 736 |
+
"eval_loss": 0.7414445281028748,
|
| 737 |
+
"eval_runtime": 6.8006,
|
| 738 |
+
"eval_samples_per_second": 8.823,
|
| 739 |
+
"eval_steps_per_second": 2.206,
|
| 740 |
+
"step": 731
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 22.25563909774436,
|
| 744 |
+
"grad_norm": 44.12444305419922,
|
| 745 |
+
"learning_rate": 4.3097643097643095e-05,
|
| 746 |
+
"loss": 0.5268,
|
| 747 |
+
"step": 740
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 22.55639097744361,
|
| 751 |
+
"grad_norm": 25.069778442382812,
|
| 752 |
+
"learning_rate": 4.292929292929293e-05,
|
| 753 |
+
"loss": 0.5273,
|
| 754 |
+
"step": 750
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 22.857142857142858,
|
| 758 |
+
"grad_norm": 9.567957878112793,
|
| 759 |
+
"learning_rate": 4.276094276094276e-05,
|
| 760 |
+
"loss": 0.4749,
|
| 761 |
+
"step": 760
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 22.977443609022558,
|
| 765 |
+
"eval_accuracy": 0.6166666666666667,
|
| 766 |
+
"eval_f1": 0.612888388247442,
|
| 767 |
+
"eval_loss": 0.7978177666664124,
|
| 768 |
+
"eval_runtime": 6.5763,
|
| 769 |
+
"eval_samples_per_second": 9.124,
|
| 770 |
+
"eval_steps_per_second": 2.281,
|
| 771 |
+
"step": 764
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 23.157894736842106,
|
| 775 |
+
"grad_norm": 32.95903396606445,
|
| 776 |
+
"learning_rate": 4.259259259259259e-05,
|
| 777 |
+
"loss": 0.5124,
|
| 778 |
+
"step": 770
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 23.458646616541355,
|
| 782 |
+
"grad_norm": 13.2031831741333,
|
| 783 |
+
"learning_rate": 4.242424242424243e-05,
|
| 784 |
+
"loss": 0.577,
|
| 785 |
+
"step": 780
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 23.7593984962406,
|
| 789 |
+
"grad_norm": 28.219993591308594,
|
| 790 |
+
"learning_rate": 4.225589225589226e-05,
|
| 791 |
+
"loss": 0.4607,
|
| 792 |
+
"step": 790
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 24.0,
|
| 796 |
+
"eval_accuracy": 0.55,
|
| 797 |
+
"eval_f1": 0.5420163334272036,
|
| 798 |
+
"eval_loss": 0.8087051510810852,
|
| 799 |
+
"eval_runtime": 6.5355,
|
| 800 |
+
"eval_samples_per_second": 9.181,
|
| 801 |
+
"eval_steps_per_second": 2.295,
|
| 802 |
+
"step": 798
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 24.06015037593985,
|
| 806 |
+
"grad_norm": 7.048239707946777,
|
| 807 |
+
"learning_rate": 4.208754208754209e-05,
|
| 808 |
+
"loss": 0.4982,
|
| 809 |
+
"step": 800
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 24.360902255639097,
|
| 813 |
+
"grad_norm": 8.705081939697266,
|
| 814 |
+
"learning_rate": 4.191919191919192e-05,
|
| 815 |
+
"loss": 0.4764,
|
| 816 |
+
"step": 810
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 24.661654135338345,
|
| 820 |
+
"grad_norm": 14.514195442199707,
|
| 821 |
+
"learning_rate": 4.175084175084175e-05,
|
| 822 |
+
"loss": 0.4705,
|
| 823 |
+
"step": 820
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 24.962406015037594,
|
| 827 |
+
"grad_norm": 15.823882102966309,
|
| 828 |
+
"learning_rate": 4.158249158249159e-05,
|
| 829 |
+
"loss": 0.5837,
|
| 830 |
+
"step": 830
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 24.992481203007518,
|
| 834 |
+
"eval_accuracy": 0.5666666666666667,
|
| 835 |
+
"eval_f1": 0.5456018518518518,
|
| 836 |
+
"eval_loss": 0.827062726020813,
|
| 837 |
+
"eval_runtime": 6.6423,
|
| 838 |
+
"eval_samples_per_second": 9.033,
|
| 839 |
+
"eval_steps_per_second": 2.258,
|
| 840 |
+
"step": 831
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 25.263157894736842,
|
| 844 |
+
"grad_norm": 40.25634765625,
|
| 845 |
+
"learning_rate": 4.141414141414142e-05,
|
| 846 |
+
"loss": 0.5328,
|
| 847 |
+
"step": 840
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 25.56390977443609,
|
| 851 |
+
"grad_norm": 35.79170227050781,
|
| 852 |
+
"learning_rate": 4.124579124579125e-05,
|
| 853 |
+
"loss": 0.5592,
|
| 854 |
+
"step": 850
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 25.86466165413534,
|
| 858 |
+
"grad_norm": 30.948213577270508,
|
| 859 |
+
"learning_rate": 4.1077441077441085e-05,
|
| 860 |
+
"loss": 0.4608,
|
| 861 |
+
"step": 860
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 25.984962406015036,
|
| 865 |
+
"eval_accuracy": 0.6,
|
| 866 |
+
"eval_f1": 0.5862857142857142,
|
| 867 |
+
"eval_loss": 0.853912353515625,
|
| 868 |
+
"eval_runtime": 7.1127,
|
| 869 |
+
"eval_samples_per_second": 8.436,
|
| 870 |
+
"eval_steps_per_second": 2.109,
|
| 871 |
+
"step": 864
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 26.165413533834588,
|
| 875 |
+
"grad_norm": 16.665449142456055,
|
| 876 |
+
"learning_rate": 4.0909090909090915e-05,
|
| 877 |
+
"loss": 0.4597,
|
| 878 |
+
"step": 870
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 26.466165413533833,
|
| 882 |
+
"grad_norm": 15.06215763092041,
|
| 883 |
+
"learning_rate": 4.074074074074074e-05,
|
| 884 |
+
"loss": 0.3808,
|
| 885 |
+
"step": 880
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 26.76691729323308,
|
| 889 |
+
"grad_norm": 16.880653381347656,
|
| 890 |
+
"learning_rate": 4.0572390572390575e-05,
|
| 891 |
+
"loss": 0.536,
|
| 892 |
+
"step": 890
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 26.977443609022558,
|
| 896 |
+
"eval_accuracy": 0.5333333333333333,
|
| 897 |
+
"eval_f1": 0.5026243634939287,
|
| 898 |
+
"eval_loss": 0.9802255034446716,
|
| 899 |
+
"eval_runtime": 6.3964,
|
| 900 |
+
"eval_samples_per_second": 9.38,
|
| 901 |
+
"eval_steps_per_second": 2.345,
|
| 902 |
+
"step": 897
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 27.06766917293233,
|
| 906 |
+
"grad_norm": 22.354326248168945,
|
| 907 |
+
"learning_rate": 4.0404040404040405e-05,
|
| 908 |
+
"loss": 0.5231,
|
| 909 |
+
"step": 900
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 27.36842105263158,
|
| 913 |
+
"grad_norm": 21.1690731048584,
|
| 914 |
+
"learning_rate": 4.0235690235690236e-05,
|
| 915 |
+
"loss": 0.4904,
|
| 916 |
+
"step": 910
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 27.669172932330827,
|
| 920 |
+
"grad_norm": 45.88932800292969,
|
| 921 |
+
"learning_rate": 4.006734006734007e-05,
|
| 922 |
+
"loss": 0.4966,
|
| 923 |
+
"step": 920
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 27.969924812030076,
|
| 927 |
+
"grad_norm": 16.252901077270508,
|
| 928 |
+
"learning_rate": 3.98989898989899e-05,
|
| 929 |
+
"loss": 0.4225,
|
| 930 |
+
"step": 930
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 28.0,
|
| 934 |
+
"eval_accuracy": 0.6,
|
| 935 |
+
"eval_f1": 0.5909502262443439,
|
| 936 |
+
"eval_loss": 0.9274828433990479,
|
| 937 |
+
"eval_runtime": 6.4885,
|
| 938 |
+
"eval_samples_per_second": 9.247,
|
| 939 |
+
"eval_steps_per_second": 2.312,
|
| 940 |
+
"step": 931
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 28.270676691729324,
|
| 944 |
+
"grad_norm": 24.86367416381836,
|
| 945 |
+
"learning_rate": 3.973063973063973e-05,
|
| 946 |
+
"loss": 0.4523,
|
| 947 |
+
"step": 940
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 28.571428571428573,
|
| 951 |
+
"grad_norm": 59.121612548828125,
|
| 952 |
+
"learning_rate": 3.956228956228956e-05,
|
| 953 |
+
"loss": 0.4359,
|
| 954 |
+
"step": 950
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 28.872180451127818,
|
| 958 |
+
"grad_norm": 23.139965057373047,
|
| 959 |
+
"learning_rate": 3.939393939393939e-05,
|
| 960 |
+
"loss": 0.4325,
|
| 961 |
+
"step": 960
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 28.992481203007518,
|
| 965 |
+
"eval_accuracy": 0.6166666666666667,
|
| 966 |
+
"eval_f1": 0.6098657655120623,
|
| 967 |
+
"eval_loss": 0.8834265470504761,
|
| 968 |
+
"eval_runtime": 6.8775,
|
| 969 |
+
"eval_samples_per_second": 8.724,
|
| 970 |
+
"eval_steps_per_second": 2.181,
|
| 971 |
+
"step": 964
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 29.172932330827066,
|
| 975 |
+
"grad_norm": 28.827287673950195,
|
| 976 |
+
"learning_rate": 3.9225589225589224e-05,
|
| 977 |
+
"loss": 0.4721,
|
| 978 |
+
"step": 970
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 29.473684210526315,
|
| 982 |
+
"grad_norm": 16.462913513183594,
|
| 983 |
+
"learning_rate": 3.905723905723906e-05,
|
| 984 |
+
"loss": 0.4941,
|
| 985 |
+
"step": 980
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 29.774436090225564,
|
| 989 |
+
"grad_norm": 21.16120719909668,
|
| 990 |
+
"learning_rate": 3.888888888888889e-05,
|
| 991 |
+
"loss": 0.4874,
|
| 992 |
+
"step": 990
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 29.984962406015036,
|
| 996 |
+
"eval_accuracy": 0.6166666666666667,
|
| 997 |
+
"eval_f1": 0.616773177734556,
|
| 998 |
+
"eval_loss": 0.8720671534538269,
|
| 999 |
+
"eval_runtime": 6.6224,
|
| 1000 |
+
"eval_samples_per_second": 9.06,
|
| 1001 |
+
"eval_steps_per_second": 2.265,
|
| 1002 |
+
"step": 997
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 30.075187969924812,
|
| 1006 |
+
"grad_norm": 14.207987785339355,
|
| 1007 |
+
"learning_rate": 3.872053872053872e-05,
|
| 1008 |
+
"loss": 0.5007,
|
| 1009 |
+
"step": 1000
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 30.37593984962406,
|
| 1013 |
+
"grad_norm": 6.5519537925720215,
|
| 1014 |
+
"learning_rate": 3.855218855218856e-05,
|
| 1015 |
+
"loss": 0.4059,
|
| 1016 |
+
"step": 1010
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 30.67669172932331,
|
| 1020 |
+
"grad_norm": 8.17160415649414,
|
| 1021 |
+
"learning_rate": 3.838383838383838e-05,
|
| 1022 |
+
"loss": 0.3861,
|
| 1023 |
+
"step": 1020
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 30.977443609022558,
|
| 1027 |
+
"grad_norm": 19.54429817199707,
|
| 1028 |
+
"learning_rate": 3.821548821548822e-05,
|
| 1029 |
+
"loss": 0.4165,
|
| 1030 |
+
"step": 1030
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 30.977443609022558,
|
| 1034 |
+
"eval_accuracy": 0.6166666666666667,
|
| 1035 |
+
"eval_f1": 0.616347133462999,
|
| 1036 |
+
"eval_loss": 1.0359536409378052,
|
| 1037 |
+
"eval_runtime": 6.5574,
|
| 1038 |
+
"eval_samples_per_second": 9.15,
|
| 1039 |
+
"eval_steps_per_second": 2.287,
|
| 1040 |
+
"step": 1030
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 31.278195488721803,
|
| 1044 |
+
"grad_norm": 23.49751091003418,
|
| 1045 |
+
"learning_rate": 3.804713804713805e-05,
|
| 1046 |
+
"loss": 0.4572,
|
| 1047 |
+
"step": 1040
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 31.57894736842105,
|
| 1051 |
+
"grad_norm": 9.040291786193848,
|
| 1052 |
+
"learning_rate": 3.787878787878788e-05,
|
| 1053 |
+
"loss": 0.3131,
|
| 1054 |
+
"step": 1050
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 31.8796992481203,
|
| 1058 |
+
"grad_norm": 7.378245830535889,
|
| 1059 |
+
"learning_rate": 3.7710437710437716e-05,
|
| 1060 |
+
"loss": 0.4773,
|
| 1061 |
+
"step": 1060
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 32.0,
|
| 1065 |
+
"eval_accuracy": 0.5833333333333334,
|
| 1066 |
+
"eval_f1": 0.575941049469633,
|
| 1067 |
+
