Model save
Browse files- all_results.json +9 -0
- generation_config.json +12 -0
- train_results.json +9 -0
- trainer_state.json +2156 -0
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.4408511233930828,
|
| 5 |
+
"train_runtime": 2418.9161,
|
| 6 |
+
"train_samples": 10466,
|
| 7 |
+
"train_samples_per_second": 4.327,
|
| 8 |
+
"train_steps_per_second": 0.541
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 128000,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
128001,
|
| 6 |
+
128008,
|
| 7 |
+
128009
|
| 8 |
+
],
|
| 9 |
+
"temperature": 0.6,
|
| 10 |
+
"top_p": 0.9,
|
| 11 |
+
"transformers_version": "4.53.3"
|
| 12 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.4408511233930828,
|
| 5 |
+
"train_runtime": 2418.9161,
|
| 6 |
+
"train_samples": 10466,
|
| 7 |
+
"train_samples_per_second": 4.327,
|
| 8 |
+
"train_steps_per_second": 0.541
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2156 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1309,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0007639419404125286,
|
| 14 |
+
"grad_norm": 258.47398808043715,
|
| 15 |
+
"learning_rate": 0.0,
|
| 16 |
+
"logits/chosen": -0.5703125,
|
| 17 |
+
"logits/rejected": -0.5234375,
|
| 18 |
+
"logps/chosen": -540.0,
|
| 19 |
+
"logps/rejected": -464.0,
|
| 20 |
+
"loss": 0.6924,
|
| 21 |
+
"nll_loss": 1.2421875,
|
| 22 |
+
"rewards/accuracies": 0.0,
|
| 23 |
+
"rewards/chosen": 0.0,
|
| 24 |
+
"rewards/margins": 0.0,
|
| 25 |
+
"rewards/rejected": 0.0,
|
| 26 |
+
"step": 1
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.007639419404125287,
|
| 30 |
+
"grad_norm": 314.4150784740708,
|
| 31 |
+
"learning_rate": 3.435114503816794e-08,
|
| 32 |
+
"logits/chosen": -0.3402777910232544,
|
| 33 |
+
"logits/rejected": -0.34063720703125,
|
| 34 |
+
"logps/chosen": -493.77777099609375,
|
| 35 |
+
"logps/rejected": -482.8888854980469,
|
| 36 |
+
"loss": 0.6921,
|
| 37 |
+
"nll_loss": 0.9296875,
|
| 38 |
+
"rewards/accuracies": 0.2083333283662796,
|
| 39 |
+
"rewards/chosen": -0.00835503451526165,
|
| 40 |
+
"rewards/margins": 0.00830756314098835,
|
| 41 |
+
"rewards/rejected": -0.0166490338742733,
|
| 42 |
+
"step": 10
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.015278838808250574,
|
| 46 |
+
"grad_norm": 260.33930694178486,
|
| 47 |
+
"learning_rate": 7.251908396946564e-08,
|
| 48 |
+
"logits/chosen": -0.4122558534145355,
|
| 49 |
+
"logits/rejected": -0.42744141817092896,
|
| 50 |
+
"logps/chosen": -515.4000244140625,
|
| 51 |
+
"logps/rejected": -470.3999938964844,
|
| 52 |
+
"loss": 0.7081,
|
| 53 |
+
"nll_loss": 0.98046875,
|
| 54 |
+
"rewards/accuracies": 0.20000000298023224,
|
| 55 |
+
"rewards/chosen": -0.02316894568502903,
|
| 56 |
+
"rewards/margins": -0.0243988037109375,
|
| 57 |
+
"rewards/rejected": 0.0012573242420330644,
|
| 58 |
+
"step": 20
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.02291825821237586,
|
| 62 |
+
"grad_norm": 231.79105073106817,
|
| 63 |
+
"learning_rate": 1.1068702290076336e-07,
|
| 64 |
+
"logits/chosen": -0.38813477754592896,
|
| 65 |
+
"logits/rejected": -0.3990234434604645,
|
| 66 |
+
"logps/chosen": -586.2000122070312,
|
| 67 |
+
"logps/rejected": -565.2000122070312,
|
| 68 |
+
"loss": 0.6695,
|
| 69 |
+
"nll_loss": 0.967968761920929,
|
| 70 |
+
"rewards/accuracies": 0.375,
|
| 71 |
+
"rewards/chosen": 0.07893677055835724,
|
| 72 |
+
"rewards/margins": 0.06370849907398224,
|
| 73 |
+
"rewards/rejected": 0.01502075232565403,
|
| 74 |
+
"step": 30
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.030557677616501147,
|
| 78 |
+
"grad_norm": 207.3625348390448,
|
| 79 |
+
"learning_rate": 1.4885496183206107e-07,
|
| 80 |
+
"logits/chosen": -0.4215331971645355,
|
| 81 |
+
"logits/rejected": -0.3983154296875,
|
| 82 |
+
"logps/chosen": -525.5999755859375,
|
| 83 |
+
"logps/rejected": -436.20001220703125,
|
| 84 |
+
"loss": 0.6524,
|
| 85 |
+
"nll_loss": 0.9371093511581421,
|
| 86 |
+
"rewards/accuracies": 0.42500001192092896,
|
| 87 |
+
"rewards/chosen": 0.13503417372703552,
|
| 88 |
+
"rewards/margins": 0.10765381157398224,
|
| 89 |
+
"rewards/rejected": 0.02747192420065403,
|
| 90 |
+
"step": 40
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.03819709702062643,
|
| 94 |
+
"grad_norm": 226.87256558590633,
|
| 95 |
+
"learning_rate": 1.8702290076335877e-07,
|
| 96 |
+
"logits/chosen": -0.176025390625,
|
| 97 |
+
"logits/rejected": -0.181396484375,
|
| 98 |
+
"logps/chosen": -536.4000244140625,
|
| 99 |
+
"logps/rejected": -475.20001220703125,
|
| 100 |
+
"loss": 0.6743,
|
| 101 |
+
"nll_loss": 0.833984375,
|
| 102 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 103 |
+
"rewards/chosen": 0.20390625298023224,
|
| 104 |
+
"rewards/margins": 0.06151122972369194,
|
| 105 |
+
"rewards/rejected": 0.14276733994483948,
|
| 106 |
+
"step": 50
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 0.04583651642475172,
|
| 110 |
+
"grad_norm": 183.79471712514464,
|
| 111 |
+
"learning_rate": 2.2519083969465648e-07,
|
| 112 |
+
"logits/chosen": -0.36970216035842896,
|
| 113 |
+
"logits/rejected": -0.38398438692092896,
|
| 114 |
+
"logps/chosen": -692.4000244140625,
|
| 115 |
+
"logps/rejected": -639.5999755859375,
|
| 116 |
+
"loss": 0.5822,
|
| 117 |
+
"nll_loss": 0.9417968988418579,
|
| 118 |
+
"rewards/accuracies": 0.612500011920929,
|
| 119 |
+
"rewards/chosen": 0.377685546875,
|
| 120 |
+
"rewards/margins": 0.3079589903354645,
|
| 121 |
+
"rewards/rejected": 0.07008667290210724,
|
| 122 |
+
"step": 60
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.053475935828877004,
|
| 126 |
+
"grad_norm": 193.89130497070676,
|
| 127 |
+
"learning_rate": 2.633587786259542e-07,
|
| 128 |
+
"logits/chosen": -0.397705078125,
|
| 129 |
+
"logits/rejected": -0.43745118379592896,
|
| 130 |
+
"logps/chosen": -535.7999877929688,
|
| 131 |
+
"logps/rejected": -487.3999938964844,
|
| 132 |
+