"eval_loss": 1.2209875583648682,
|
| 1068 |
+
"eval_runtime": 6.5797,
|
| 1069 |
+
"eval_samples_per_second": 9.119,
|
| 1070 |
+
"eval_steps_per_second": 2.28,
|
| 1071 |
+
"step": 1064
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 32.18045112781955,
|
| 1075 |
+
"grad_norm": 15.682769775390625,
|
| 1076 |
+
"learning_rate": 3.7542087542087546e-05,
|
| 1077 |
+
"loss": 0.3388,
|
| 1078 |
+
"step": 1070
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"epoch": 32.4812030075188,
|
| 1082 |
+
"grad_norm": 19.28175163269043,
|
| 1083 |
+
"learning_rate": 3.7373737373737376e-05,
|
| 1084 |
+
"loss": 0.3307,
|
| 1085 |
+
"step": 1080
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 32.78195488721804,
|
| 1089 |
+
"grad_norm": 14.588042259216309,
|
| 1090 |
+
"learning_rate": 3.720538720538721e-05,
|
| 1091 |
+
"loss": 0.3756,
|
| 1092 |
+
"step": 1090
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 32.99248120300752,
|
| 1096 |
+
"eval_accuracy": 0.5833333333333334,
|
| 1097 |
+
"eval_f1": 0.5829860146336946,
|
| 1098 |
+
"eval_loss": 1.1291446685791016,
|
| 1099 |
+
"eval_runtime": 6.8244,
|
| 1100 |
+
"eval_samples_per_second": 8.792,
|
| 1101 |
+
"eval_steps_per_second": 2.198,
|
| 1102 |
+
"step": 1097
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 33.08270676691729,
|
| 1106 |
+
"grad_norm": 13.086041450500488,
|
| 1107 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 1108 |
+
"loss": 0.4182,
|
| 1109 |
+
"step": 1100
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 33.38345864661654,
|
| 1113 |
+
"grad_norm": 5.811966419219971,
|
| 1114 |
+
"learning_rate": 3.686868686868687e-05,
|
| 1115 |
+
"loss": 0.3508,
|
| 1116 |
+
"step": 1110
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 33.68421052631579,
|
| 1120 |
+
"grad_norm": 81.32879638671875,
|
| 1121 |
+
"learning_rate": 3.6700336700336704e-05,
|
| 1122 |
+
"loss": 0.3672,
|
| 1123 |
+
"step": 1120
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 33.984962406015036,
|
| 1127 |
+
"grad_norm": 5.003579616546631,
|
| 1128 |
+
"learning_rate": 3.6531986531986534e-05,
|
| 1129 |
+
"loss": 0.636,
|
| 1130 |
+
"step": 1130
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 33.984962406015036,
|
| 1134 |
+
"eval_accuracy": 0.5833333333333334,
|
| 1135 |
+
"eval_f1": 0.5829860146336946,
|
| 1136 |
+
"eval_loss": 1.0177818536758423,
|
| 1137 |
+
"eval_runtime": 6.5487,
|
| 1138 |
+
"eval_samples_per_second": 9.162,
|
| 1139 |
+
"eval_steps_per_second": 2.291,
|
| 1140 |
+
"step": 1130
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 34.285714285714285,
|
| 1144 |
+
"grad_norm": 8.781743049621582,
|
| 1145 |
+
"learning_rate": 3.6363636363636364e-05,
|
| 1146 |
+
"loss": 0.3774,
|
| 1147 |
+
"step": 1140
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 34.58646616541353,
|
| 1151 |
+
"grad_norm": 57.26716232299805,
|
| 1152 |
+
"learning_rate": 3.61952861952862e-05,
|
| 1153 |
+
"loss": 0.4032,
|
| 1154 |
+
"step": 1150
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 34.88721804511278,
|
| 1158 |
+
"grad_norm": 21.142290115356445,
|
| 1159 |
+
"learning_rate": 3.602693602693603e-05,
|
| 1160 |
+
"loss": 0.5474,
|
| 1161 |
+
"step": 1160
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 34.97744360902256,
|
| 1165 |
+
"eval_accuracy": 0.5666666666666667,
|
| 1166 |
+
"eval_f1": 0.5608305274971942,
|
| 1167 |
+
"eval_loss": 0.9478945136070251,
|
| 1168 |
+
"eval_runtime": 6.7769,
|
| 1169 |
+
"eval_samples_per_second": 8.854,
|
| 1170 |
+
"eval_steps_per_second": 2.213,
|
| 1171 |
+
"step": 1163
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 35.18796992481203,
|
| 1175 |
+
"grad_norm": 27.60926055908203,
|
| 1176 |
+
"learning_rate": 3.5858585858585855e-05,
|
| 1177 |
+
"loss": 0.3429,
|
| 1178 |
+
"step": 1170
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 35.48872180451128,
|
| 1182 |
+
"grad_norm": 12.224372863769531,
|
| 1183 |
+
"learning_rate": 3.569023569023569e-05,
|
| 1184 |
+
"loss": 0.4284,
|
| 1185 |
+
"step": 1180
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 35.78947368421053,
|
| 1189 |
+
"grad_norm": 13.833395004272461,
|
| 1190 |
+
"learning_rate": 3.552188552188552e-05,
|
| 1191 |
+
"loss": 0.3462,
|
| 1192 |
+
"step": 1190
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 36.0,
|
| 1196 |
+
"eval_accuracy": 0.6166666666666667,
|
| 1197 |
+
"eval_f1": 0.6163464216095794,
|
| 1198 |
+
"eval_loss": 0.9584738612174988,
|
| 1199 |
+
"eval_runtime": 6.5533,
|
| 1200 |
+
"eval_samples_per_second": 9.156,
|
| 1201 |
+
"eval_steps_per_second": 2.289,
|
| 1202 |
+
"step": 1197
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 36.090225563909776,
|
| 1206 |
+
"grad_norm": 10.387247085571289,
|
| 1207 |
+
"learning_rate": 3.535353535353535e-05,
|
| 1208 |
+
"loss": 0.2622,
|
| 1209 |
+
"step": 1200
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 36.390977443609025,
|
| 1213 |
+
"grad_norm": 16.11662483215332,
|
| 1214 |
+
"learning_rate": 3.518518518518519e-05,
|
| 1215 |
+
"loss": 0.3035,
|
| 1216 |
+
"step": 1210
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 36.69172932330827,
|
| 1220 |
+
"grad_norm": 60.57283401489258,
|
| 1221 |
+
"learning_rate": 3.501683501683502e-05,
|
| 1222 |
+
"loss": 0.4646,
|
| 1223 |
+
"step": 1220
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 36.99248120300752,
|
| 1227 |
+
"grad_norm": 11.045464515686035,
|
| 1228 |
+
"learning_rate": 3.484848484848485e-05,
|
| 1229 |
+
"loss": 0.3057,
|
| 1230 |
+
"step": 1230
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 36.99248120300752,
|
| 1234 |
+
"eval_accuracy": 0.6166666666666667,
|
| 1235 |
+
"eval_f1": 0.6163464216095794,
|
| 1236 |
+
"eval_loss": 1.2013615369796753,
|
| 1237 |
+
"eval_runtime": 7.009,
|
| 1238 |
+
"eval_samples_per_second": 8.56,
|
| 1239 |
+
"eval_steps_per_second": 2.14,
|
| 1240 |
+
"step": 1230
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 37.29323308270677,
|
| 1244 |
+
"grad_norm": 20.90575408935547,
|
| 1245 |
+
"learning_rate": 3.468013468013468e-05,
|
| 1246 |
+
"loss": 0.3392,
|
| 1247 |
+
"step": 1240
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 37.59398496240601,
|
| 1251 |
+
"grad_norm": 5.260161399841309,
|
| 1252 |
+
"learning_rate": 3.451178451178451e-05,
|
| 1253 |
+
"loss": 0.2822,
|
| 1254 |
+
"step": 1250
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 37.89473684210526,
|
| 1258 |
+
"grad_norm": 10.245043754577637,
|
| 1259 |
+
"learning_rate": 3.434343434343435e-05,
|
| 1260 |
+
"loss": 0.2304,
|
| 1261 |
+
"step": 1260
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 37.984962406015036,
|
| 1265 |
+
"eval_accuracy": 0.6333333333333333,
|
| 1266 |
+
"eval_f1": 0.6333333333333333,
|
| 1267 |
+
"eval_loss": 1.1975445747375488,
|
| 1268 |
+
"eval_runtime": 6.8757,
|
| 1269 |
+
"eval_samples_per_second": 8.726,
|
| 1270 |
+
"eval_steps_per_second": 2.182,
|
| 1271 |
+
"step": 1263
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 38.19548872180451,
|
| 1275 |
+
"grad_norm": 5.8279242515563965,
|
| 1276 |
+
"learning_rate": 3.417508417508418e-05,
|
| 1277 |
+
"loss": 0.3548,
|
| 1278 |
+
"step": 1270
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 38.49624060150376,
|
| 1282 |
+
"grad_norm": 30.216154098510742,
|
| 1283 |
+
"learning_rate": 3.400673400673401e-05,
|
| 1284 |
+
"loss": 0.246,
|
| 1285 |
+
"step": 1280
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 38.796992481203006,
|
| 1289 |
+
"grad_norm": 18.291364669799805,
|
| 1290 |
+
"learning_rate": 3.3838383838383844e-05,
|
| 1291 |
+
"loss": 0.2628,
|
| 1292 |
+
"step": 1290
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 38.97744360902256,
|
| 1296 |
+
"eval_accuracy": 0.5833333333333334,
|
| 1297 |
+
"eval_f1": 0.5792540792540792,
|
| 1298 |
+
"eval_loss": 1.5224032402038574,
|
| 1299 |
+
"eval_runtime": 6.6333,
|
| 1300 |
+
"eval_samples_per_second": 9.045,
|
| 1301 |
+
"eval_steps_per_second": 2.261,
|
| 1302 |
+
"step": 1296
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 39.097744360902254,
|
| 1306 |
+
"grad_norm": 29.24486541748047,
|
| 1307 |
+
"learning_rate": 3.3670033670033675e-05,
|
| 1308 |
+
"loss": 0.2907,
|
| 1309 |
+
"step": 1300
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 39.3984962406015,
|
| 1313 |
+
"grad_norm": 10.511560440063477,
|
| 1314 |
+
"learning_rate": 3.35016835016835e-05,
|
| 1315 |
+
"loss": 0.3166,
|
| 1316 |
+
"step": 1310
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 39.69924812030075,
|
| 1320 |
+
"grad_norm": 9.040928840637207,
|
| 1321 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 1322 |
+
"loss": 0.2413,
|
| 1323 |
+
"step": 1320
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 40.0,
|
| 1327 |
+
"grad_norm": 10.92233657836914,
|
| 1328 |
+
"learning_rate": 3.3164983164983165e-05,
|
| 1329 |
+
"loss": 0.3774,
|
| 1330 |
+
"step": 1330
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 40.0,
|
| 1334 |
+
"eval_accuracy": 0.5666666666666667,
|
| 1335 |
+
"eval_f1": 0.5516203703703704,
|
| 1336 |
+
"eval_loss": 1.2903040647506714,
|
| 1337 |
+
"eval_runtime": 6.6687,
|
| 1338 |
+
"eval_samples_per_second": 8.997,
|
| 1339 |
+
"eval_steps_per_second": 2.249,
|
| 1340 |
+
"step": 1330
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 40.30075187969925,
|
| 1344 |
+
"grad_norm": 14.264669418334961,
|
| 1345 |
+
"learning_rate": 3.2996632996632995e-05,
|
| 1346 |
+
"loss": 0.3152,
|
| 1347 |
+
"step": 1340
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 40.6015037593985,
|
| 1351 |
+
"grad_norm": 21.567790985107422,
|
| 1352 |
+
"learning_rate": 3.282828282828283e-05,
|
| 1353 |
+
"loss": 0.3674,
|
| 1354 |
+
"step": 1350
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 40.902255639097746,
|
| 1358 |
+
"grad_norm": 42.34912109375,
|
| 1359 |
+
"learning_rate": 3.265993265993266e-05,
|
| 1360 |
+
"loss": 0.2604,
|
| 1361 |
+
"step": 1360
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 40.99248120300752,
|
| 1365 |
+
"eval_accuracy": 0.5666666666666667,
|
| 1366 |
+
"eval_f1": 0.5608305274971942,
|
| 1367 |
+
"eval_loss": 1.408166527748108,
|
| 1368 |
+
"eval_runtime": 6.9727,
|
| 1369 |
+
"eval_samples_per_second": 8.605,
|
| 1370 |
+
"eval_steps_per_second": 2.151,
|
| 1371 |
+
"step": 1363
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 41.203007518796994,
|
| 1375 |
+
"grad_norm": 10.318882942199707,
|
| 1376 |
+
"learning_rate": 3.249158249158249e-05,
|
| 1377 |
+
"loss": 0.2682,
|
| 1378 |
+
"step": 1370
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 41.50375939849624,
|
| 1382 |
+
"grad_norm": 12.875001907348633,
|
| 1383 |
+
"learning_rate": 3.232323232323233e-05,
|
| 1384 |
+
"loss": 0.2848,
|
| 1385 |
+
"step": 1380
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 41.80451127819549,
|
| 1389 |
+
"grad_norm": 50.29880905151367,
|
| 1390 |
+
"learning_rate": 3.215488215488215e-05,
|
| 1391 |
+