"loss": 0.5212,
|
| 133 |
+
"nll_loss": 0.9515625238418579,
|
| 134 |
+
"rewards/accuracies": 0.675000011920929,
|
| 135 |
+
"rewards/chosen": 0.46650391817092896,
|
| 136 |
+
"rewards/margins": 0.5068359375,
|
| 137 |
+
"rewards/rejected": -0.0400390625,
|
| 138 |
+
"step": 70
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.061115355233002294,
|
| 142 |
+
"grad_norm": 204.2303212377364,
|
| 143 |
+
"learning_rate": 3.015267175572519e-07,
|
| 144 |
+
"logits/chosen": -0.32470703125,
|
| 145 |
+
"logits/rejected": -0.3303466737270355,
|
| 146 |
+
"logps/chosen": -473.6000061035156,
|
| 147 |
+
"logps/rejected": -454.0,
|
| 148 |
+
"loss": 0.5459,
|
| 149 |
+
"nll_loss": 0.876171886920929,
|
| 150 |
+
"rewards/accuracies": 0.637499988079071,
|
| 151 |
+
"rewards/chosen": 0.556835949420929,
|
| 152 |
+
"rewards/margins": 0.4847656190395355,
|
| 153 |
+
"rewards/rejected": 0.07137451320886612,
|
| 154 |
+
"step": 80
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.06875477463712758,
|
| 158 |
+
"grad_norm": 155.46344646489163,
|
| 159 |
+
"learning_rate": 3.396946564885496e-07,
|
| 160 |
+
"logits/chosen": -0.32172852754592896,
|
| 161 |
+
"logits/rejected": -0.33759766817092896,
|
| 162 |
+
"logps/chosen": -715.7999877929688,
|
| 163 |
+
"logps/rejected": -700.5999755859375,
|
| 164 |
+
"loss": 0.5522,
|
| 165 |
+
"nll_loss": 1.0656249523162842,
|
| 166 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 167 |
+
"rewards/chosen": 0.6611328125,
|
| 168 |
+
"rewards/margins": 0.5273803472518921,
|
| 169 |
+
"rewards/rejected": 0.1341552734375,
|
| 170 |
+
"step": 90
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.07639419404125286,
|
| 174 |
+
"grad_norm": 141.18575275944212,
|
| 175 |
+
"learning_rate": 3.7786259541984735e-07,
|
| 176 |
+
"logits/chosen": -0.3919433653354645,
|
| 177 |
+
"logits/rejected": -0.40632325410842896,
|
| 178 |
+
"logps/chosen": -533.0,
|
| 179 |
+
"logps/rejected": -520.7999877929688,
|
| 180 |
+
"loss": 0.5137,
|
| 181 |
+
"nll_loss": 0.940234363079071,
|
| 182 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 183 |
+
"rewards/chosen": 0.7120116949081421,
|
| 184 |
+
"rewards/margins": 0.7798827886581421,
|
| 185 |
+
"rewards/rejected": -0.06877441704273224,
|
| 186 |
+
"step": 100
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.08403361344537816,
|
| 190 |
+
"grad_norm": 231.1316603751942,
|
| 191 |
+
"learning_rate": 4.1603053435114506e-07,
|
| 192 |
+
"logits/chosen": -0.2874999940395355,
|
| 193 |
+
"logits/rejected": -0.3013549745082855,
|
| 194 |
+
"logps/chosen": -599.2000122070312,
|
| 195 |
+
"logps/rejected": -513.0,
|
| 196 |
+
"loss": 0.4408,
|
| 197 |
+
"nll_loss": 0.859375,
|
| 198 |
+
"rewards/accuracies": 0.675000011920929,
|
| 199 |
+
"rewards/chosen": 1.0283203125,
|
| 200 |
+
"rewards/margins": 1.157617211341858,
|
| 201 |
+
"rewards/rejected": -0.12840576469898224,
|
| 202 |
+
"step": 110
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.09167303284950344,
|
| 206 |
+
"grad_norm": 132.89992060448196,
|
| 207 |
+
"learning_rate": 4.541984732824427e-07,
|
| 208 |
+
"logits/chosen": -0.25129395723342896,
|
| 209 |
+
"logits/rejected": -0.27121275663375854,
|
| 210 |
+
"logps/chosen": -602.4000244140625,
|
| 211 |
+
"logps/rejected": -581.7999877929688,
|
| 212 |
+
"loss": 0.4933,
|
| 213 |
+
"nll_loss": 0.8753906488418579,
|
| 214 |
+
"rewards/accuracies": 0.675000011920929,
|
| 215 |
+
"rewards/chosen": 0.906445324420929,
|
| 216 |
+
"rewards/margins": 1.029296875,
|
| 217 |
+
"rewards/rejected": -0.121826171875,
|
| 218 |
+
"step": 120
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.09931245225362872,
|
| 222 |
+
"grad_norm": 133.95238732292788,
|
| 223 |
+
"learning_rate": 4.923664122137405e-07,
|
| 224 |
+
"logits/chosen": -0.21683350205421448,
|
| 225 |
+
"logits/rejected": -0.2558349668979645,
|
| 226 |
+
"logps/chosen": -614.2000122070312,
|
| 227 |
+
"logps/rejected": -569.5999755859375,
|
| 228 |
+
"loss": 0.3928,
|
| 229 |
+
"nll_loss": 0.8531249761581421,
|
| 230 |
+
"rewards/accuracies": 0.800000011920929,
|
| 231 |
+
"rewards/chosen": 0.694927990436554,
|
| 232 |
+
"rewards/margins": 1.353906273841858,
|
| 233 |
+
"rewards/rejected": -0.658007800579071,
|
| 234 |
+
"step": 130
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.10695187165775401,
|
| 238 |
+
"grad_norm": 168.1618677127587,
|
| 239 |
+
"learning_rate": 4.966044142614601e-07,
|
| 240 |
+
"logits/chosen": -0.3291992247104645,
|
| 241 |
+
"logits/rejected": -0.2532714903354645,
|
| 242 |
+
"logps/chosen": -496.79998779296875,
|
| 243 |
+
"logps/rejected": -428.79998779296875,
|
| 244 |
+
"loss": 0.5624,
|
| 245 |
+
"nll_loss": 0.9046875238418579,
|
| 246 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 247 |
+
"rewards/chosen": 0.8631836175918579,
|
| 248 |
+
"rewards/margins": 1.0146484375,
|
| 249 |
+
"rewards/rejected": -0.15104980766773224,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.11459129106187929,
|
| 254 |
+
"grad_norm": 162.0020969653275,
|
| 255 |
+
"learning_rate": 4.923599320882851e-07,
|
| 256 |
+
"logits/chosen": -0.3076171875,
|
| 257 |
+
"logits/rejected": -0.3332763612270355,
|
| 258 |
+
"logps/chosen": -540.4000244140625,
|
| 259 |
+
"logps/rejected": -498.20001220703125,
|
| 260 |
+
"loss": 0.4469,
|
| 261 |
+
"nll_loss": 0.893750011920929,
|
| 262 |
+
"rewards/accuracies": 0.762499988079071,
|
| 263 |
+
"rewards/chosen": 1.2221190929412842,
|
| 264 |
+
"rewards/margins": 1.4638671875,
|
| 265 |
+
"rewards/rejected": -0.24140624701976776,
|
| 266 |
+
"step": 150
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.12223071046600459,
|
| 270 |
+
"grad_norm": 174.01769486132392,
|
| 271 |
+
"learning_rate": 4.881154499151103e-07,
|
| 272 |
+
"logits/chosen": -0.32622069120407104,
|
| 273 |
+
"logits/rejected": -0.3348449766635895,
|
| 274 |
+
"logps/chosen": -591.0,