"loss": 0.2522,
|
| 1392 |
+
"step": 1390
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 41.984962406015036,
|
| 1396 |
+
"eval_accuracy": 0.6166666666666667,
|
| 1397 |
+
"eval_f1": 0.616347133462999,
|
| 1398 |
+
"eval_loss": 1.178326964378357,
|
| 1399 |
+
"eval_runtime": 6.8032,
|
| 1400 |
+
"eval_samples_per_second": 8.819,
|
| 1401 |
+
"eval_steps_per_second": 2.205,
|
| 1402 |
+
"step": 1396
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 42.10526315789474,
|
| 1406 |
+
"grad_norm": 47.25193405151367,
|
| 1407 |
+
"learning_rate": 3.198653198653199e-05,
|
| 1408 |
+
"loss": 0.4457,
|
| 1409 |
+
"step": 1400
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 42.40601503759399,
|
| 1413 |
+
"grad_norm": 12.840446472167969,
|
| 1414 |
+
"learning_rate": 3.181818181818182e-05,
|
| 1415 |
+
"loss": 0.2781,
|
| 1416 |
+
"step": 1410
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 42.70676691729323,
|
| 1420 |
+
"grad_norm": 12.884856224060059,
|
| 1421 |
+
"learning_rate": 3.164983164983165e-05,
|
| 1422 |
+
"loss": 0.1925,
|
| 1423 |
+
"step": 1420
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 42.97744360902256,
|
| 1427 |
+
"eval_accuracy": 0.6166666666666667,
|
| 1428 |
+
"eval_f1": 0.616347133462999,
|
| 1429 |
+
"eval_loss": 1.3612515926361084,
|
| 1430 |
+
"eval_runtime": 6.6918,
|
| 1431 |
+
"eval_samples_per_second": 8.966,
|
| 1432 |
+
"eval_steps_per_second": 2.242,
|
| 1433 |
+
"step": 1429
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 43.00751879699248,
|
| 1437 |
+
"grad_norm": 24.348125457763672,
|
| 1438 |
+
"learning_rate": 3.148148148148148e-05,
|
| 1439 |
+
"loss": 0.2576,
|
| 1440 |
+
"step": 1430
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 43.30827067669173,
|
| 1444 |
+
"grad_norm": 15.819385528564453,
|
| 1445 |
+
"learning_rate": 3.131313131313132e-05,
|
| 1446 |
+
"loss": 0.2574,
|
| 1447 |
+
"step": 1440
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 43.609022556390975,
|
| 1451 |
+
"grad_norm": 7.910970211029053,
|
| 1452 |
+
"learning_rate": 3.114478114478115e-05,
|
| 1453 |
+
"loss": 0.2357,
|
| 1454 |
+
"step": 1450
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 43.909774436090224,
|
| 1458 |
+
"grad_norm": 20.98566436767578,
|
| 1459 |
+
"learning_rate": 3.097643097643098e-05,
|
| 1460 |
+
"loss": 0.3436,
|
| 1461 |
+
"step": 1460
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 44.0,
|
| 1465 |
+
"eval_accuracy": 0.5333333333333333,
|
| 1466 |
+
"eval_f1": 0.5173333333333333,
|
| 1467 |
+
"eval_loss": 1.6382521390914917,
|
| 1468 |
+
"eval_runtime": 6.4628,
|
| 1469 |
+
"eval_samples_per_second": 9.284,
|
| 1470 |
+
"eval_steps_per_second": 2.321,
|
| 1471 |
+
"step": 1463
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 44.21052631578947,
|
| 1475 |
+
"grad_norm": 15.531021118164062,
|
| 1476 |
+
"learning_rate": 3.080808080808081e-05,
|
| 1477 |
+
"loss": 0.2849,
|
| 1478 |
+
"step": 1470
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 44.51127819548872,
|
| 1482 |
+
"grad_norm": 20.71696662902832,
|
| 1483 |
+
"learning_rate": 3.063973063973064e-05,
|
| 1484 |
+
"loss": 0.274,
|
| 1485 |
+
"step": 1480
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 44.81203007518797,
|
| 1489 |
+
"grad_norm": 20.764507293701172,
|
| 1490 |
+
"learning_rate": 3.0471380471380472e-05,
|
| 1491 |
+
"loss": 0.1955,
|
| 1492 |
+
"step": 1490
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 44.99248120300752,
|
| 1496 |
+
"eval_accuracy": 0.5,
|
| 1497 |
+
"eval_f1": 0.4828571428571428,
|
| 1498 |
+
"eval_loss": 1.8947157859802246,
|
| 1499 |
+
"eval_runtime": 6.9171,
|
| 1500 |
+
"eval_samples_per_second": 8.674,
|
| 1501 |
+
"eval_steps_per_second": 2.169,
|
| 1502 |
+
"step": 1496
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 45.11278195488722,
|
| 1506 |
+
"grad_norm": 22.713884353637695,
|
| 1507 |
+
"learning_rate": 3.0303030303030306e-05,
|
| 1508 |
+
"loss": 0.2448,
|
| 1509 |
+
"step": 1500
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 45.41353383458647,
|
| 1513 |
+
"grad_norm": 10.142008781433105,
|
| 1514 |
+
"learning_rate": 3.013468013468014e-05,
|
| 1515 |
+
"loss": 0.1694,
|
| 1516 |
+
"step": 1510
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 45.714285714285715,
|
| 1520 |
+
"grad_norm": 24.03466796875,
|
| 1521 |
+
"learning_rate": 2.996632996632997e-05,
|
| 1522 |
+
"loss": 0.2206,
|
| 1523 |
+
"step": 1520
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 45.984962406015036,
|
| 1527 |
+
"eval_accuracy": 0.6,
|
| 1528 |
+
"eval_f1": 0.6,
|
| 1529 |
+
"eval_loss": 1.4390347003936768,
|
| 1530 |
+
"eval_runtime": 6.8263,
|
| 1531 |
+
"eval_samples_per_second": 8.79,
|
| 1532 |
+
"eval_steps_per_second": 2.197,
|
| 1533 |
+
"step": 1529
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 46.015037593984964,
|
| 1537 |
+
"grad_norm": 13.955698013305664,
|
| 1538 |
+
"learning_rate": 2.9797979797979796e-05,
|
| 1539 |
+
"loss": 0.2549,
|
| 1540 |
+
"step": 1530
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 46.31578947368421,
|
| 1544 |
+
"grad_norm": 11.930123329162598,
|
| 1545 |
+
"learning_rate": 2.962962962962963e-05,
|
| 1546 |
+
"loss": 0.2461,
|
| 1547 |
+
"step": 1540
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 46.61654135338346,
|
| 1551 |
+
"grad_norm": 12.452401161193848,
|
| 1552 |
+
"learning_rate": 2.946127946127946e-05,
|
| 1553 |
+
"loss": 0.1838,
|
| 1554 |
+
"step": 1550
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 46.91729323308271,
|
| 1558 |
+
"grad_norm": 6.408818244934082,
|
| 1559 |
+
"learning_rate": 2.9292929292929294e-05,
|
| 1560 |
+
"loss": 0.1912,
|
| 1561 |
+
"step": 1560
|
| 1562 |
+
},
|
| 1563 |
+
{
|
| 1564 |
+
"epoch": 46.97744360902256,
|
| 1565 |
+
"eval_accuracy": 0.65,
|
| 1566 |
+
"eval_f1": 0.6400402414486922,
|
| 1567 |
+
"eval_loss": 1.5288487672805786,
|
| 1568 |
+
"eval_runtime": 6.7766,
|
| 1569 |
+
"eval_samples_per_second": 8.854,
|
| 1570 |
+
"eval_steps_per_second": 2.213,
|
| 1571 |
+
"step": 1562
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 47.21804511278196,
|
| 1575 |
+
"grad_norm": 1.8607676029205322,
|
| 1576 |
+
"learning_rate": 2.9124579124579127e-05,
|
| 1577 |
+
"loss": 0.2388,
|
| 1578 |
+
"step": 1570
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 47.5187969924812,
|
| 1582 |
+
"grad_norm": 47.768798828125,
|
| 1583 |
+
"learning_rate": 2.8956228956228958e-05,
|
| 1584 |
+
"loss": 0.2501,
|
| 1585 |
+
"step": 1580
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 47.81954887218045,
|
| 1589 |
+
"grad_norm": 48.893470764160156,
|
| 1590 |
+
"learning_rate": 2.878787878787879e-05,
|
| 1591 |
+
"loss": 0.2794,
|
| 1592 |
+
"step": 1590
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 48.0,
|
| 1596 |
+
"eval_accuracy": 0.55,
|
| 1597 |
+
"eval_f1": 0.5420163334272036,
|
| 1598 |
+
"eval_loss": 1.7392734289169312,
|
| 1599 |
+
"eval_runtime": 6.7034,
|
| 1600 |
+
"eval_samples_per_second": 8.951,
|
| 1601 |
+
"eval_steps_per_second": 2.238,
|
| 1602 |
+
"step": 1596
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 48.1203007518797,
|
| 1606 |
+
"grad_norm": 3.5548160076141357,
|
| 1607 |
+
"learning_rate": 2.8619528619528618e-05,
|
| 1608 |
+
"loss": 0.2652,
|
| 1609 |
+
"step": 1600
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 48.421052631578945,
|
| 1613 |
+
"grad_norm": 19.021543502807617,
|
| 1614 |
+
"learning_rate": 2.845117845117845e-05,
|
| 1615 |
+
"loss": 0.0985,
|
| 1616 |
+
"step": 1610
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 48.721804511278194,
|
| 1620 |
+
"grad_norm": 33.23078918457031,
|
| 1621 |
+
"learning_rate": 2.8282828282828282e-05,
|
| 1622 |
+
"loss": 0.3166,
|
| 1623 |
+
"step": 1620
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 48.99248120300752,
|
| 1627 |
+
"eval_accuracy": 0.5666666666666667,
|
| 1628 |
+
"eval_f1": 0.5608305274971942,
|
| 1629 |
+
"eval_loss": 2.041395664215088,
|
| 1630 |
+
"eval_runtime": 6.6771,
|
| 1631 |
+
"eval_samples_per_second": 8.986,
|
| 1632 |
+
"eval_steps_per_second": 2.246,
|
| 1633 |
+
"step": 1629
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 49.02255639097744,
|
| 1637 |
+
"grad_norm": 24.95660972595215,
|
| 1638 |
+
"learning_rate": 2.8114478114478115e-05,
|
| 1639 |
+
"loss": 0.2412,
|
| 1640 |
+
"step": 1630
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 49.32330827067669,
|
| 1644 |
+
"grad_norm": 19.062162399291992,
|
| 1645 |
+
"learning_rate": 2.794612794612795e-05,
|
| 1646 |
+
"loss": 0.2181,
|
| 1647 |
+
"step": 1640
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 49.62406015037594,
|
| 1651 |
+
"grad_norm": 5.83784294128418,
|
| 1652 |
+
"learning_rate": 2.777777777777778e-05,
|
| 1653 |
+
"loss": 0.2117,
|
| 1654 |
+
"step": 1650
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 49.92481203007519,
|
| 1658 |
+
"grad_norm": 22.206602096557617,
|
| 1659 |
+
"learning_rate": 2.7609427609427613e-05,
|
| 1660 |
+
"loss": 0.173,
|
| 1661 |
+
"step": 1660
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 49.984962406015036,
|
| 1665 |
+
"eval_accuracy": 0.6,
|
| 1666 |
+
"eval_f1": 0.5991071428571428,
|
| 1667 |
+
"eval_loss": 1.6376549005508423,
|
| 1668 |
+
"eval_runtime": 6.909,
|
| 1669 |
+
"eval_samples_per_second": 8.684,
|
| 1670 |
+
"eval_steps_per_second": 2.171,
|
| 1671 |
+
"step": 1662
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 50.225563909774436,
|
| 1675 |
+
"grad_norm": 12.871647834777832,
|
| 1676 |
+
"learning_rate": 2.7441077441077446e-05,
|
| 1677 |
+
"loss": 0.194,
|
| 1678 |
+
"step": 1670
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 50.526315789473685,
|
| 1682 |
+
"grad_norm": 27.346282958984375,
|
| 1683 |
+
"learning_rate": 2.7272727272727273e-05,
|
| 1684 |
+
"loss": 0.1649,
|
| 1685 |
+
"step": 1680
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 50.82706766917293,
|
| 1689 |
+
"grad_norm": 9.746476173400879,
|
| 1690 |
+
"learning_rate": 2.7104377104377103e-05,
|
| 1691 |
+
"loss": 0.1375,
|
| 1692 |
+
"step": 1690
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 50.97744360902256,
|
| 1696 |
+
"eval_accuracy": 0.6,
|
| 1697 |
+
"eval_f1": 0.6,
|
| 1698 |
+
"eval_loss": 1.622830867767334,
|
| 1699 |
+
"eval_runtime": 6.6844,
|
| 1700 |
+
"eval_samples_per_second": 8.976,
|
| 1701 |
+
"eval_steps_per_second": 2.244,
|
| 1702 |
+
"step": 1695
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 51.12781954887218,
|
| 1706 |
+
"grad_norm": 12.960921287536621,
|
| 1707 |
+
"learning_rate": 2.6936026936026937e-05,
|
| 1708 |
+
"loss": 0.2868,
|
| 1709 |
+
"step": 1700
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 51.42857142857143,
|
| 1713 |
+
"grad_norm": 10.291200637817383,
|
| 1714 |
+
"learning_rate": 2.676767676767677e-05,
|
| 1715 |
+
"loss": 0.1947,
|
| 1716 |
+
"step": 1710
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 51.72932330827068,
|
| 1720 |
+
"grad_norm": 39.014408111572266,
|
| 1721 |
+
"learning_rate": 2.65993265993266e-05,
|
| 1722 |
+
"loss": 0.2659,
|
| 1723 |
+
"step": 1720
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 52.0,
|
| 1727 |
+
"eval_accuracy": 0.6333333333333333,
|
| 1728 |
+
"eval_f1": 0.6333333333333333,
|
| 1729 |
+
"eval_loss": 1.6452404260635376,
|
| 1730 |
+
"eval_runtime": 6.7167,
|
| 1731 |
+
"eval_samples_per_second": 8.933,
|
| 1732 |
+
"eval_steps_per_second": 2.233,
|
| 1733 |
+
"step": 1729
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 52.03007518796993,
|
| 1737 |
+
"grad_norm": 25.84610366821289,
|
| 1738 |
+
"learning_rate": 2.6430976430976434e-05,
|
| 1739 |
+
"loss": 0.1805,
|
| 1740 |
+
"step": 1730
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 52.330827067669176,
|
| 1744 |
+
"grad_norm": 20.498271942138672,
|
| 1745 |
+
"learning_rate": 2.6262626262626268e-05,
|
| 1746 |
+
"loss": 0.2289,
|
| 1747 |
+
"step": 1740
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 52.63157894736842,
|
| 1751 |
+
"grad_norm": 13.4515962600708,
|
| 1752 |
+
"learning_rate": 2.6094276094276095e-05,
|
| 1753 |
+
"loss": 0.0952,
|
| 1754 |
+
"step": 1750
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 52.932330827067666,
|
| 1758 |
+
"grad_norm": 27.01753044128418,
|
| 1759 |
+
"learning_rate": 2.5925925925925925e-05,
|
| 1760 |
+
"loss": 0.2045,
|
| 1761 |
+
"step": 1760
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 52.99248120300752,
|
| 1765 |
+
"eval_accuracy": 0.5666666666666667,
|
| 1766 |
+
"eval_f1": 0.5608305274971942,
|
| 1767 |
+
"eval_loss": 1.970557451248169,
|
| 1768 |
+
"eval_runtime": 6.8136,
|
| 1769 |
+
"eval_samples_per_second": 8.806,
|
| 1770 |
+
"eval_steps_per_second": 2.201,
|
| 1771 |
+
"step": 1762
|
| 1772 |
+
},
|
| 1773 |
+
{
|
| 1774 |
+
"epoch": 53.233082706766915,
|
| 1775 |
+
"grad_norm": 53.2229118347168,
|
| 1776 |
+
"learning_rate": 2.575757575757576e-05,
|
| 1777 |
+
"loss": 0.2113,
|
| 1778 |
+
"step": 1770
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 53.53383458646616,
|
| 1782 |
+
"grad_norm": 1.9475269317626953,
|
| 1783 |
+
"learning_rate": 2.5589225589225592e-05,
|
| 1784 |
+
"loss": 0.159,
|
| 1785 |
+
"step": 1780
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 53.83458646616541,
|
| 1789 |
+
"grad_norm": 50.679840087890625,
|
| 1790 |
+
"learning_rate": 2.5420875420875422e-05,
|
| 1791 |
+
"loss": 0.1081,
|
| 1792 |
+
"step": 1790
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"epoch": 53.984962406015036,
|
| 1796 |
+
"eval_accuracy": 0.6166666666666667,
|
| 1797 |
+
"eval_f1": 0.6009102232442273,
|
| 1798 |
+
"eval_loss": 1.954608678817749,
|
| 1799 |
+
"eval_runtime": 6.9274,
|
| 1800 |
+
"eval_samples_per_second": 8.661,
|
| 1801 |
+
"eval_steps_per_second": 2.165,
|
| 1802 |
+
"step": 1795
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 54.13533834586466,
|
| 1806 |
+
"grad_norm": 42.00580596923828,
|
| 1807 |
+
"learning_rate": 2.5252525252525256e-05,
|
| 1808 |
+
"loss": 0.2169,
|
| 1809 |
+
"step": 1800
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 54.43609022556391,
|
| 1813 |
+
"grad_norm": 34.89802169799805,
|
| 1814 |
+
"learning_rate": 2.5084175084175086e-05,
|
| 1815 |
+
"loss": 0.1815,
|
| 1816 |
+
"step": 1810
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 54.73684210526316,
|
| 1820 |
+
"grad_norm": 43.92179489135742,
|
| 1821 |
+
"learning_rate": 2.4915824915824916e-05,
|
| 1822 |
+
"loss": 0.1782,
|
| 1823 |
+
"step": 1820
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 54.97744360902256,
|
| 1827 |
+
"eval_accuracy": 0.5666666666666667,
|
| 1828 |
+
"eval_f1": 0.5608305274971942,
|
| 1829 |
+
"eval_loss": 2.1268014907836914,
|
| 1830 |
+
"eval_runtime": 6.6378,
|
| 1831 |
+
"eval_samples_per_second": 9.039,
|
| 1832 |
+
"eval_steps_per_second": 2.26,
|
| 1833 |
+
"step": 1828
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 55.037593984962406,
|
| 1837 |
+
"grad_norm": 23.048227310180664,
|
| 1838 |
+
"learning_rate": 2.474747474747475e-05,
|
| 1839 |
+
"loss": 0.2837,
|
| 1840 |
+
"step": 1830
|
| 1841 |
+
},
|
| 1842 |
+
{
|
| 1843 |
+
"epoch": 55.338345864661655,
|
| 1844 |
+
"grad_norm": 31.097103118896484,
|
| 1845 |
+
"learning_rate": 2.457912457912458e-05,
|
| 1846 |
+
"loss": 0.1967,
|
| 1847 |
+
"step": 1840
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 55.6390977443609,
|
| 1851 |
+
"grad_norm": 31.70062255859375,
|
| 1852 |
+
"learning_rate": 2.441077441077441e-05,
|
| 1853 |
+
"loss": 0.1528,
|
| 1854 |
+
"step": 1850
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 55.93984962406015,
|
| 1858 |
+
"grad_norm": 25.655332565307617,
|
| 1859 |
+
"learning_rate": 2.4242424242424244e-05,
|
| 1860 |
+
"loss": 0.244,
|
| 1861 |
+
"step": 1860
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 56.0,
|
| 1865 |
+
"eval_accuracy": 0.6166666666666667,
|
| 1866 |
+
"eval_f1": 0.6098039215686274,
|
| 1867 |
+
"eval_loss": 1.8301317691802979,
|
| 1868 |
+
"eval_runtime": 6.6758,
|
| 1869 |
+
"eval_samples_per_second": 8.988,
|
| 1870 |
+
"eval_steps_per_second": 2.247,
|
| 1871 |
+
"step": 1862
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 56.2406015037594,
|
| 1875 |
+
"grad_norm": 9.54874038696289,
|
| 1876 |
+
"learning_rate": 2.4074074074074074e-05,
|
| 1877 |
+
"loss": 0.1363,
|
| 1878 |
+
"step": 1870
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 56.54135338345865,
|
| 1882 |
+
"grad_norm": 32.316490173339844,
|
| 1883 |
+
"learning_rate": 2.3905723905723908e-05,
|
| 1884 |
+
"loss": 0.2024,
|
| 1885 |
+
"step": 1880
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 56.8421052631579,
|
| 1889 |
+
"grad_norm": 8.345525741577148,
|
| 1890 |
+
"learning_rate": 2.3737373737373738e-05,
|
| 1891 |
+
"loss": 0.1783,
|
| 1892 |
+
"step": 1890
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 56.99248120300752,
|
| 1896 |
+
"eval_accuracy": 0.5666666666666667,
|
| 1897 |
+
"eval_f1": 0.5070833333333334,
|
| 1898 |
+
"eval_loss": 2.5808091163635254,
|
| 1899 |
+
"eval_runtime": 6.6634,
|
| 1900 |
+
"eval_samples_per_second": 9.004,
|
| 1901 |
+
"eval_steps_per_second": 2.251,
|
| 1902 |
+
"step": 1895
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 57.142857142857146,
|
| 1906 |
+
"grad_norm": 16.978073120117188,
|
| 1907 |
+
"learning_rate": 2.356902356902357e-05,
|
| 1908 |
+
"loss": 0.246,
|
| 1909 |
+
"step": 1900
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 57.443609022556394,
|
| 1913 |
+
"grad_norm": 23.652156829833984,
|
| 1914 |
+
"learning_rate": 2.34006734006734e-05,
|
| 1915 |
+
"loss": 0.1651,
|
| 1916 |
+
"step": 1910
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 57.744360902255636,
|
| 1920 |
+
"grad_norm": 67.35960388183594,
|
| 1921 |
+
"learning_rate": 2.3232323232323232e-05,
|
| 1922 |
+
"loss": 0.2429,
|
| 1923 |
+
"step": 1920
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 57.984962406015036,
|
| 1927 |
+
"eval_accuracy": 0.6166666666666667,
|
| 1928 |
+
"eval_f1": 0.6058823529411764,
|
| 1929 |
+
"eval_loss": 2.1214308738708496,
|
| 1930 |
+
"eval_runtime": 6.6413,
|
| 1931 |
+
"eval_samples_per_second": 9.034,
|
| 1932 |
+
"eval_steps_per_second": 2.259,
|
| 1933 |
+
"step": 1928
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"epoch": 58.045112781954884,
|
| 1937 |
+
"grad_norm": 21.2883243560791,
|
| 1938 |
+
"learning_rate": 2.3063973063973065e-05,
|
| 1939 |
+
"loss": 0.1294,
|
| 1940 |
+
"step": 1930
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 58.34586466165413,
|
| 1944 |
+
"grad_norm": 33.893524169921875,
|
| 1945 |
+
"learning_rate": 2.28956228956229e-05,
|
| 1946 |
+
"loss": 0.1717,
|
| 1947 |
+
"step": 1940
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 58.64661654135338,
|
| 1951 |
+
"grad_norm": 7.831489086151123,
|
| 1952 |
+
"learning_rate": 2.272727272727273e-05,
|
| 1953 |
+
"loss": 0.1625,
|
| 1954 |
+
"step": 1950
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 58.94736842105263,
|
| 1958 |
+
"grad_norm": 36.57512283325195,
|
| 1959 |
+
"learning_rate": 2.255892255892256e-05,
|
| 1960 |
+
"loss": 0.2,
|
| 1961 |
+
"step": 1960
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 58.97744360902256,
|
| 1965 |
+
"eval_accuracy": 0.5666666666666667,
|
| 1966 |
+
"eval_f1": 0.5657026325546904,
|
| 1967 |
+
"eval_loss": 2.2281510829925537,
|
| 1968 |
+
"eval_runtime": 6.9221,
|
| 1969 |
+
"eval_samples_per_second": 8.668,
|
| 1970 |
+
"eval_steps_per_second": 2.167,
|
| 1971 |
+
"step": 1961
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 59.24812030075188,
|
| 1975 |
+
"grad_norm": 41.826107025146484,
|
| 1976 |
+
"learning_rate": 2.2390572390572393e-05,
|
| 1977 |
+
"loss": 0.1346,
|
| 1978 |
+
"step": 1970
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"epoch": 59.54887218045113,
|
| 1982 |
+
"grad_norm": 44.347206115722656,
|
| 1983 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 1984 |
+
"loss": 0.2132,
|
| 1985 |
+
"step": 1980
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 59.849624060150376,
|
| 1989 |
+
"grad_norm": 36.68692398071289,
|
| 1990 |
+
"learning_rate": 2.2053872053872053e-05,
|
| 1991 |
+
"loss": 0.1646,
|
| 1992 |
+
"step": 1990
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 60.0,
|
| 1996 |
+
"eval_accuracy": 0.5833333333333334,
|
| 1997 |
+
"eval_f1": 0.5662067643958992,
|
| 1998 |
+
"eval_loss": 2.327244281768799,
|
| 1999 |
+
"eval_runtime": 6.9668,
|
| 2000 |
+
"eval_samples_per_second": 8.612,
|
| 2001 |
+
"eval_steps_per_second": 2.153,
|
| 2002 |
+
"step": 1995
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 60.150375939849624,
|
| 2006 |
+
"grad_norm": 32.38457107543945,
|
| 2007 |
+
"learning_rate": 2.1885521885521887e-05,
|
| 2008 |
+
"loss": 0.1555,
|
| 2009 |
+
"step": 2000
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 60.45112781954887,
|
| 2013 |
+
"grad_norm": 40.30613708496094,
|
| 2014 |
+
"learning_rate": 2.171717171717172e-05,
|
| 2015 |
+
"loss": 0.1694,
|
| 2016 |
+
"step": 2010
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 60.75187969924812,
|
| 2020 |
+
"grad_norm": 32.216827392578125,
|
| 2021 |
+
"learning_rate": 2.1548821548821547e-05,
|
| 2022 |
+
"loss": 0.1663,
|
| 2023 |
+
"step": 2020
|
| 2024 |
+
},
|
| 2025 |
+
{
|
| 2026 |
+
"epoch": 60.99248120300752,
|
| 2027 |
+
"eval_accuracy": 0.5333333333333333,
|
| 2028 |
+
"eval_f1": 0.5322916666666667,
|
| 2029 |
+
"eval_loss": 2.472301959991455,
|
| 2030 |
+
"eval_runtime": 6.5598,
|
| 2031 |
+
"eval_samples_per_second": 9.147,
|
| 2032 |
+
"eval_steps_per_second": 2.287,
|
| 2033 |
+
"step": 2028
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 61.05263157894737,
|
| 2037 |
+
"grad_norm": 14.56987190246582,
|
| 2038 |
+
"learning_rate": 2.138047138047138e-05,
|
| 2039 |
+
"loss": 0.142,
|
| 2040 |
+
"step": 2030
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 61.35338345864662,