|
| 275 |
+
"logps/rejected": -580.7999877929688,
|
| 276 |
+
"loss": 0.5754,
|
| 277 |
+
"nll_loss": 1.0691406726837158,
|
| 278 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 279 |
+
"rewards/chosen": 1.2189452648162842,
|
| 280 |
+
"rewards/margins": 1.0265624523162842,
|
| 281 |
+
"rewards/rejected": 0.19337157905101776,
|
| 282 |
+
"step": 160
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.12987012987012986,
|
| 286 |
+
"grad_norm": 190.37882731718457,
|
| 287 |
+
"learning_rate": 4.838709677419355e-07,
|
| 288 |
+
"logits/chosen": -0.2706542909145355,
|
| 289 |
+
"logits/rejected": -0.33831483125686646,
|
| 290 |
+
"logps/chosen": -535.5999755859375,
|
| 291 |
+
"logps/rejected": -530.2000122070312,
|
| 292 |
+
"loss": 0.5538,
|
| 293 |
+
"nll_loss": 0.8460937738418579,
|
| 294 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 295 |
+
"rewards/chosen": 1.3611328601837158,
|
| 296 |
+
"rewards/margins": 0.967968761920929,
|
| 297 |
+
"rewards/rejected": 0.3932128846645355,
|
| 298 |
+
"step": 170
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.13750954927425516,
|
| 302 |
+
"grad_norm": 145.8998739490587,
|
| 303 |
+
"learning_rate": 4.796264855687606e-07,
|
| 304 |
+
"logits/chosen": -0.2790161073207855,
|
| 305 |
+
"logits/rejected": -0.29090577363967896,
|
| 306 |
+
"logps/chosen": -595.5999755859375,
|
| 307 |
+
"logps/rejected": -568.4000244140625,
|
| 308 |
+
"loss": 0.4184,
|
| 309 |
+
"nll_loss": 0.8812500238418579,
|
| 310 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 311 |
+
"rewards/chosen": 0.9041992425918579,
|
| 312 |
+
"rewards/margins": 1.43359375,
|
| 313 |
+
"rewards/rejected": -0.53125,
|
| 314 |
+
"step": 180
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.14514896867838045,
|
| 318 |
+
"grad_norm": 102.7664998097143,
|
| 319 |
+
"learning_rate": 4.753820033955857e-07,
|
| 320 |
+
"logits/chosen": -0.36333006620407104,
|
| 321 |
+
"logits/rejected": -0.30805665254592896,
|
| 322 |
+
"logps/chosen": -529.4000244140625,
|
| 323 |
+
"logps/rejected": -484.79998779296875,
|
| 324 |
+
"loss": 0.3831,
|
| 325 |
+
"nll_loss": 0.8929687738418579,
|
| 326 |
+
"rewards/accuracies": 0.800000011920929,
|
| 327 |
+
"rewards/chosen": 1.4363281726837158,
|
| 328 |
+
"rewards/margins": 1.766992211341858,
|
| 329 |
+
"rewards/rejected": -0.33027344942092896,
|
| 330 |
+
"step": 190
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.15278838808250572,
|
| 334 |
+
"grad_norm": 62.90690254682499,
|
| 335 |
+
"learning_rate": 4.7113752122241087e-07,
|
| 336 |
+
"logits/chosen": -0.16816405951976776,
|
| 337 |
+
"logits/rejected": -0.15498046576976776,
|
| 338 |
+
"logps/chosen": -583.4000244140625,
|
| 339 |
+
"logps/rejected": -536.5999755859375,
|
| 340 |
+
"loss": 0.4999,
|
| 341 |
+
"nll_loss": 0.892578125,
|
| 342 |
+
"rewards/accuracies": 0.6875,
|
| 343 |
+
"rewards/chosen": 1.306249976158142,
|
| 344 |
+
"rewards/margins": 1.3748047351837158,
|
| 345 |
+
"rewards/rejected": -0.06831054389476776,
|
| 346 |
+
"step": 200
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.16042780748663102,
|
| 350 |
+
"grad_norm": 84.68520470211422,
|
| 351 |
+
"learning_rate": 4.66893039049236e-07,
|
| 352 |
+
"logits/chosen": -0.19780273735523224,
|
| 353 |
+
"logits/rejected": -0.24296875298023224,
|
| 354 |
+
"logps/chosen": -525.0,
|
| 355 |
+
"logps/rejected": -510.79998779296875,
|
| 356 |
+
"loss": 0.6017,
|
| 357 |
+
"nll_loss": 0.946093738079071,
|
| 358 |
+
"rewards/accuracies": 0.637499988079071,
|
| 359 |
+
"rewards/chosen": 0.9354492425918579,
|
| 360 |
+
"rewards/margins": 1.27734375,
|
| 361 |
+
"rewards/rejected": -0.34150391817092896,
|
| 362 |
+
"step": 210
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.16806722689075632,
|
| 366 |
+
"grad_norm": 185.43708476964164,
|
| 367 |
+
"learning_rate": 4.6264855687606106e-07,
|
| 368 |
+
"logits/chosen": -0.27099609375,
|
| 369 |
+
"logits/rejected": -0.23505859076976776,
|
| 370 |
+
"logps/chosen": -484.3999938964844,
|
| 371 |
+
"logps/rejected": -489.20001220703125,
|
| 372 |
+
"loss": 0.487,
|
| 373 |
+
"nll_loss": 0.8511718511581421,
|
| 374 |
+
"rewards/accuracies": 0.675000011920929,
|
| 375 |
+
"rewards/chosen": 0.3269897401332855,
|
| 376 |
+
"rewards/margins": 1.3175780773162842,
|
| 377 |
+
"rewards/rejected": -0.990429699420929,
|
| 378 |
+
"step": 220
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.17570664629488159,
|
| 382 |
+
"grad_norm": 108.47585085006368,
|
| 383 |
+
"learning_rate": 4.5840407470288624e-07,
|
| 384 |
+
"logits/chosen": -0.2373046875,
|
| 385 |
+
"logits/rejected": -0.26665037870407104,
|
| 386 |
+
"logps/chosen": -519.2000122070312,
|
| 387 |
+
"logps/rejected": -494.6000061035156,
|
| 388 |
+
"loss": 0.3942,
|
| 389 |
+
"nll_loss": 0.8656250238418579,
|
| 390 |
+
"rewards/accuracies": 0.75,
|
| 391 |
+
"rewards/chosen": 0.721728503704071,
|
| 392 |
+
"rewards/margins": 2.1226563453674316,
|
| 393 |
+
"rewards/rejected": -1.4031250476837158,
|
| 394 |
+
"step": 230
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.18334606569900688,
|
| 398 |
+
"grad_norm": 167.38681415802225,
|
| 399 |
+
"learning_rate": 4.5415959252971136e-07,
|
| 400 |
+
"logits/chosen": -0.18267822265625,
|
| 401 |
+
"logits/rejected": -0.17292480170726776,
|
| 402 |
+
"logps/chosen": -525.0,
|
| 403 |
+
"logps/rejected": -501.6000061035156,
|
| 404 |
+
"loss": 0.5013,
|
| 405 |
+
"nll_loss": 0.892578125,
|
| 406 |
+
"rewards/accuracies": 0.6875,
|
| 407 |
+
"rewards/chosen": 0.698437511920929,
|
| 408 |
+
"rewards/margins": 1.597265601158142,
|
| 409 |
+
"rewards/rejected": -0.89697265625,
|
| 410 |
+
"step": 240
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.19098548510313215,
|
| 414 |
+
"grad_norm": 101.64538233833237,
|
| 415 |
+
"learning_rate": 4.499151103565365e-07,
|
| 416 |
+
"logits/chosen": -0.15837402641773224,
|
| 417 |
+
"logits/rejected": -0.19194336235523224,
|
| 418 |
+
"logps/chosen": -677.4000244140625,
|
| 419 |
+
"logps/rejected": -586.4000244140625,
|
| 420 |
+
"loss": 0.4529,
|
| 421 |
+
"nll_loss": 0.887890636920929,
|
| 422 |
+
"rewards/accuracies": 0.699999988079071,
|
| 423 |
+
"rewards/chosen": 0.6534668207168579,
|
| 424 |
+
"rewards/margins": 1.8076171875,
|
| 425 |
+
"rewards/rejected": -1.1537597179412842,
|
| 426 |
+
"step": 250
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.19862490450725745,
|
| 430 |
+
"grad_norm": 294.75439921434145,
|
| 431 |
+
"learning_rate": 4.456706281833616e-07,
|
| 432 |
+
"logits/chosen": -0.2868896424770355,
|
| 433 |
+
"logits/rejected": -0.22414550185203552,
|
| 434 |
+
"logps/chosen": -544.0,
|
| 435 |
+
"logps/rejected": -496.3999938964844,
|
| 436 |
+
"loss": 0.4019,
|
| 437 |
+
"nll_loss": 0.926562488079071,
|
| 438 |
+
"rewards/accuracies": 0.737500011920929,
|
| 439 |
+
"rewards/chosen": 1.142968773841858,
|
| 440 |
+
"rewards/margins": 1.655859351158142,
|
| 441 |
+
"rewards/rejected": -0.5120483636856079,
|
| 442 |
+
"step": 260
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.20626432391138275,
|
| 446 |
+
"grad_norm": 117.11210831501775,
|
| 447 |
+
"learning_rate": 4.4142614601018673e-07,
|
| 448 |
+
"logits/chosen": -0.19429931044578552,
|
| 449 |
+
"logits/rejected": -0.218994140625,
|
| 450 |
+
"logps/chosen": -575.2000122070312,
|
| 451 |
+
"logps/rejected": -544.2000122070312,
|
| 452 |
+
"loss": 0.5657,
|
| 453 |
+
"nll_loss": 0.871874988079071,
|
| 454 |
+
"rewards/accuracies": 0.6875,
|
| 455 |
+
"rewards/chosen": 1.008203148841858,
|
| 456 |
+
"rewards/margins": 1.3722655773162842,
|
| 457 |
+
"rewards/rejected": -0.3637451231479645,
|
| 458 |
+
"step": 270
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.21390374331550802,
|
| 462 |
+
"grad_norm": 163.8676544002689,
|
| 463 |
+
"learning_rate": 4.3718166383701186e-07,
|
| 464 |
+
"logits/chosen": -0.25664061307907104,
|
| 465 |
+
"logits/rejected": -0.23786620795726776,
|
| 466 |
+
"logps/chosen": -637.7999877929688,
|
| 467 |
+
"logps/rejected": -592.0,
|
| 468 |
+
"loss": 0.3987,
|
| 469 |
+
"nll_loss": 0.989453136920929,
|
| 470 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 471 |
+
"rewards/chosen": 0.5733398199081421,
|
| 472 |
+
"rewards/margins": 1.9832031726837158,
|
| 473 |
+
"rewards/rejected": -1.4110839366912842,
|
| 474 |
+
"step": 280
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.2215431627196333,
|
| 478 |
+
"grad_norm": 264.70174523111024,
|
| 479 |
+
"learning_rate": 4.32937181663837e-07,
|
| 480 |
+
"logits/chosen": -0.16657714545726776,
|
| 481 |
+
"logits/rejected": -0.16425780951976776,
|
| 482 |
+
"logps/chosen": -623.0,
|
| 483 |
+
"logps/rejected": -573.7999877929688,
|
| 484 |
+
"loss": 0.5026,
|
| 485 |
+
"nll_loss": 0.8687499761581421,
|
| 486 |
+
"rewards/accuracies": 0.6875,
|
| 487 |
+
"rewards/chosen": 0.6055663824081421,
|
| 488 |
+
"rewards/margins": 1.7492187023162842,
|
| 489 |
+
"rewards/rejected": -1.142822265625,
|
| 490 |
+
"step": 290
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.22918258212375858,
|
| 494 |
+
"grad_norm": 188.9470989355876,
|
| 495 |
+
"learning_rate": 4.286926994906621e-07,
|
| 496 |
+
"logits/chosen": -0.35651856660842896,
|
| 497 |
+
"logits/rejected": -0.402099609375,
|
| 498 |
+
"logps/chosen": -510.6000061035156,
|
| 499 |
+
"logps/rejected": -459.3999938964844,
|
| 500 |
+
"loss": 0.4784,
|
| 501 |
+
"nll_loss": 0.896484375,
|
| 502 |
+
"rewards/accuracies": 0.6875,
|
| 503 |
+
"rewards/chosen": 1.013574242591858,
|
| 504 |
+
"rewards/margins": 1.8142578601837158,
|
| 505 |
+
"rewards/rejected": -0.8011718988418579,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.23682200152788388,
|
| 510 |
+
"grad_norm": 225.67003396669838,
|
| 511 |
+
"learning_rate": 4.244482173174873e-07,
|
| 512 |
+
"logits/chosen": -0.1279296875,
|
| 513 |
+
"logits/rejected": -0.13811035454273224,
|
| 514 |
+
"logps/chosen": -545.0,
|
| 515 |
+
"logps/rejected": -516.7999877929688,
|
| 516 |
+
"loss": 0.4794,
|
| 517 |
+
"nll_loss": 0.8207031488418579,
|
| 518 |
+
"rewards/accuracies": 0.6875,
|
| 519 |
+
"rewards/chosen": 1.391210913658142,
|
| 520 |
+
"rewards/margins": 1.7058594226837158,
|
| 521 |
+
"rewards/rejected": -0.3139282166957855,
|
| 522 |
+
"step": 310
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.24446142093200918,
|
| 526 |
+
"grad_norm": 105.64614366845814,
|
| 527 |
+
"learning_rate": 4.202037351443124e-07,
|
| 528 |
+
"logits/chosen": -0.16025391221046448,
|
| 529 |
+
"logits/rejected": -0.19588622450828552,
|
| 530 |
+
"logps/chosen": -463.79998779296875,
|
| 531 |
+
"logps/rejected": -460.6000061035156,
|
| 532 |
+
"loss": 0.2819,
|
| 533 |
+
"nll_loss": 0.8589843511581421,
|
| 534 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 535 |
+
"rewards/chosen": 1.2197265625,
|
| 536 |
+
"rewards/margins": 2.479687452316284,
|
| 537 |
+
"rewards/rejected": -1.260498046875,
|
| 538 |
+
"step": 320
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.25210084033613445,
|
| 542 |
+
"grad_norm": 287.90740376642924,
|
| 543 |
+
"learning_rate": 4.159592529711375e-07,
|
| 544 |
+
"logits/chosen": -0.2608642578125,
|
| 545 |
+
"logits/rejected": -0.25861817598342896,
|
| 546 |
+
"logps/chosen": -548.4000244140625,
|
| 547 |
+
"logps/rejected": -534.7999877929688,
|
| 548 |
+
"loss": 0.6499,
|
| 549 |
+
"nll_loss": 0.977734386920929,
|
| 550 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 551 |
+
"rewards/chosen": 0.809765636920929,
|
| 552 |
+
"rewards/margins": 1.789453148841858,
|
| 553 |
+
"rewards/rejected": -0.979785144329071,
|
| 554 |
+
"step": 330
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.2597402597402597,
|
| 558 |
+
"grad_norm": 184.4996723823412,
|
| 559 |
+
"learning_rate": 4.1171477079796265e-07,
|
| 560 |