|
| 2044 |
+
"grad_norm": 83.00841522216797,
|
| 2045 |
+
"learning_rate": 2.1212121212121215e-05,
|
| 2046 |
+
"loss": 0.1869,
|
| 2047 |
+
"step": 2040
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 61.65413533834587,
|
| 2051 |
+
"grad_norm": 16.63633918762207,
|
| 2052 |
+
"learning_rate": 2.1043771043771045e-05,
|
| 2053 |
+
"loss": 0.0878,
|
| 2054 |
+
"step": 2050
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 61.954887218045116,
|
| 2058 |
+
"grad_norm": 21.885534286499023,
|
| 2059 |
+
"learning_rate": 2.0875420875420875e-05,
|
| 2060 |
+
"loss": 0.1935,
|
| 2061 |
+
"step": 2060
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 61.984962406015036,
|
| 2065 |
+
"eval_accuracy": 0.6,
|
| 2066 |
+
"eval_f1": 0.5973214285714284,
|
| 2067 |
+
"eval_loss": 2.338430166244507,
|
| 2068 |
+
"eval_runtime": 6.6897,
|
| 2069 |
+
"eval_samples_per_second": 8.969,
|
| 2070 |
+
"eval_steps_per_second": 2.242,
|
| 2071 |
+
"step": 2061
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 62.255639097744364,
|
| 2075 |
+
"grad_norm": 2.885319471359253,
|
| 2076 |
+
"learning_rate": 2.070707070707071e-05,
|
| 2077 |
+
"loss": 0.1802,
|
| 2078 |
+
"step": 2070
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 62.556390977443606,
|
| 2082 |
+
"grad_norm": 13.658590316772461,
|
| 2083 |
+
"learning_rate": 2.0538720538720542e-05,
|
| 2084 |
+
"loss": 0.1909,
|
| 2085 |
+
"step": 2080
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 62.857142857142854,
|
| 2089 |
+
"grad_norm": 12.268779754638672,
|
| 2090 |
+
"learning_rate": 2.037037037037037e-05,
|
| 2091 |
+
"loss": 0.2079,
|
| 2092 |
+
"step": 2090
|
| 2093 |
+
},
|
| 2094 |
+
{
|
| 2095 |
+
"epoch": 62.97744360902256,
|
| 2096 |
+
"eval_accuracy": 0.5833333333333334,
|
| 2097 |
+
"eval_f1": 0.5829860146336946,
|
| 2098 |
+
"eval_loss": 1.9270540475845337,
|
| 2099 |
+
"eval_runtime": 6.7469,
|
| 2100 |
+
"eval_samples_per_second": 8.893,
|
| 2101 |
+
"eval_steps_per_second": 2.223,
|
| 2102 |
+
"step": 2094
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 63.1578947368421,
|
| 2106 |
+
"grad_norm": 27.859628677368164,
|
| 2107 |
+
"learning_rate": 2.0202020202020203e-05,
|
| 2108 |
+
"loss": 0.1491,
|
| 2109 |
+
"step": 2100
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 63.45864661654135,
|
| 2113 |
+
"grad_norm": 22.086532592773438,
|
| 2114 |
+
"learning_rate": 2.0033670033670036e-05,
|
| 2115 |
+
"loss": 0.1178,
|
| 2116 |
+
"step": 2110
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 63.7593984962406,
|
| 2120 |
+
"grad_norm": 12.724394798278809,
|
| 2121 |
+
"learning_rate": 1.9865319865319866e-05,
|
| 2122 |
+
"loss": 0.1797,
|
| 2123 |
+
"step": 2120
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 64.0,
|
| 2127 |
+
"eval_accuracy": 0.6166666666666667,
|
| 2128 |
+
"eval_f1": 0.6150582750582749,
|
| 2129 |
+
"eval_loss": 1.870719313621521,
|
| 2130 |
+
"eval_runtime": 6.5187,
|
| 2131 |
+
"eval_samples_per_second": 9.204,
|
| 2132 |
+
"eval_steps_per_second": 2.301,
|
| 2133 |
+
"step": 2128
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 64.06015037593986,
|
| 2137 |
+
"grad_norm": 2.5529205799102783,
|
| 2138 |
+
"learning_rate": 1.9696969696969697e-05,
|
| 2139 |
+
"loss": 0.2255,
|
| 2140 |
+
"step": 2130
|
| 2141 |
+
},
|
| 2142 |
+
{
|
| 2143 |
+
"epoch": 64.3609022556391,
|
| 2144 |
+
"grad_norm": 46.02068328857422,
|
| 2145 |
+
"learning_rate": 1.952861952861953e-05,
|
| 2146 |
+
"loss": 0.1442,
|
| 2147 |
+
"step": 2140
|
| 2148 |
+
},
|
| 2149 |
+
{
|
| 2150 |
+
"epoch": 64.66165413533835,
|
| 2151 |
+
"grad_norm": 15.257735252380371,
|
| 2152 |
+
"learning_rate": 1.936026936026936e-05,
|
| 2153 |
+
"loss": 0.1668,
|
| 2154 |
+
"step": 2150
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 64.9624060150376,
|
| 2158 |
+
"grad_norm": 10.21757698059082,
|
| 2159 |
+
"learning_rate": 1.919191919191919e-05,
|
| 2160 |
+
"loss": 0.173,
|
| 2161 |
+
"step": 2160
|
| 2162 |
+
},
|
| 2163 |
+
{
|
| 2164 |
+
"epoch": 64.99248120300751,
|
| 2165 |
+
"eval_accuracy": 0.5166666666666667,
|
| 2166 |
+
"eval_f1": 0.5030690537084399,
|
| 2167 |
+
"eval_loss": 2.629154682159424,
|
| 2168 |
+
"eval_runtime": 6.56,
|
| 2169 |
+
"eval_samples_per_second": 9.146,
|
| 2170 |
+
"eval_steps_per_second": 2.287,
|
| 2171 |
+
"step": 2161
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 65.26315789473684,
|
| 2175 |
+
"grad_norm": 9.854087829589844,
|
| 2176 |
+
"learning_rate": 1.9023569023569024e-05,
|
| 2177 |
+
"loss": 0.1248,
|
| 2178 |
+
"step": 2170
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"epoch": 65.56390977443608,
|
| 2182 |
+
"grad_norm": 4.802936553955078,
|
| 2183 |
+
"learning_rate": 1.8855218855218858e-05,
|
| 2184 |
+
"loss": 0.1242,
|
| 2185 |
+
"step": 2180
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 65.86466165413533,
|
| 2189 |
+
"grad_norm": 38.107017517089844,
|
| 2190 |
+
"learning_rate": 1.8686868686868688e-05,
|
| 2191 |
+
"loss": 0.1815,
|
| 2192 |
+
"step": 2190
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 65.98496240601504,
|
| 2196 |
+
"eval_accuracy": 0.6,
|
| 2197 |
+
"eval_f1": 0.5973214285714284,
|
| 2198 |
+
"eval_loss": 2.6566758155822754,
|
| 2199 |
+
"eval_runtime": 6.7429,
|
| 2200 |
+
"eval_samples_per_second": 8.898,
|
| 2201 |
+
"eval_steps_per_second": 2.225,
|
| 2202 |
+
"step": 2194
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 66.16541353383458,
|
| 2206 |
+
"grad_norm": 7.107358455657959,
|
| 2207 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 2208 |
+
"loss": 0.1251,
|
| 2209 |
+
"step": 2200
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 66.46616541353383,
|
| 2213 |
+
"grad_norm": 71.81128692626953,
|
| 2214 |
+
"learning_rate": 1.8350168350168352e-05,
|
| 2215 |
+
"loss": 0.3104,
|
| 2216 |
+
"step": 2210
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 66.76691729323308,
|
| 2220 |
+
"grad_norm": 0.22506819665431976,
|
| 2221 |
+
"learning_rate": 1.8181818181818182e-05,
|
| 2222 |
+
"loss": 0.0665,
|
| 2223 |
+
"step": 2220
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 66.97744360902256,
|
| 2227 |
+
"eval_accuracy": 0.5166666666666667,
|
| 2228 |
+
"eval_f1": 0.5030690537084399,
|
| 2229 |
+
"eval_loss": 3.2103826999664307,
|
| 2230 |
+
"eval_runtime": 7.0616,
|
| 2231 |
+
"eval_samples_per_second": 8.497,
|
| 2232 |
+
"eval_steps_per_second": 2.124,
|
| 2233 |
+
"step": 2227
|
| 2234 |
+
},
|
| 2235 |
+
{
|
| 2236 |
+
"epoch": 67.06766917293233,
|
| 2237 |
+
"grad_norm": 7.794451713562012,
|
| 2238 |
+
"learning_rate": 1.8013468013468016e-05,
|
| 2239 |
+
"loss": 0.1232,
|
| 2240 |
+
"step": 2230
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 67.36842105263158,
|
| 2244 |
+
"grad_norm": 0.3173252046108246,
|
| 2245 |
+
"learning_rate": 1.7845117845117846e-05,
|
| 2246 |
+
"loss": 0.0533,
|
| 2247 |
+
"step": 2240
|
| 2248 |
+
},
|
| 2249 |
+
{
|
| 2250 |
+
"epoch": 67.66917293233082,
|
| 2251 |
+
"grad_norm": 14.248137474060059,
|
| 2252 |
+
"learning_rate": 1.7676767676767676e-05,
|
| 2253 |
+
"loss": 0.1336,
|
| 2254 |
+
"step": 2250
|
| 2255 |
+
},
|
| 2256 |
+
{
|
| 2257 |
+
"epoch": 67.96992481203007,
|
| 2258 |
+
"grad_norm": 28.17351531982422,
|
| 2259 |
+
"learning_rate": 1.750841750841751e-05,
|
| 2260 |
+
"loss": 0.1084,
|
| 2261 |
+
"step": 2260
|
| 2262 |
+
},
|
| 2263 |
+
{
|
| 2264 |
+
"epoch": 68.0,
|
| 2265 |
+
"eval_accuracy": 0.5333333333333333,
|
| 2266 |
+
"eval_f1": 0.5227752639517346,
|
| 2267 |
+
"eval_loss": 3.669238567352295,
|
| 2268 |
+
"eval_runtime": 6.5484,
|
| 2269 |
+
"eval_samples_per_second": 9.163,
|
| 2270 |
+
"eval_steps_per_second": 2.291,
|
| 2271 |
+
"step": 2261
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 68.27067669172932,
|
| 2275 |
+
"grad_norm": 20.75010871887207,
|
| 2276 |
+
"learning_rate": 1.734006734006734e-05,
|
| 2277 |
+
"loss": 0.1947,
|
| 2278 |
+
"step": 2270
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 68.57142857142857,
|
| 2282 |
+
"grad_norm": 0.9663441777229309,
|
| 2283 |
+
"learning_rate": 1.7171717171717173e-05,
|
| 2284 |
+
"loss": 0.1185,
|
| 2285 |
+
"step": 2280
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 68.87218045112782,
|
| 2289 |
+
"grad_norm": 21.510988235473633,
|
| 2290 |
+
"learning_rate": 1.7003367003367004e-05,
|
| 2291 |
+
"loss": 0.1298,
|
| 2292 |
+
"step": 2290
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 68.99248120300751,
|
| 2296 |
+
"eval_accuracy": 0.55,
|
| 2297 |
+
"eval_f1": 0.5373401534526855,
|
| 2298 |
+
"eval_loss": 3.4104325771331787,
|
| 2299 |
+
"eval_runtime": 6.6129,
|
| 2300 |
+
"eval_samples_per_second": 9.073,
|
| 2301 |
+
"eval_steps_per_second": 2.268,
|
| 2302 |
+
"step": 2294
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 69.17293233082707,
|
| 2306 |
+
"grad_norm": 52.829627990722656,
|
| 2307 |
+
"learning_rate": 1.6835016835016837e-05,
|
| 2308 |
+
"loss": 0.1408,
|
| 2309 |
+
"step": 2300
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 69.47368421052632,
|
| 2313 |
+
"grad_norm": 14.396146774291992,
|
| 2314 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 2315 |
+
"loss": 0.0992,
|
| 2316 |
+
"step": 2310
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 69.77443609022556,
|
| 2320 |
+
"grad_norm": 19.629337310791016,
|
| 2321 |
+
"learning_rate": 1.6498316498316498e-05,
|
| 2322 |
+
"loss": 0.1338,
|
| 2323 |
+
"step": 2320
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 69.98496240601504,
|
| 2327 |
+
"eval_accuracy": 0.6,
|
| 2328 |
+
"eval_f1": 0.5973063973063972,
|
| 2329 |
+
"eval_loss": 2.82153058052063,
|
| 2330 |
+
"eval_runtime": 6.722,
|
| 2331 |
+
"eval_samples_per_second": 8.926,
|
| 2332 |
+
"eval_steps_per_second": 2.231,
|
| 2333 |
+
"step": 2327
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 70.07518796992481,
|
| 2337 |
+
"grad_norm": 18.8348445892334,
|
| 2338 |
+
"learning_rate": 1.632996632996633e-05,
|
| 2339 |
+
"loss": 0.1428,
|
| 2340 |
+
"step": 2330
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 70.37593984962406,
|
| 2344 |
+
"grad_norm": 30.958715438842773,
|
| 2345 |
+
"learning_rate": 1.6161616161616165e-05,
|
| 2346 |
+
"loss": 0.1255,
|
| 2347 |
+
"step": 2340
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 70.67669172932331,
|
| 2351 |
+
"grad_norm": 16.776973724365234,
|
| 2352 |
+
"learning_rate": 1.5993265993265995e-05,
|
| 2353 |
+
"loss": 0.1582,
|
| 2354 |
+
"step": 2350
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 70.97744360902256,
|
| 2358 |
+
"grad_norm": 8.8987398147583,
|
| 2359 |
+
"learning_rate": 1.5824915824915825e-05,
|
| 2360 |
+
"loss": 0.0795,
|
| 2361 |
+
"step": 2360
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 70.97744360902256,
|
| 2365 |
+
"eval_accuracy": 0.5833333333333334,
|
| 2366 |
+
"eval_f1": 0.5829852408799777,
|
| 2367 |
+
"eval_loss": 2.920820713043213,
|
| 2368 |
+
"eval_runtime": 6.6561,