+
"logits/chosen": -0.15278320014476776,
|
| 561 |
+
"logits/rejected": -0.20881347358226776,
|
| 562 |
+
"logps/chosen": -534.4000244140625,
|
| 563 |
+
"logps/rejected": -502.20001220703125,
|
| 564 |
+
"loss": 0.3265,
|
| 565 |
+
"nll_loss": 0.8968750238418579,
|
| 566 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 567 |
+
"rewards/chosen": 0.8492431640625,
|
| 568 |
+
"rewards/margins": 2.390625,
|
| 569 |
+
"rewards/rejected": -1.5436522960662842,
|
| 570 |
+
"step": 340
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.26737967914438504,
|
| 574 |
+
"grad_norm": 176.1001324023726,
|
| 575 |
+
"learning_rate": 4.074702886247878e-07,
|
| 576 |
+
"logits/chosen": -0.291748046875,
|
| 577 |
+
"logits/rejected": -0.30827635526657104,
|
| 578 |
+
"logps/chosen": -536.7999877929688,
|
| 579 |
+
"logps/rejected": -502.20001220703125,
|
| 580 |
+
"loss": 0.5069,
|
| 581 |
+
"nll_loss": 0.903124988079071,
|
| 582 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 583 |
+
"rewards/chosen": 0.542651355266571,
|
| 584 |
+
"rewards/margins": 1.62109375,
|
| 585 |
+
"rewards/rejected": -1.079248070716858,
|
| 586 |
+
"step": 350
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.2750190985485103,
|
| 590 |
+
"grad_norm": 215.75426907278032,
|
| 591 |
+
"learning_rate": 4.0322580645161285e-07,
|
| 592 |
+
"logits/chosen": -0.19513702392578125,
|
| 593 |
+
"logits/rejected": -0.18853148818016052,
|
| 594 |
+
"logps/chosen": -659.4000244140625,
|
| 595 |
+
"logps/rejected": -631.2000122070312,
|
| 596 |
+
"loss": 0.4296,
|
| 597 |
+
"nll_loss": 0.93359375,
|
| 598 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 599 |
+
"rewards/chosen": 1.0828125476837158,
|
| 600 |
+
"rewards/margins": 1.869531273841858,
|
| 601 |
+
"rewards/rejected": -0.787548840045929,
|
| 602 |
+
"step": 360
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.2826585179526356,
|
| 606 |
+
"grad_norm": 108.88771986132863,
|
| 607 |
+
"learning_rate": 3.98981324278438e-07,
|
| 608 |
+
"logits/chosen": -0.3078857362270355,
|
| 609 |
+
"logits/rejected": -0.29511719942092896,
|
| 610 |
+
"logps/chosen": -550.2000122070312,
|
| 611 |
+
"logps/rejected": -524.5999755859375,
|
| 612 |
+
"loss": 0.4382,
|
| 613 |
+
"nll_loss": 0.8999999761581421,
|
| 614 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 615 |
+
"rewards/chosen": 0.9032226800918579,
|
| 616 |
+
"rewards/margins": 1.9148437976837158,
|
| 617 |
+
"rewards/rejected": -1.0095703601837158,
|
| 618 |
+
"step": 370
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.2902979373567609,
|
| 622 |
+
"grad_norm": 157.57914105986907,
|
| 623 |
+
"learning_rate": 3.9473684210526315e-07,
|
| 624 |
+
"logits/chosen": -0.24609375,
|
| 625 |
+
"logits/rejected": -0.2859863340854645,
|
| 626 |
+
"logps/chosen": -623.5999755859375,
|
| 627 |
+
"logps/rejected": -571.5999755859375,
|
| 628 |
+
"loss": 0.4434,
|
| 629 |
+
"nll_loss": 0.9203125238418579,
|
| 630 |
+
"rewards/accuracies": 0.75,
|
| 631 |
+
"rewards/chosen": 1.6277344226837158,
|
| 632 |
+
"rewards/margins": 1.9421875476837158,
|
| 633 |
+
"rewards/rejected": -0.314453125,
|
| 634 |
+
"step": 380
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.2979373567608862,
|
| 638 |
+
"grad_norm": 95.48144608914365,
|
| 639 |
+
"learning_rate": 3.9049235993208827e-07,
|
| 640 |
+
"logits/chosen": -0.24777832627296448,
|
| 641 |
+
"logits/rejected": -0.23862305283546448,
|
| 642 |
+
"logps/chosen": -589.5999755859375,
|
| 643 |
+
"logps/rejected": -546.5999755859375,
|
| 644 |
+
"loss": 0.414,
|
| 645 |
+
"nll_loss": 0.923046886920929,
|
| 646 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 647 |
+
"rewards/chosen": 1.652441382408142,
|
| 648 |
+
"rewards/margins": 2.092968702316284,
|
| 649 |
+
"rewards/rejected": -0.43925780057907104,
|
| 650 |
+
"step": 390
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.30557677616501144,
|
| 654 |
+
"grad_norm": 132.87548084265904,
|
| 655 |
+
"learning_rate": 3.862478777589134e-07,
|
| 656 |
+
"logits/chosen": -0.21234130859375,
|
| 657 |
+
"logits/rejected": -0.22281494736671448,
|
| 658 |
+
"logps/chosen": -596.5999755859375,
|
| 659 |
+
"logps/rejected": -545.7999877929688,
|
| 660 |
+
"loss": 0.5241,
|
| 661 |
+
"nll_loss": 1.0109374523162842,
|
| 662 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 663 |
+
"rewards/chosen": 1.0762450695037842,
|
| 664 |
+
"rewards/margins": 1.8078124523162842,
|
| 665 |
+
"rewards/rejected": -0.730297863483429,
|
| 666 |
+
"step": 400
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 0.31321619556913677,
|
| 670 |
+
"grad_norm": 94.35786051595302,
|
| 671 |
+
"learning_rate": 3.820033955857385e-07,
|
| 672 |
+
"logits/chosen": -0.16411133110523224,
|
| 673 |
+
"logits/rejected": -0.12357177585363388,
|
| 674 |
+
"logps/chosen": -545.5999755859375,
|
| 675 |
+
"logps/rejected": -495.0,
|
| 676 |
+
"loss": 0.4153,
|
| 677 |
+
"nll_loss": 0.915234386920929,
|
| 678 |
+
"rewards/accuracies": 0.737500011920929,
|
| 679 |
+
"rewards/chosen": 1.5988280773162842,
|
| 680 |
+
"rewards/margins": 2.166015625,
|
| 681 |
+
"rewards/rejected": -0.5667968988418579,
|
| 682 |
+
"step": 410
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.32085561497326204,
|
| 686 |
+
"grad_norm": 136.65656778443866,
|
| 687 |
+
"learning_rate": 3.7775891341256364e-07,
|
| 688 |
+
"logits/chosen": -0.20586547255516052,
|
| 689 |
+
"logits/rejected": -0.18314209580421448,
|
| 690 |
+
"logps/chosen": -649.5999755859375,
|
| 691 |
+
"logps/rejected": -575.4000244140625,
|
| 692 |
+
"loss": 0.3882,
|
| 693 |
+
"nll_loss": 0.908984363079071,
|
| 694 |
+
"rewards/accuracies": 0.737500011920929,
|
| 695 |
+
"rewards/chosen": 1.666894555091858,
|
| 696 |
+
"rewards/margins": 2.0859375,
|
| 697 |
+
"rewards/rejected": -0.41822510957717896,
|
| 698 |
+
"step": 420
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.3284950343773873,
|
| 702 |
+
"grad_norm": 245.401265321346,
|
| 703 |
+
"learning_rate": 3.735144312393888e-07,
|
| 704 |
+
"logits/chosen": -0.11831054836511612,
|
| 705 |
+
"logits/rejected": -0.17479248344898224,
|
| 706 |
+
"logps/chosen": -665.5999755859375,
|
| 707 |
+
"logps/rejected": -618.4000244140625,
|
| 708 |
+
"loss": 0.5723,
|
| 709 |
+
"nll_loss": 0.8828125,
|
| 710 |
+
"rewards/accuracies": 0.6875,
|
| 711 |
+
"rewards/chosen": 1.431249976158142,
|
| 712 |
+
"rewards/margins": 1.7414062023162842,
|
| 713 |
+
"rewards/rejected": -0.3086914122104645,
|
| 714 |
+
"step": 430
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.33613445378151263,
|
| 718 |
+
"grad_norm": 78.84105453853267,
|
| 719 |
+
"learning_rate": 3.692699490662139e-07,
|
| 720 |
+
"logits/chosen": -0.07949218899011612,
|
| 721 |
+
"logits/rejected": -0.07717285305261612,
|
| 722 |
+
"logps/chosen": -481.79998779296875,
|
| 723 |
+
"logps/rejected": -444.20001220703125,
|
| 724 |
+
"loss": 0.3382,
|
| 725 |
+
"nll_loss": 0.89453125,
|
| 726 |
+
"rewards/accuracies": 0.75,
|
| 727 |
+
"rewards/chosen": 1.306640625,
|
| 728 |
+
"rewards/margins": 2.328125,
|
| 729 |
+
"rewards/rejected": -1.0234375,
|
| 730 |
+
"step": 440
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.3437738731856379,
|
| 734 |
+
"grad_norm": 126.23202663934781,
|
| 735 |
+
"learning_rate": 3.65025466893039e-07,
|
| 736 |
+
"logits/chosen": -0.29511719942092896,
|
| 737 |
+
"logits/rejected": -0.2608398497104645,
|
| 738 |
+
"logps/chosen": -547.5999755859375,
|
| 739 |
+
"logps/rejected": -525.5999755859375,
|
| 740 |
+
"loss": 0.4856,
|
| 741 |
+
"nll_loss": 0.9339843988418579,
|
| 742 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 743 |
+
"rewards/chosen": 1.1731445789337158,
|
| 744 |
+
"rewards/margins": 1.9373047351837158,
|
| 745 |
+
"rewards/rejected": -0.7630859613418579,
|
| 746 |
+
"step": 450
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.35141329258976317,
|
| 750 |
+
"grad_norm": 76.23435884387047,
|
| 751 |
+
"learning_rate": 3.607809847198642e-07,
|
| 752 |
+
"logits/chosen": -0.3000244200229645,
|
| 753 |
+
"logits/rejected": -0.30839842557907104,
|
| 754 |
+
"logps/chosen": -572.4000244140625,
|
| 755 |
+
"logps/rejected": -551.7999877929688,
|
| 756 |
+
"loss": 0.3774,
|
| 757 |
+
"nll_loss": 0.9281250238418579,
|
| 758 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 759 |
+
"rewards/chosen": 2.0835938453674316,
|
| 760 |
+
"rewards/margins": 2.276562452316284,
|
| 761 |
+
"rewards/rejected": -0.19318847358226776,
|
| 762 |
+
"step": 460
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 0.35905271199388844,
|
| 766 |
+
"grad_norm": 114.53029164371587,
|
| 767 |
+
"learning_rate": 3.5653650254668926e-07,
|
| 768 |
+
"logits/chosen": -0.17070312798023224,
|
| 769 |
+
"logits/rejected": -0.16102905571460724,
|
| 770 |
+
"logps/chosen": -531.2000122070312,
|
| 771 |
+
"logps/rejected": -517.7999877929688,
|
| 772 |
+
"loss": 0.4004,
|
| 773 |
+
"nll_loss": 0.873046875,
|
| 774 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 775 |
+
"rewards/chosen": 1.2610352039337158,
|
| 776 |
+
"rewards/margins": 1.8546874523162842,
|
| 777 |
+
"rewards/rejected": -0.590991199016571,
|
| 778 |
+
"step": 470
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.36669213139801377,
|
| 782 |
+
"grad_norm": 243.54108183416537,
|
| 783 |
+
"learning_rate": 3.5229202037351443e-07,
|
| 784 |
+
"logits/chosen": -0.30506592988967896,
|
| 785 |
+
"logits/rejected": -0.24588623642921448,
|
| 786 |
+
"logps/chosen": -532.5999755859375,
|
| 787 |
+
"logps/rejected": -517.0,
|
| 788 |
+
"loss": 0.3923,
|
| 789 |
+
"nll_loss": 0.908203125,
|
| 790 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 791 |
+
"rewards/chosen": 1.15625,
|
| 792 |
+
"rewards/margins": 2.299999952316284,
|
| 793 |
+
"rewards/rejected": -1.144677758216858,
|
| 794 |
+
"step": 480
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.37433155080213903,
|
| 798 |
+
"grad_norm": 174.36984130018106,
|
| 799 |
+
"learning_rate": 3.4804753820033956e-07,
|
| 800 |
+
"logits/chosen": -0.29462891817092896,
|
| 801 |
+
"logits/rejected": -0.34052735567092896,
|
| 802 |
+
"logps/chosen": -650.7999877929688,
|
| 803 |
+
"logps/rejected": -604.5999755859375,
|
| 804 |
+
"loss": 0.3593,
|
| 805 |
+
"nll_loss": 0.997265636920929,
|
| 806 |
+
"rewards/accuracies": 0.824999988079071,
|
| 807 |
+
"rewards/chosen": 1.580468773841858,
|
| 808 |
+
"rewards/margins": 2.7640624046325684,
|
| 809 |
+
"rewards/rejected": -1.182226538658142,
|
| 810 |
+
"step": 490
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 0.3819709702062643,
|
| 814 |
+
"grad_norm": 161.0508842577597,
|
| 815 |
+
"learning_rate": 3.438030560271647e-07,
|
| 816 |
+
"logits/chosen": -0.2794433534145355,
|
| 817 |
+
"logits/rejected": -0.26513671875,
|
| 818 |
+
"logps/chosen": -607.2000122070312,
|
| 819 |
+
"logps/rejected": -620.0,
|
| 820 |
+
"loss": 0.4101,
|
| 821 |
+
"nll_loss": 0.9351562261581421,
|
| 822 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 823 |
+
"rewards/chosen": 1.4298827648162842,
|
| 824 |
+
"rewards/margins": 2.234375,
|
| 825 |
+
"rewards/rejected": -0.80517578125,
|
| 826 |
+
"step": 500
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 0.38961038961038963,
|
| 830 |
+
"grad_norm": 110.76591319339768,
|
| 831 |
+
"learning_rate": 3.395585738539898e-07,
|
| 832 |
+
"logits/chosen": -0.16807861626148224,
|
| 833 |
+
"logits/rejected": -0.12810058891773224,
|
| 834 |
+
"logps/chosen": -636.5999755859375,
|
| 835 |
+
"logps/rejected": -607.4000244140625,
|
| 836 |
+
"loss": 0.3684,
|
| 837 |
+
"nll_loss": 0.936718761920929,
|
| 838 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 839 |
+
"rewards/chosen": 0.9383300542831421,
|
| 840 |
+
"rewards/margins": 2.294921875,
|
| 841 |
+
"rewards/rejected": -1.353674292564392,