|
| 2369 |
+
"eval_samples_per_second": 9.014,
|
| 2370 |
+
"eval_steps_per_second": 2.254,
|
| 2371 |
+
"step": 2360
|
| 2372 |
+
},
|
| 2373 |
+
{
|
| 2374 |
+
"epoch": 71.2781954887218,
|
| 2375 |
+
"grad_norm": 18.05194664001465,
|
| 2376 |
+
"learning_rate": 1.565656565656566e-05,
|
| 2377 |
+
"loss": 0.1047,
|
| 2378 |
+
"step": 2370
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 71.57894736842105,
|
| 2382 |
+
"grad_norm": 20.946168899536133,
|
| 2383 |
+
"learning_rate": 1.548821548821549e-05,
|
| 2384 |
+
"loss": 0.1055,
|
| 2385 |
+
"step": 2380
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 71.8796992481203,
|
| 2389 |
+
"grad_norm": 25.874860763549805,
|
| 2390 |
+
"learning_rate": 1.531986531986532e-05,
|
| 2391 |
+
"loss": 0.1138,
|
| 2392 |
+
"step": 2390
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"epoch": 72.0,
|
| 2396 |
+
"eval_accuracy": 0.5333333333333333,
|
| 2397 |
+
"eval_f1": 0.5302083333333333,
|
| 2398 |
+
"eval_loss": 3.427683115005493,
|
| 2399 |
+
"eval_runtime": 6.9445,
|
| 2400 |
+
"eval_samples_per_second": 8.64,
|
| 2401 |
+
"eval_steps_per_second": 2.16,
|
| 2402 |
+
"step": 2394
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 72.18045112781955,
|
| 2406 |
+
"grad_norm": 18.978233337402344,
|
| 2407 |
+
"learning_rate": 1.5151515151515153e-05,
|
| 2408 |
+
"loss": 0.155,
|
| 2409 |
+
"step": 2400
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 72.4812030075188,
|
| 2413 |
+
"grad_norm": 100.18550109863281,
|
| 2414 |
+
"learning_rate": 1.4983164983164985e-05,
|
| 2415 |
+
"loss": 0.1628,
|
| 2416 |
+
"step": 2410
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 72.78195488721805,
|
| 2420 |
+
"grad_norm": 65.66423797607422,
|
| 2421 |
+
"learning_rate": 1.4814814814814815e-05,
|
| 2422 |
+
"loss": 0.1644,
|
| 2423 |
+
"step": 2420
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 72.99248120300751,
|
| 2427 |
+
"eval_accuracy": 0.5833333333333334,
|
| 2428 |
+
"eval_f1": 0.5829852408799777,
|
| 2429 |
+
"eval_loss": 2.8140501976013184,
|
| 2430 |
+
"eval_runtime": 6.6318,
|
| 2431 |
+
"eval_samples_per_second": 9.047,
|
| 2432 |
+
"eval_steps_per_second": 2.262,
|
| 2433 |
+
"step": 2427
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 73.0827067669173,
|
| 2437 |
+
"grad_norm": 2.0067760944366455,
|
| 2438 |
+
"learning_rate": 1.4646464646464647e-05,
|
| 2439 |
+
"loss": 0.1517,
|
| 2440 |
+
"step": 2430
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 73.38345864661655,
|
| 2444 |
+
"grad_norm": 0.03711702302098274,
|
| 2445 |
+
"learning_rate": 1.4478114478114479e-05,
|
| 2446 |
+
"loss": 0.1144,
|
| 2447 |
+
"step": 2440
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 73.6842105263158,
|
| 2451 |
+
"grad_norm": 4.453554153442383,
|
| 2452 |
+
"learning_rate": 1.4309764309764309e-05,
|
| 2453 |
+
"loss": 0.0904,
|
| 2454 |
+
"step": 2450
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 73.98496240601504,
|
| 2458 |
+
"grad_norm": 39.73163604736328,
|
| 2459 |
+
"learning_rate": 1.4141414141414141e-05,
|
| 2460 |
+
"loss": 0.1659,
|
| 2461 |
+
"step": 2460
|
| 2462 |
+
},
|
| 2463 |
+
{
|
| 2464 |
+
"epoch": 73.98496240601504,
|
| 2465 |
+
"eval_accuracy": 0.6,
|
| 2466 |
+
"eval_f1": 0.6,
|
| 2467 |
+
"eval_loss": 2.872349500656128,
|
| 2468 |
+
"eval_runtime": 6.965,
|
| 2469 |
+
"eval_samples_per_second": 8.614,
|
| 2470 |
+
"eval_steps_per_second": 2.154,
|
| 2471 |
+
"step": 2460
|
| 2472 |
+
},
|
| 2473 |
+
{
|
| 2474 |
+
"epoch": 74.28571428571429,
|
| 2475 |
+
"grad_norm": 3.167804479598999,
|
| 2476 |
+
"learning_rate": 1.3973063973063974e-05,
|
| 2477 |
+
"loss": 0.0658,
|
| 2478 |
+
"step": 2470
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 74.58646616541354,
|
| 2482 |
+
"grad_norm": 6.628215789794922,
|
| 2483 |
+
"learning_rate": 1.3804713804713806e-05,
|
| 2484 |
+
"loss": 0.1165,
|
| 2485 |
+
"step": 2480
|
| 2486 |
+
},
|
| 2487 |
+
{
|
| 2488 |
+
"epoch": 74.88721804511279,
|
| 2489 |
+
"grad_norm": 36.84265899658203,
|
| 2490 |
+
"learning_rate": 1.3636363636363637e-05,
|
| 2491 |
+
"loss": 0.0453,
|
| 2492 |
+
"step": 2490
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 74.97744360902256,
|
| 2496 |
+
"eval_accuracy": 0.6333333333333333,
|
| 2497 |
+
"eval_f1": 0.6308641975308641,
|
| 2498 |
+
"eval_loss": 2.876894950866699,
|
| 2499 |
+
"eval_runtime": 6.9372,
|
| 2500 |
+
"eval_samples_per_second": 8.649,
|
| 2501 |
+
"eval_steps_per_second": 2.162,
|
| 2502 |
+
"step": 2493
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 75.18796992481202,
|
| 2506 |
+
"grad_norm": 40.516727447509766,
|
| 2507 |
+
"learning_rate": 1.3468013468013468e-05,
|
| 2508 |
+
"loss": 0.095,
|
| 2509 |
+
"step": 2500
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 75.48872180451127,
|
| 2513 |
+
"grad_norm": 10.00910758972168,
|
| 2514 |
+
"learning_rate": 1.32996632996633e-05,
|
| 2515 |
+
"loss": 0.078,
|
| 2516 |
+
"step": 2510
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 75.78947368421052,
|
| 2520 |
+
"grad_norm": 28.663267135620117,
|
| 2521 |
+
"learning_rate": 1.3131313131313134e-05,
|
| 2522 |
+
"loss": 0.0956,
|
| 2523 |
+
"step": 2520
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 76.0,
|
| 2527 |
+
"eval_accuracy": 0.6166666666666667,
|
| 2528 |
+
"eval_f1": 0.6098039215686274,
|
| 2529 |
+
"eval_loss": 3.29703426361084,
|
| 2530 |
+
"eval_runtime": 6.6814,
|
| 2531 |
+
"eval_samples_per_second": 8.98,
|
| 2532 |
+
"eval_steps_per_second": 2.245,
|
| 2533 |
+
"step": 2527
|
| 2534 |
+
},
|
| 2535 |
+
{
|
| 2536 |
+
"epoch": 76.09022556390977,
|
| 2537 |
+
"grad_norm": 0.029593205079436302,
|
| 2538 |
+
"learning_rate": 1.2962962962962962e-05,
|
| 2539 |
+
"loss": 0.1537,
|
| 2540 |
+
"step": 2530
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 76.39097744360902,
|
| 2544 |
+
"grad_norm": 117.96146392822266,
|
| 2545 |
+
"learning_rate": 1.2794612794612796e-05,
|
| 2546 |
+
"loss": 0.1007,
|
| 2547 |
+
"step": 2540
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 76.69172932330827,
|
| 2551 |
+
"grad_norm": 3.2077600955963135,
|
| 2552 |
+
"learning_rate": 1.2626262626262628e-05,
|
| 2553 |
+
"loss": 0.0296,
|
| 2554 |
+
"step": 2550
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 76.99248120300751,
|
| 2558 |
+
"grad_norm": 17.509721755981445,
|
| 2559 |
+
"learning_rate": 1.2457912457912458e-05,
|
| 2560 |
+
"loss": 0.1581,
|
| 2561 |
+
"step": 2560
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 76.99248120300751,
|
| 2565 |
+
"eval_accuracy": 0.5833333333333334,
|
| 2566 |
+
"eval_f1": 0.5815850815850816,
|
| 2567 |
+
"eval_loss": 3.6671502590179443,
|
| 2568 |
+
"eval_runtime": 6.7787,
|
| 2569 |
+
"eval_samples_per_second": 8.851,
|
| 2570 |
+
"eval_steps_per_second": 2.213,
|
| 2571 |
+
"step": 2560
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 77.29323308270676,
|
| 2575 |
+
"grad_norm": 17.232837677001953,
|
| 2576 |
+
"learning_rate": 1.228956228956229e-05,
|
| 2577 |
+
"loss": 0.0953,
|
| 2578 |
+
"step": 2570
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 77.59398496240601,
|
| 2582 |
+
"grad_norm": 4.897756099700928,
|
| 2583 |
+
"learning_rate": 1.2121212121212122e-05,
|
| 2584 |
+
"loss": 0.1592,
|
| 2585 |
+
"step": 2580
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 77.89473684210526,
|
| 2589 |
+
"grad_norm": 7.417236328125,
|
| 2590 |
+
"learning_rate": 1.1952861952861954e-05,
|
| 2591 |
+
"loss": 0.157,
|
| 2592 |
+
"step": 2590
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 77.98496240601504,
|
| 2596 |
+
"eval_accuracy": 0.55,
|
| 2597 |
+
"eval_f1": 0.55012503473187,
|
| 2598 |
+
"eval_loss": 3.531651258468628,
|
| 2599 |
+
"eval_runtime": 6.7182,
|
| 2600 |
+
"eval_samples_per_second": 8.931,
|
| 2601 |
+
"eval_steps_per_second": 2.233,
|
| 2602 |
+
"step": 2593
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 78.19548872180451,
|
| 2606 |
+
"grad_norm": 2.9548983573913574,
|
| 2607 |
+
"learning_rate": 1.1784511784511786e-05,
|
| 2608 |
+
"loss": 0.1002,
|
| 2609 |
+
"step": 2600
|
| 2610 |
+
},
|
| 2611 |
+
{
|
| 2612 |
+
"epoch": 78.49624060150376,
|
| 2613 |
+
"grad_norm": 0.129615917801857,
|
| 2614 |
+
"learning_rate": 1.1616161616161616e-05,
|
| 2615 |
+
"loss": 0.2083,
|
| 2616 |
+
"step": 2610
|
| 2617 |
+
},
|
| 2618 |
+
{
|
| 2619 |
+
"epoch": 78.796992481203,
|
| 2620 |
+
"grad_norm": 13.61913013458252,
|
| 2621 |
+
"learning_rate": 1.144781144781145e-05,
|
| 2622 |
+
"loss": 0.0662,
|
| 2623 |
+
"step": 2620
|
| 2624 |
+
},
|
| 2625 |
+
{
|
| 2626 |
+
"epoch": 78.97744360902256,
|
| 2627 |
+
"eval_accuracy": 0.55,
|
| 2628 |
+
"eval_f1": 0.5455944055944055,
|
| 2629 |
+
"eval_loss": 3.9002785682678223,
|
| 2630 |
+
"eval_runtime": 6.7274,
|
| 2631 |
+
"eval_samples_per_second": 8.919,
|
| 2632 |
+
"eval_steps_per_second": 2.23,
|
| 2633 |
+
"step": 2626
|
| 2634 |
+
},
|
| 2635 |
+
{
|
| 2636 |
+
"epoch": 79.09774436090225,
|
| 2637 |
+
"grad_norm": 39.20783233642578,
|
| 2638 |
+
"learning_rate": 1.127946127946128e-05,
|
| 2639 |
+
"loss": 0.0954,
|
| 2640 |
+
"step": 2630
|
| 2641 |
+
},
|
| 2642 |
+
{
|
| 2643 |
+
"epoch": 79.3984962406015,
|
| 2644 |
+
"grad_norm": 5.543479919433594,
|
| 2645 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 2646 |
+
"loss": 0.0968,
|
| 2647 |
+
"step": 2640
|
| 2648 |
+
},
|
| 2649 |
+
{
|
| 2650 |
+
"epoch": 79.69924812030075,
|
| 2651 |
+
"grad_norm": 1.9669289588928223,
|
| 2652 |
+
"learning_rate": 1.0942760942760944e-05,
|
| 2653 |
+
"loss": 0.061,
|
| 2654 |
+
"step": 2650
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 80.0,
|
| 2658 |
+
"grad_norm": 0.08515916019678116,
|
| 2659 |
+
"learning_rate": 1.0774410774410774e-05,
|
| 2660 |
+
"loss": 0.1954,
|
| 2661 |
+
"step": 2660
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 80.0,
|
| 2665 |
+
"eval_accuracy": 0.5833333333333334,
|
| 2666 |
+
"eval_f1": 0.5834491062332129,
|
| 2667 |
+
"eval_loss": 3.2999682426452637,
|
| 2668 |
+
"eval_runtime": 6.8436,
|
| 2669 |
+
"eval_samples_per_second": 8.767,
|
| 2670 |
+
"eval_steps_per_second": 2.192,
|
| 2671 |
+
"step": 2660
|
| 2672 |
+
},
|
| 2673 |
+
{
|
| 2674 |
+
"epoch": 80.30075187969925,
|
| 2675 |
+
"grad_norm": 129.70567321777344,
|
| 2676 |
+
"learning_rate": 1.0606060606060607e-05,
|
| 2677 |
+
"loss": 0.0925,
|
| 2678 |
+
"step": 2670
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 80.6015037593985,
|
| 2682 |
+
"grad_norm": 31.757814407348633,
|
| 2683 |
+
"learning_rate": 1.0437710437710438e-05,
|
| 2684 |
+
"loss": 0.0628,
|
| 2685 |
+
"step": 2680
|
| 2686 |
+
},
|
| 2687 |
+
{
|
| 2688 |
+
"epoch": 80.90225563909775,
|
| 2689 |
+
"grad_norm": 24.929340362548828,
|
| 2690 |
+
"learning_rate": 1.0269360269360271e-05,
|
| 2691 |
+
"loss": 0.0527,
|
| 2692 |
+
"step": 2690
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"epoch": 80.99248120300751,
|
| 2696 |
+