|
| 842 |
+
"step": 510
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.3972498090145149,
|
| 846 |
+
"grad_norm": 180.59465535461902,
|
| 847 |
+
"learning_rate": 3.3531409168081493e-07,
|
| 848 |
+
"logits/chosen": -0.18726806342601776,
|
| 849 |
+
"logits/rejected": -0.16033935546875,
|
| 850 |
+
"logps/chosen": -668.2000122070312,
|
| 851 |
+
"logps/rejected": -634.5999755859375,
|
| 852 |
+
"loss": 0.4376,
|
| 853 |
+
"nll_loss": 0.938671886920929,
|
| 854 |
+
"rewards/accuracies": 0.699999988079071,
|
| 855 |
+
"rewards/chosen": 1.7078125476837158,
|
| 856 |
+
"rewards/margins": 2.2789063453674316,
|
| 857 |
+
"rewards/rejected": -0.5711914300918579,
|
| 858 |
+
"step": 520
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 0.40488922841864017,
|
| 862 |
+
"grad_norm": 138.49457254902308,
|
| 863 |
+
"learning_rate": 3.3106960950764005e-07,
|
| 864 |
+
"logits/chosen": -0.2599121034145355,
|
| 865 |
+
"logits/rejected": -0.22050781548023224,
|
| 866 |
+
"logps/chosen": -661.5999755859375,
|
| 867 |
+
"logps/rejected": -634.5999755859375,
|
| 868 |
+
"loss": 0.4448,
|
| 869 |
+
"nll_loss": 1.0085937976837158,
|
| 870 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 871 |
+
"rewards/chosen": 0.7669922113418579,
|
| 872 |
+
"rewards/margins": 1.933203101158142,
|
| 873 |
+
"rewards/rejected": -1.1654784679412842,
|
| 874 |
+
"step": 530
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 0.4125286478227655,
|
| 878 |
+
"grad_norm": 231.1850522757914,
|
| 879 |
+
"learning_rate": 3.268251273344652e-07,
|
| 880 |
+
"logits/chosen": -0.17841796576976776,
|
| 881 |
+
"logits/rejected": -0.18508300185203552,
|
| 882 |
+
"logps/chosen": -628.2000122070312,
|
| 883 |
+
"logps/rejected": -584.2000122070312,
|
| 884 |
+
"loss": 0.4491,
|
| 885 |
+
"nll_loss": 1.060546875,
|
| 886 |
+
"rewards/accuracies": 0.6875,
|
| 887 |
+
"rewards/chosen": 0.76904296875,
|
| 888 |
+
"rewards/margins": 2.246875047683716,
|
| 889 |
+
"rewards/rejected": -1.480224609375,
|
| 890 |
+
"step": 540
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.42016806722689076,
|
| 894 |
+
"grad_norm": 160.7987831948394,
|
| 895 |
+
"learning_rate": 3.225806451612903e-07,
|
| 896 |
+
"logits/chosen": -0.14698180556297302,
|
| 897 |
+
"logits/rejected": -0.19716796278953552,
|
| 898 |
+
"logps/chosen": -602.2000122070312,
|
| 899 |
+
"logps/rejected": -539.7999877929688,
|
| 900 |
+
"loss": 0.3708,
|
| 901 |
+
"nll_loss": 0.871874988079071,
|
| 902 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 903 |
+
"rewards/chosen": 0.7313476800918579,
|
| 904 |
+
"rewards/margins": 2.765625,
|
| 905 |
+
"rewards/rejected": -2.0337891578674316,
|
| 906 |
+
"step": 550
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.42780748663101603,
|
| 910 |
+
"grad_norm": 90.82442844767456,
|
| 911 |
+
"learning_rate": 3.183361629881154e-07,
|
| 912 |
+
"logits/chosen": -0.16887207329273224,
|
| 913 |
+
"logits/rejected": -0.19123534858226776,
|
| 914 |
+
"logps/chosen": -600.2000122070312,
|
| 915 |
+
"logps/rejected": -548.5999755859375,
|
| 916 |
+
"loss": 0.4537,
|
| 917 |
+
"nll_loss": 0.9097656011581421,
|
| 918 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 919 |
+
"rewards/chosen": 0.8675781488418579,
|
| 920 |
+
"rewards/margins": 2.024218797683716,
|
| 921 |
+
"rewards/rejected": -1.1591796875,
|
| 922 |
+
"step": 560
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 0.43544690603514136,
|
| 926 |
+
"grad_norm": 138.92459223223193,
|
| 927 |
+
"learning_rate": 3.140916808149406e-07,
|
| 928 |
+
"logits/chosen": -0.11478271335363388,
|
| 929 |
+
"logits/rejected": -0.1641845703125,
|
| 930 |
+
"logps/chosen": -657.2000122070312,
|
| 931 |
+
"logps/rejected": -604.2000122070312,
|
| 932 |
+
"loss": 0.5563,
|
| 933 |
+
"nll_loss": 0.9136718511581421,
|
| 934 |
+
"rewards/accuracies": 0.675000011920929,
|
| 935 |
+
"rewards/chosen": 1.17626953125,
|
| 936 |
+
"rewards/margins": 1.947656273841858,
|
| 937 |
+
"rewards/rejected": -0.772656261920929,
|
| 938 |
+
"step": 570
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.4430863254392666,
|
| 942 |
+
"grad_norm": 123.1501820354294,
|
| 943 |
+
"learning_rate": 3.0984719864176567e-07,
|
| 944 |
+
"logits/chosen": -0.11948242038488388,
|
| 945 |
+
"logits/rejected": -0.10875244438648224,
|
| 946 |
+
"logps/chosen": -553.0,
|
| 947 |
+
"logps/rejected": -479.79998779296875,
|
| 948 |
+
"loss": 0.3869,
|
| 949 |
+
"nll_loss": 0.857421875,
|
| 950 |
+
"rewards/accuracies": 0.737500011920929,
|
| 951 |
+
"rewards/chosen": 1.2839844226837158,
|
| 952 |
+
"rewards/margins": 2.4769530296325684,
|
| 953 |
+
"rewards/rejected": -1.1936523914337158,
|
| 954 |
+
"step": 580
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.4507257448433919,
|
| 958 |
+
"grad_norm": 127.32337855269951,
|
| 959 |
+
"learning_rate": 3.056027164685908e-07,
|
| 960 |
+
"logits/chosen": -0.2039794921875,
|
| 961 |
+
"logits/rejected": -0.17943115532398224,
|
| 962 |
+
"logps/chosen": -564.4000244140625,
|
| 963 |
+
"logps/rejected": -573.7999877929688,
|
| 964 |
+
"loss": 0.4879,
|
| 965 |
+
"nll_loss": 0.991015613079071,
|
| 966 |
+
"rewards/accuracies": 0.6875,
|
| 967 |
+
"rewards/chosen": 0.6973632574081421,
|
| 968 |
+
"rewards/margins": 1.755468726158142,
|
| 969 |
+
"rewards/rejected": -1.059179663658142,
|
| 970 |
+
"step": 590
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 0.45836516424751717,
|
| 974 |
+
"grad_norm": 107.25927292937021,
|
| 975 |
+
"learning_rate": 3.0135823429541597e-07,
|
| 976 |
+
"logits/chosen": -0.18730469048023224,
|
| 977 |
+
"logits/rejected": -0.17886963486671448,
|
| 978 |
+
"logps/chosen": -597.2000122070312,
|
| 979 |
+
"logps/rejected": -592.0,
|
| 980 |
+
"loss": 0.4284,
|
| 981 |
+
"nll_loss": 1.019140601158142,
|
| 982 |
+
"rewards/accuracies": 0.699999988079071,
|
| 983 |
+