"eval_accuracy": 0.5666666666666667,
|
| 2697 |
+
"eval_f1": 0.563764880952381,
|
| 2698 |
+
"eval_loss": 3.9595704078674316,
|
| 2699 |
+
"eval_runtime": 6.6214,
|
| 2700 |
+
"eval_samples_per_second": 9.061,
|
| 2701 |
+
"eval_steps_per_second": 2.265,
|
| 2702 |
+
"step": 2693
|
| 2703 |
+
},
|
| 2704 |
+
{
|
| 2705 |
+
"epoch": 81.203007518797,
|
| 2706 |
+
"grad_norm": 29.07752799987793,
|
| 2707 |
+
"learning_rate": 1.0101010101010101e-05,
|
| 2708 |
+
"loss": 0.2314,
|
| 2709 |
+
"step": 2700
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 81.50375939849624,
|
| 2713 |
+
"grad_norm": 3.3806166648864746,
|
| 2714 |
+
"learning_rate": 9.932659932659933e-06,
|
| 2715 |
+
"loss": 0.1772,
|
| 2716 |
+
"step": 2710
|
| 2717 |
+
},
|
| 2718 |
+
{
|
| 2719 |
+
"epoch": 81.80451127819549,
|
| 2720 |
+
"grad_norm": 44.8973388671875,
|
| 2721 |
+
"learning_rate": 9.764309764309765e-06,
|
| 2722 |
+
"loss": 0.1578,
|
| 2723 |
+
"step": 2720
|
| 2724 |
+
},
|
| 2725 |
+
{
|
| 2726 |
+
"epoch": 81.98496240601504,
|
| 2727 |
+
"eval_accuracy": 0.55,
|
| 2728 |
+
"eval_f1": 0.5481203007518796,
|
| 2729 |
+
"eval_loss": 3.67238712310791,
|
| 2730 |
+
"eval_runtime": 6.7367,
|
| 2731 |
+
"eval_samples_per_second": 8.906,
|
| 2732 |
+
"eval_steps_per_second": 2.227,
|
| 2733 |
+
"step": 2726
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 82.10526315789474,
|
| 2737 |
+
"grad_norm": 35.64302062988281,
|
| 2738 |
+
"learning_rate": 9.595959595959595e-06,
|
| 2739 |
+
"loss": 0.1126,
|
| 2740 |
+
"step": 2730
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 82.40601503759399,
|
| 2744 |
+
"grad_norm": 1.958249568939209,
|
| 2745 |
+
"learning_rate": 9.427609427609429e-06,
|
| 2746 |
+
"loss": 0.0772,
|
| 2747 |
+
"step": 2740
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 82.70676691729324,
|
| 2751 |
+
"grad_norm": 15.566617965698242,
|
| 2752 |
+
"learning_rate": 9.259259259259259e-06,
|
| 2753 |
+
"loss": 0.0737,
|
| 2754 |
+
"step": 2750
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 82.97744360902256,
|
| 2758 |
+
"eval_accuracy": 0.5166666666666667,
|
| 2759 |
+
"eval_f1": 0.5119347319347319,
|
| 2760 |
+
"eval_loss": 4.022154331207275,
|
| 2761 |
+
"eval_runtime": 6.7543,
|
| 2762 |
+
"eval_samples_per_second": 8.883,
|
| 2763 |
+
"eval_steps_per_second": 2.221,
|
| 2764 |
+
"step": 2759
|
| 2765 |
+
},
|
| 2766 |
+
{
|
| 2767 |
+
"epoch": 83.00751879699249,
|
| 2768 |
+
"grad_norm": 29.270156860351562,
|
| 2769 |
+
"learning_rate": 9.090909090909091e-06,
|
| 2770 |
+
"loss": 0.1313,
|
| 2771 |
+
"step": 2760
|
| 2772 |
+
},
|
| 2773 |
+
{
|
| 2774 |
+
"epoch": 83.30827067669173,
|
| 2775 |
+
"grad_norm": 28.969818115234375,
|
| 2776 |
+
"learning_rate": 8.922558922558923e-06,
|
| 2777 |
+
"loss": 0.0702,
|
| 2778 |
+
"step": 2770
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 83.60902255639098,
|
| 2782 |
+
"grad_norm": 129.39942932128906,
|
| 2783 |
+
"learning_rate": 8.754208754208755e-06,
|
| 2784 |
+
"loss": 0.1426,
|
| 2785 |
+
"step": 2780
|
| 2786 |
+
},
|
| 2787 |
+
{
|
| 2788 |
+
"epoch": 83.90977443609023,
|
| 2789 |
+
"grad_norm": 38.763427734375,
|
| 2790 |
+
"learning_rate": 8.585858585858587e-06,
|
| 2791 |
+
"loss": 0.0617,
|
| 2792 |
+
"step": 2790
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 84.0,
|
| 2796 |
+
"eval_accuracy": 0.5833333333333334,
|
| 2797 |
+
"eval_f1": 0.5834491062332129,
|
| 2798 |
+
"eval_loss": 3.551039934158325,
|
| 2799 |
+
"eval_runtime": 6.5857,
|
| 2800 |
+
"eval_samples_per_second": 9.111,
|
| 2801 |
+
"eval_steps_per_second": 2.278,
|
| 2802 |
+
"step": 2793
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 84.21052631578948,
|
| 2806 |
+
"grad_norm": 4.888305187225342,
|
| 2807 |
+
"learning_rate": 8.417508417508419e-06,
|
| 2808 |
+
"loss": 0.0564,
|
| 2809 |
+
"step": 2800
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 84.51127819548873,
|
| 2813 |
+
"grad_norm": 58.47932434082031,
|
| 2814 |
+
"learning_rate": 8.249158249158249e-06,
|
| 2815 |
+
"loss": 0.0997,
|
| 2816 |
+
"step": 2810
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 84.81203007518798,
|
| 2820 |
+
"grad_norm": 2.1397640705108643,
|
| 2821 |
+
"learning_rate": 8.080808080808082e-06,
|
| 2822 |
+
"loss": 0.0531,
|
| 2823 |
+
"step": 2820
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 84.99248120300751,
|
| 2827 |
+
"eval_accuracy": 0.6,
|
| 2828 |
+
"eval_f1": 0.6,
|
| 2829 |
+
"eval_loss": 3.510986328125,
|
| 2830 |
+
"eval_runtime": 6.6781,
|
| 2831 |
+
"eval_samples_per_second": 8.985,
|
| 2832 |
+
"eval_steps_per_second": 2.246,
|
| 2833 |
+
"step": 2826
|
| 2834 |
+
},
|
| 2835 |
+
{
|
| 2836 |
+
"epoch": 85.11278195488721,
|
| 2837 |
+
"grad_norm": 9.81313705444336,
|
| 2838 |
+
"learning_rate": 7.912457912457913e-06,
|
| 2839 |
+
"loss": 0.2663,
|
| 2840 |
+
"step": 2830
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 85.41353383458646,
|
| 2844 |
+
"grad_norm": 89.6509017944336,
|
| 2845 |
+
"learning_rate": 7.744107744107745e-06,
|
| 2846 |
+
"loss": 0.1666,
|
| 2847 |
+
"step": 2840
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 85.71428571428571,
|
| 2851 |
+
"grad_norm": 1.1382590532302856,
|
| 2852 |
+
"learning_rate": 7.5757575757575764e-06,
|
| 2853 |
+
"loss": 0.0993,
|
| 2854 |
+
"step": 2850
|
| 2855 |
+
},
|
| 2856 |
+
{
|
| 2857 |
+
"epoch": 85.98496240601504,
|
| 2858 |
+
"eval_accuracy": 0.55,
|
| 2859 |
+
"eval_f1": 0.5481203007518796,
|
| 2860 |
+
"eval_loss": 4.06988000869751,
|
| 2861 |
+
"eval_runtime": 6.9333,
|
| 2862 |
+
"eval_samples_per_second": 8.654,
|
| 2863 |
+
"eval_steps_per_second": 2.163,
|
| 2864 |
+
"step": 2859
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 86.01503759398496,
|
| 2868 |
+
"grad_norm": 1.276601791381836,
|
| 2869 |
+
"learning_rate": 7.4074074074074075e-06,
|
| 2870 |
+
"loss": 0.0773,
|
| 2871 |
+
"step": 2860
|
| 2872 |
+
},
|
| 2873 |
+
{
|
| 2874 |
+
"epoch": 86.3157894736842,
|
| 2875 |
+
"grad_norm": 2.0897693634033203,
|
| 2876 |
+
"learning_rate": 7.239057239057239e-06,
|
| 2877 |
+
"loss": 0.0431,
|
| 2878 |
+
"step": 2870
|
| 2879 |
+
},
|
| 2880 |
+
{
|
| 2881 |
+
"epoch": 86.61654135338345,
|
| 2882 |
+
"grad_norm": 25.9661922454834,
|
| 2883 |
+
"learning_rate": 7.0707070707070704e-06,
|
| 2884 |
+
"loss": 0.1435,
|
| 2885 |
+
"step": 2880
|
| 2886 |
+
},
|
| 2887 |
+
{
|
| 2888 |
+
"epoch": 86.9172932330827,
|
| 2889 |
+
"grad_norm": 49.344520568847656,
|
| 2890 |
+
"learning_rate": 6.902356902356903e-06,
|
| 2891 |
+
"loss": 0.1545,
|
| 2892 |
+
"step": 2890
|
| 2893 |
+
},
|
| 2894 |
+
{
|
| 2895 |
+
"epoch": 86.97744360902256,
|
| 2896 |
+
"eval_accuracy": 0.5666666666666667,
|
| 2897 |
+
"eval_f1": 0.5666666666666667,
|
| 2898 |
+
"eval_loss": 3.6860132217407227,
|
| 2899 |
+
"eval_runtime": 7.1718,
|
| 2900 |
+
"eval_samples_per_second": 8.366,
|
| 2901 |
+
"eval_steps_per_second": 2.092,
|
| 2902 |
+
"step": 2892
|
| 2903 |
+
},
|
| 2904 |
+
{
|
| 2905 |
+
"epoch": 87.21804511278195,
|
| 2906 |
+
"grad_norm": 20.44532585144043,
|
| 2907 |
+
"learning_rate": 6.734006734006734e-06,
|
| 2908 |
+
"loss": 0.0467,
|
| 2909 |
+
"step": 2900
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 87.5187969924812,
|
| 2913 |
+
"grad_norm": 0.8258923292160034,
|
| 2914 |
+
"learning_rate": 6.565656565656567e-06,
|
| 2915 |
+
"loss": 0.1309,
|
| 2916 |
+
"step": 2910
|
| 2917 |
+
},
|
| 2918 |
+
{
|
| 2919 |
+
"epoch": 87.81954887218045,
|
| 2920 |
+
"grad_norm": 0.22264444828033447,
|
| 2921 |
+
"learning_rate": 6.397306397306398e-06,
|
| 2922 |
+
"loss": 0.0554,
|
| 2923 |
+
"step": 2920
|
| 2924 |
+
},
|
| 2925 |
+
{
|
| 2926 |
+
"epoch": 88.0,
|
| 2927 |
+
"eval_accuracy": 0.6,
|
| 2928 |
+
"eval_f1": 0.6,
|
| 2929 |
+
"eval_loss": 3.440887212753296,
|
| 2930 |
+
"eval_runtime": 6.7259,
|
| 2931 |
+
"eval_samples_per_second": 8.921,
|
| 2932 |
+
"eval_steps_per_second": 2.23,
|
| 2933 |
+
"step": 2926
|
| 2934 |
+
},
|
| 2935 |
+
{
|
| 2936 |
+
"epoch": 88.1203007518797,
|
| 2937 |
+
"grad_norm": 26.937969207763672,
|
| 2938 |
+
"learning_rate": 6.228956228956229e-06,
|
| 2939 |
+
"loss": 0.1753,
|
| 2940 |
+
"step": 2930
|
| 2941 |
+
},
|
| 2942 |
+
{
|
| 2943 |
+
"epoch": 88.42105263157895,
|
| 2944 |
+
"grad_norm": 5.258349418640137,
|
| 2945 |
+
"learning_rate": 6.060606060606061e-06,
|
| 2946 |
+
"loss": 0.0585,
|
| 2947 |
+
"step": 2940
|
| 2948 |
+
},
|
| 2949 |
+
{
|
| 2950 |
+
"epoch": 88.7218045112782,
|
| 2951 |
+
"grad_norm": 0.008519310504198074,
|
| 2952 |
+
"learning_rate": 5.892255892255893e-06,
|
| 2953 |
+
"loss": 0.0641,
|
| 2954 |
+
"step": 2950
|
| 2955 |
+
},
|
| 2956 |
+
{
|
| 2957 |
+
"epoch": 88.99248120300751,
|
| 2958 |
+
"eval_accuracy": 0.55,
|
| 2959 |
+
"eval_f1": 0.5496248958043901,
|
| 2960 |
+
"eval_loss": 3.8303909301757812,
|
| 2961 |
+
"eval_runtime": 6.6141,
|
| 2962 |
+
"eval_samples_per_second": 9.072,
|
| 2963 |
+
"eval_steps_per_second": 2.268,
|
| 2964 |
+
"step": 2959
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 89.02255639097744,
|
| 2968 |
+
"grad_norm": 0.1844397634267807,
|
| 2969 |
+
"learning_rate": 5.723905723905725e-06,
|
| 2970 |
+
"loss": 0.0186,
|
| 2971 |
+
"step": 2960
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 89.32330827067669,
|
| 2975 |
+
"grad_norm": 15.114968299865723,
|
| 2976 |
+
"learning_rate": 5.555555555555556e-06,
|
| 2977 |
+
"loss": 0.0863,
|
| 2978 |
+
"step": 2970
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 89.62406015037594,
|
| 2982 |
+
"grad_norm": 2.7161004543304443,
|
| 2983 |
+
"learning_rate": 5.387205387205387e-06,
|
| 2984 |
+
"loss": 0.0546,
|
| 2985 |
+
"step": 2980
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 89.92481203007519,
|
| 2989 |
+
"grad_norm": 24.26571273803711,
|
| 2990 |
+
"learning_rate": 5.218855218855219e-06,
|
| 2991 |
+
"loss": 0.0633,
|
| 2992 |
+
"step": 2990
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 89.98496240601504,
|
| 2996 |
+
"eval_accuracy": 0.55,
|
| 2997 |
+
"eval_f1": 0.5455944055944055,
|
| 2998 |
+
"eval_loss": 4.089901924133301,
|
| 2999 |
+
"eval_runtime": 6.6774,
|
| 3000 |
+
"eval_samples_per_second": 8.986,
|
| 3001 |
+
"eval_steps_per_second": 2.246,
|
| 3002 |
+
"step": 2992
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 90.22556390977444,
|
| 3006 |
+
"grad_norm": 1.2879589796066284,
|
| 3007 |
+
"learning_rate": 5.050505050505051e-06,
|
| 3008 |
+
"loss": 0.0261,
|
| 3009 |
+
"step": 3000
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 90.52631578947368,
|
| 3013 |
+
"grad_norm": 60.2087287902832,
|
| 3014 |
+
"learning_rate": 4.8821548821548826e-06,
|
| 3015 |
+
"loss": 0.0305,
|
| 3016 |
+
"step": 3010
|
| 3017 |
+
},
|
| 3018 |
+
{
|
| 3019 |
+
"epoch": 90.82706766917293,
|
| 3020 |
+
"grad_norm": 0.27982065081596375,
|
| 3021 |
+
"learning_rate": 4.7138047138047145e-06,
|
| 3022 |
+
"loss": 0.0991,
|
| 3023 |
+
"step": 3020
|
| 3024 |
+
},
|
| 3025 |
+
{
|
| 3026 |
+
"epoch": 90.97744360902256,
|
| 3027 |
+
"eval_accuracy": 0.6,
|
| 3028 |
+
"eval_f1": 0.6,
|
| 3029 |
+
"eval_loss": 3.734445095062256,
|
| 3030 |
+
"eval_runtime": 6.8383,
|
| 3031 |
+
"eval_samples_per_second": 8.774,
|
| 3032 |
+
"eval_steps_per_second": 2.194,
|
| 3033 |
+
"step": 3025
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 91.12781954887218,
|
| 3037 |
+
"grad_norm": 21.401033401489258,
|
| 3038 |
+
"learning_rate": 4.5454545454545455e-06,
|
| 3039 |
+
"loss": 0.099,
|
| 3040 |
+
"step": 3030
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 91.42857142857143,
|
| 3044 |
+
"grad_norm": 83.67088317871094,
|
| 3045 |
+
"learning_rate": 4.377104377104377e-06,
|
| 3046 |
+
"loss": 0.2015,
|
| 3047 |
+
"step": 3040
|
| 3048 |
+
},
|
| 3049 |
+
{
|
| 3050 |
+
"epoch": 91.72932330827068,
|
| 3051 |
+
"grad_norm": 4.662288188934326,
|
| 3052 |
+
"learning_rate": 4.208754208754209e-06,
|
| 3053 |
+
"loss": 0.0772,
|
| 3054 |
+
"step": 3050
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 92.0,
|
| 3058 |
+
"eval_accuracy": 0.6,
|
| 3059 |
+
"eval_f1": 0.5991101223581758,
|
| 3060 |
+
"eval_loss": 3.8447837829589844,
|
| 3061 |
+
"eval_runtime": 6.7081,
|
| 3062 |
+
"eval_samples_per_second": 8.944,
|
| 3063 |
+
"eval_steps_per_second": 2.236,
|
| 3064 |
+
"step": 3059
|
| 3065 |
+
},
|
| 3066 |
+
{
|
| 3067 |
+
"epoch": 92.03007518796993,
|
| 3068 |
+
"grad_norm": 8.341154098510742,
|
| 3069 |
+
"learning_rate": 4.040404040404041e-06,
|
| 3070 |
+
"loss": 0.1092,
|
| 3071 |
+
"step": 3060
|
| 3072 |
+
},
|
| 3073 |
+
{
|
| 3074 |
+
"epoch": 92.33082706766918,
|
| 3075 |
+
"grad_norm": 0.6782325506210327,
|
| 3076 |
+
"learning_rate": 3.872053872053872e-06,
|
| 3077 |
+
"loss": 0.1804,
|
| 3078 |
+
"step": 3070
|
| 3079 |
+
},
|
| 3080 |
+
{
|
| 3081 |
+
"epoch": 92.63157894736842,
|
| 3082 |
+
"grad_norm": 0.04346880316734314,
|
| 3083 |
+
"learning_rate": 3.7037037037037037e-06,
|
| 3084 |
+
"loss": 0.1252,
|
| 3085 |
+
"step": 3080
|
| 3086 |
+
},
|
| 3087 |
+
{
|
| 3088 |
+
"epoch": 92.93233082706767,
|
| 3089 |
+
"grad_norm": 8.484058380126953,
|
| 3090 |
+
"learning_rate": 3.5353535353535352e-06,
|
| 3091 |
+
"loss": 0.0646,
|
| 3092 |
+
"step": 3090
|
| 3093 |
+
},
|
| 3094 |
+
{
|
| 3095 |
+
"epoch": 92.99248120300751,
|
| 3096 |
+
"eval_accuracy": 0.6,
|
| 3097 |
+
"eval_f1": 0.5991101223581758,
|
| 3098 |
+
"eval_loss": 3.779362440109253,
|
| 3099 |
+
"eval_runtime": 6.7205,
|
| 3100 |
+
"eval_samples_per_second": 8.928,
|
| 3101 |
+
"eval_steps_per_second": 2.232,
|
| 3102 |
+
"step": 3092
|
| 3103 |
+
},
|
| 3104 |
+
{
|
| 3105 |
+
"epoch": 93.23308270676692,
|
| 3106 |
+
"grad_norm": 27.746423721313477,
|
| 3107 |
+
"learning_rate": 3.367003367003367e-06,
|
| 3108 |
+
"loss": 0.1189,
|
| 3109 |
+
"step": 3100
|
| 3110 |
+
},
|
| 3111 |
+
{
|
| 3112 |
+
"epoch": 93.53383458646617,
|
| 3113 |
+
"grad_norm": 73.70268249511719,
|
| 3114 |
+
"learning_rate": 3.198653198653199e-06,
|
| 3115 |
+
"loss": 0.0597,
|
| 3116 |
+
"step": 3110
|
| 3117 |
+
},
|
| 3118 |
+
{
|
| 3119 |
+
"epoch": 93.83458646616542,
|
| 3120 |
+
"grad_norm": 1.809439778327942,
|
| 3121 |
+
"learning_rate": 3.0303030303030305e-06,
|
| 3122 |
+
"loss": 0.0562,
|
| 3123 |
+
"step": 3120
|
| 3124 |
+
},
|
| 3125 |
+
{
|
| 3126 |
+
"epoch": 93.98496240601504,
|
| 3127 |
+
"eval_accuracy": 0.5833333333333334,
|
| 3128 |
+
"eval_f1": 0.5829860146336946,
|
| 3129 |
+
"eval_loss": 3.934034824371338,
|
| 3130 |
+
"eval_runtime": 6.724,
|
| 3131 |
+
"eval_samples_per_second": 8.923,
|
| 3132 |
+
"eval_steps_per_second": 2.231,
|
| 3133 |
+
"step": 3125
|
| 3134 |
+
},
|
| 3135 |
+
{
|
| 3136 |
+
"epoch": 94.13533834586467,
|
| 3137 |
+
"grad_norm": 28.367788314819336,
|
| 3138 |
+
"learning_rate": 2.8619528619528624e-06,
|
| 3139 |
+
"loss": 0.12,
|
| 3140 |
+
"step": 3130
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 94.43609022556392,
|
| 3144 |
+
"grad_norm": 18.267364501953125,
|
| 3145 |
+
"learning_rate": 2.6936026936026934e-06,
|
| 3146 |
+
"loss": 0.0473,
|
| 3147 |
+
"step": 3140
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 94.73684210526316,
|
| 3151 |
+
"grad_norm": 3.8993523120880127,
|
| 3152 |
+
"learning_rate": 2.5252525252525253e-06,
|
| 3153 |
+
"loss": 0.0475,
|
| 3154 |
+
"step": 3150
|
| 3155 |
+
},
|
| 3156 |
+
{
|
| 3157 |
+
"epoch": 94.97744360902256,
|
| 3158 |
+
"eval_accuracy": 0.55,
|
| 3159 |
+
"eval_f1": 0.5481203007518796,
|
| 3160 |
+
"eval_loss": 4.238762378692627,
|
| 3161 |
+
"eval_runtime": 6.722,
|
| 3162 |
+
"eval_samples_per_second": 8.926,
|
| 3163 |
+
"eval_steps_per_second": 2.231,
|
| 3164 |
+
"step": 3158
|
| 3165 |
+
},
|
| 3166 |
+
{
|
| 3167 |
+
"epoch": 95.0375939849624,
|
| 3168 |
+
"grad_norm": 1.1250429153442383,
|
| 3169 |
+
"learning_rate": 2.3569023569023572e-06,
|
| 3170 |
+
"loss": 0.1145,
|
| 3171 |
+
"step": 3160
|
| 3172 |
+
},
|
| 3173 |
+
{
|
| 3174 |
+
"epoch": 95.33834586466165,
|
| 3175 |
+
"grad_norm": 0.013231492601335049,
|
| 3176 |
+
"learning_rate": 2.1885521885521887e-06,
|
| 3177 |
+
"loss": 0.0648,
|
| 3178 |
+
"step": 3170
|
| 3179 |
+
},
|
| 3180 |
+
{
|
| 3181 |
+
"epoch": 95.6390977443609,
|
| 3182 |
+
"grad_norm": 0.1679408848285675,
|
| 3183 |
+
"learning_rate": 2.0202020202020206e-06,
|
| 3184 |
+
"loss": 0.0993,
|
| 3185 |
+
"step": 3180
|
| 3186 |
+
},
|
| 3187 |
+
{
|
| 3188 |
+
"epoch": 95.93984962406014,
|
| 3189 |
+
"grad_norm": 45.490089416503906,
|
| 3190 |
+
"learning_rate": 1.8518518518518519e-06,
|
| 3191 |
+
"loss": 0.0715,
|
| 3192 |
+
"step": 3190
|
| 3193 |
+
},
|
| 3194 |
+
{
|
| 3195 |
+
"epoch": 96.0,
|
| 3196 |
+
"eval_accuracy": 0.5333333333333333,
|
| 3197 |
+
"eval_f1": 0.5302083333333333,
|
| 3198 |
+
"eval_loss": 4.273224830627441,
|
| 3199 |
+
"eval_runtime": 6.8339,
|
| 3200 |
+
"eval_samples_per_second": 8.78,
|
| 3201 |
+
"eval_steps_per_second": 2.195,
|
| 3202 |
+
"step": 3192
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 96.2406015037594,
|
| 3206 |
+
"grad_norm": 6.96300745010376,
|
| 3207 |
+
"learning_rate": 1.6835016835016836e-06,
|
| 3208 |
+
"loss": 0.048,
|
| 3209 |
+
"step": 3200
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 96.54135338345864,
|
| 3213 |
+
"grad_norm": 2.091585397720337,
|
| 3214 |
+
"learning_rate": 1.5151515151515152e-06,
|
| 3215 |
+
"loss": 0.0434,
|
| 3216 |
+
"step": 3210
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 96.84210526315789,
|
| 3220 |
+
"grad_norm": 73.60545349121094,
|
| 3221 |
+
"learning_rate": 1.3468013468013467e-06,
|
| 3222 |
+
"loss": 0.0875,
|
| 3223 |
+
"step": 3220
|
| 3224 |
+
},
|
| 3225 |
+
{
|
| 3226 |
+
"epoch": 96.99248120300751,
|
| 3227 |
+
"eval_accuracy": 0.5666666666666667,
|
| 3228 |
+
"eval_f1": 0.5657026325546904,
|
| 3229 |
+
"eval_loss": 4.152093410491943,
|
| 3230 |
+
"eval_runtime": 6.4099,
|
| 3231 |
+
"eval_samples_per_second": 9.36,
|
| 3232 |
+
"eval_steps_per_second": 2.34,
|
| 3233 |
+
"step": 3225
|
| 3234 |
+
},
|
| 3235 |
+
{
|
| 3236 |
+
"epoch": 97.14285714285714,
|
| 3237 |
+
"grad_norm": 0.0025759534910321236,
|
| 3238 |
+
"learning_rate": 1.1784511784511786e-06,
|
| 3239 |
+
"loss": 0.0766,
|
| 3240 |
+
"step": 3230
|
| 3241 |
+
},
|
| 3242 |
+
{
|
| 3243 |
+
"epoch": 97.44360902255639,
|
| 3244 |
+
"grad_norm": 0.3083040416240692,
|
| 3245 |
+
"learning_rate": 1.0101010101010103e-06,
|
| 3246 |
+
"loss": 0.1051,
|
| 3247 |
+
"step": 3240
|
| 3248 |
+
},
|
| 3249 |
+
{
|
| 3250 |
+
"epoch": 97.74436090225564,
|
| 3251 |
+
"grad_norm": 49.4640998840332,
|
| 3252 |
+
"learning_rate": 8.417508417508418e-07,
|
| 3253 |
+
"loss": 0.0253,
|
| 3254 |
+
"step": 3250
|
| 3255 |
+
},
|
| 3256 |
+
{
|
| 3257 |
+
"epoch": 97.98496240601504,
|
| 3258 |
+
"eval_accuracy": 0.5666666666666667,
|
| 3259 |
+
"eval_f1": 0.5666666666666667,
|
| 3260 |
+
"eval_loss": 4.0813493728637695,
|
| 3261 |
+
"eval_runtime": 6.8426,
|
| 3262 |
+
"eval_samples_per_second": 8.769,
|
| 3263 |
+
"eval_steps_per_second": 2.192,
|
| 3264 |
+
"step": 3258
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 98.04511278195488,
|
| 3268 |
+
"grad_norm": 26.789464950561523,
|
| 3269 |
+
"learning_rate": 6.734006734006734e-07,
|
| 3270 |
+
"loss": 0.2367,
|
| 3271 |
+
"step": 3260
|
| 3272 |
+
},
|
| 3273 |
+
{
|
| 3274 |
+
"epoch": 98.34586466165413,
|
| 3275 |
+
"grad_norm": 2.038747549057007,
|
| 3276 |
+
"learning_rate": 5.050505050505052e-07,
|
| 3277 |
+
"loss": 0.1146,
|
| 3278 |
+
"step": 3270
|
| 3279 |
+
},
|
| 3280 |
+
{
|
| 3281 |
+
"epoch": 98.64661654135338,
|
| 3282 |
+
"grad_norm": 0.2587548494338989,
|
| 3283 |
+
"learning_rate": 3.367003367003367e-07,
|
| 3284 |
+
"loss": 0.0966,
|
| 3285 |
+
"step": 3280
|
| 3286 |
+
},
|
| 3287 |
+
{
|
| 3288 |
+
"epoch": 98.94736842105263,
|
| 3289 |
+
"grad_norm": 43.22801971435547,
|
| 3290 |
+
"learning_rate": 1.6835016835016834e-07,
|
| 3291 |
+
"loss": 0.1037,
|
| 3292 |
+
"step": 3290
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 98.97744360902256,
|
| 3296 |
+
"eval_accuracy": 0.5666666666666667,
|
| 3297 |
+
"eval_f1": 0.5666666666666667,
|
| 3298 |
+
"eval_loss": 4.107425212860107,
|
| 3299 |
+
"eval_runtime": 6.6385,
|
| 3300 |
+
"eval_samples_per_second": 9.038,
|
| 3301 |
+
"eval_steps_per_second": 2.26,
|
| 3302 |
+
"step": 3291
|
| 3303 |
+
},
|
| 3304 |
+
{
|
| 3305 |
+
"epoch": 99.24812030075188,
|
| 3306 |
+
"grad_norm": 1.0782755613327026,
|
| 3307 |
+
"learning_rate": 0.0,
|
| 3308 |
+
"loss": 0.1094,
|
| 3309 |
+
"step": 3300
|
| 3310 |
+
},
|
| 3311 |
+
{
|
| 3312 |
+
"epoch": 99.24812030075188,
|
| 3313 |
+
"eval_accuracy": 0.5666666666666667,
|
| 3314 |
+
"eval_f1": 0.5666666666666667,
|
| 3315 |
+
"eval_loss": 4.108070373535156,
|
| 3316 |
+
"eval_runtime": 8.9284,
|
| 3317 |
+
"eval_samples_per_second": 6.72,
|
| 3318 |
+
"eval_steps_per_second": 1.68,
|
| 3319 |
+
"step": 3300
|
| 3320 |
+
},
|
| 3321 |
+
{
|
| 3322 |
+
"epoch": 99.24812030075188,
|
| 3323 |
+
"step": 3300,
|
| 3324 |
+
"total_flos": 2.7357484620290458e+19,
|
| 3325 |
+
"train_loss": 0.32434610511768946,
|
| 3326 |
+
"train_runtime": 9596.5736,
|
| 3327 |
+
"train_samples_per_second": 5.544,
|
| 3328 |
+
"train_steps_per_second": 0.344
|
| 3329 |
+
}
|
| 3330 |
+
],
|
| 3331 |
+
"logging_steps": 10,
|
| 3332 |
+
"max_steps": 3300,
|
| 3333 |
+
"num_input_tokens_seen": 0,
|
| 3334 |
+
"num_train_epochs": 100,
|
| 3335 |
+
"save_steps": 500,
|
| 3336 |
+
"stateful_callbacks": {
|
| 3337 |
+
"TrainerControl": {
|
| 3338 |
+
"args": {
|
| 3339 |
+
"should_epoch_stop": false,
|
| 3340 |
+
"should_evaluate": false,
|
| 3341 |
+
"should_log": false,
|
| 3342 |
+
"should_save": true,
|
| 3343 |
+
"should_training_stop": true
|
| 3344 |
+
},
|
| 3345 |
+
"attributes": {}
|
| 3346 |
+
}
|
| 3347 |
+
},
|
| 3348 |
+
"total_flos": 2.7357484620290458e+19,
|
| 3349 |
+
"train_batch_size": 4,
|
| 3350 |
+
"trial_name": null,
|
| 3351 |
+
"trial_params": null
|
| 3352 |
+
}
|