"rewards/chosen": 0.826953113079071,
|
| 984 |
+
"rewards/margins": 2.2720704078674316,
|
| 985 |
+
"rewards/rejected": -1.450781226158142,
|
| 986 |
+
"step": 600
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 0.4660045836516425,
|
| 990 |
+
"grad_norm": 181.85392573395544,
|
| 991 |
+
"learning_rate": 2.9711375212224104e-07,
|
| 992 |
+
"logits/chosen": -0.10260009765625,
|
| 993 |
+
"logits/rejected": -0.1400146484375,
|
| 994 |
+
"logps/chosen": -596.7999877929688,
|
| 995 |
+
"logps/rejected": -581.5999755859375,
|
| 996 |
+
"loss": 0.4928,
|
| 997 |
+
"nll_loss": 0.938281238079071,
|
| 998 |
+
"rewards/accuracies": 0.6875,
|
| 999 |
+
"rewards/chosen": 0.825878918170929,
|
| 1000 |
+
"rewards/margins": 1.986718773841858,
|
| 1001 |
+
"rewards/rejected": -1.161718726158142,
|
| 1002 |
+
"step": 610
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 0.47364400305576776,
|
| 1006 |
+
"grad_norm": 498.482891329986,
|
| 1007 |
+
"learning_rate": 2.928692699490662e-07,
|
| 1008 |
+
"logits/chosen": -0.32451170682907104,
|
| 1009 |
+
"logits/rejected": -0.33247071504592896,
|
| 1010 |
+
"logps/chosen": -636.5999755859375,
|
| 1011 |
+
"logps/rejected": -602.7999877929688,
|
| 1012 |
+
"loss": 0.4174,
|
| 1013 |
+
"nll_loss": 1.0402343273162842,
|
| 1014 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1015 |
+
"rewards/chosen": 0.42646485567092896,
|
| 1016 |
+
"rewards/margins": 2.5859375,
|
| 1017 |
+
"rewards/rejected": -2.1615233421325684,
|
| 1018 |
+
"step": 620
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.48128342245989303,
|
| 1022 |
+
"grad_norm": 90.59697238922614,
|
| 1023 |
+
"learning_rate": 2.8862478777589134e-07,
|
| 1024 |
+
"logits/chosen": -0.2081756591796875,
|
| 1025 |
+
"logits/rejected": -0.24516601860523224,
|
| 1026 |
+
"logps/chosen": -705.7999877929688,
|
| 1027 |
+
"logps/rejected": -635.5999755859375,
|
| 1028 |
+
"loss": 0.475,
|
| 1029 |
+
"nll_loss": 0.9750000238418579,
|
| 1030 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1031 |
+
"rewards/chosen": 0.801025390625,
|
| 1032 |
+
"rewards/margins": 2.24609375,
|
| 1033 |
+
"rewards/rejected": -1.4406249523162842,
|
| 1034 |
+
"step": 630
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.48892284186401835,
|
| 1038 |
+
"grad_norm": 70.07533244919708,
|
| 1039 |
+
"learning_rate": 2.8438030560271646e-07,
|
| 1040 |
+
"logits/chosen": -0.23149414360523224,
|
| 1041 |
+
"logits/rejected": -0.24143067002296448,
|
| 1042 |
+
"logps/chosen": -601.7999877929688,
|
| 1043 |
+
"logps/rejected": -607.0,
|
| 1044 |
+
"loss": 0.4966,
|
| 1045 |
+
"nll_loss": 1.02734375,
|
| 1046 |
+
"rewards/accuracies": 0.6875,
|
| 1047 |
+
"rewards/chosen": 0.388671875,
|
| 1048 |
+
"rewards/margins": 2.166015625,
|
| 1049 |
+
"rewards/rejected": -1.7734375,
|
| 1050 |
+
"step": 640
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.4965622612681436,
|
| 1054 |
+
"grad_norm": 155.4435454773161,
|
| 1055 |
+
"learning_rate": 2.801358234295416e-07,
|
| 1056 |
+
"logits/chosen": -0.12587890028953552,
|
| 1057 |
+
"logits/rejected": -0.12092284858226776,
|
| 1058 |
+
"logps/chosen": -496.6000061035156,
|
| 1059 |
+
"logps/rejected": -498.20001220703125,
|
| 1060 |
+
"loss": 0.4247,
|
| 1061 |
+
"nll_loss": 0.916015625,
|
| 1062 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1063 |
+
"rewards/chosen": 0.5431762933731079,
|
| 1064 |
+
"rewards/margins": 2.1624999046325684,
|
| 1065 |
+
"rewards/rejected": -1.619140625,
|
| 1066 |
+
"step": 650
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.5042016806722689,
|
| 1070 |
+
"grad_norm": 262.5809391278708,
|
| 1071 |
+
"learning_rate": 2.758913412563667e-07,
|
| 1072 |
+
"logits/chosen": -0.12770995497703552,
|
| 1073 |
+
"logits/rejected": -0.14760741591453552,
|
| 1074 |
+
"logps/chosen": -530.4000244140625,
|
| 1075 |
+
"logps/rejected": -482.0,
|
| 1076 |
+
"loss": 0.5208,
|
| 1077 |
+
"nll_loss": 0.903515636920929,
|
| 1078 |
+
"rewards/accuracies": 0.6875,
|
| 1079 |
+
"rewards/chosen": 1.02490234375,
|
| 1080 |
+
"rewards/margins": 1.9609375,
|
| 1081 |
+
"rewards/rejected": -0.9371093511581421,
|
| 1082 |
+
"step": 660
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.5118411000763942,
|
| 1086 |
+
"grad_norm": 116.54142626086018,
|
| 1087 |
+
"learning_rate": 2.7164685908319183e-07,
|
| 1088 |
+
"logits/chosen": -0.09782715141773224,
|
| 1089 |
+
"logits/rejected": -0.11018066108226776,
|
| 1090 |
+
"logps/chosen": -588.2000122070312,
|
| 1091 |
+
"logps/rejected": -562.0,
|
| 1092 |
+
"loss": 0.4879,
|
| 1093 |
+
"nll_loss": 0.98828125,
|
| 1094 |
+
"rewards/accuracies": 0.6875,
|
| 1095 |
+
"rewards/chosen": 1.236718773841858,
|
| 1096 |
+
"rewards/margins": 2.2178711891174316,
|
| 1097 |
+
"rewards/rejected": -0.982714831829071,
|
| 1098 |
+
"step": 670
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 0.5194805194805194,
|
| 1102 |
+
"grad_norm": 147.62557960833126,
|
| 1103 |
+
"learning_rate": 2.67402376910017e-07,
|
| 1104 |
+
"logits/chosen": -0.2564453184604645,
|
| 1105 |
+
"logits/rejected": -0.24064941704273224,
|
| 1106 |
+
"logps/chosen": -534.0,
|
| 1107 |
+
"logps/rejected": -507.79998779296875,
|
| 1108 |
+
"loss": 0.3249,
|
| 1109 |
+
"nll_loss": 0.9417968988418579,
|
| 1110 |
+
"rewards/accuracies": 0.8125,
|
| 1111 |
+
"rewards/chosen": 0.764941394329071,
|
| 1112 |
+
"rewards/margins": 2.680859327316284,
|
| 1113 |
+
"rewards/rejected": -1.916015625,
|
| 1114 |
+
"step": 680
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 0.5271199388846448,
|
| 1118 |
+
"grad_norm": 73.38717469252549,
|
| 1119 |
+
"learning_rate": 2.631578947368421e-07,
|
| 1120 |
+
"logits/chosen": -0.169677734375,
|
| 1121 |
+
"logits/rejected": -0.21303710341453552,
|
| 1122 |
+
"logps/chosen": -604.7999877929688,
|
| 1123 |
+
"logps/rejected": -589.5999755859375,
|
| 1124 |
+
"loss": 0.5577,
|
| 1125 |
+
"nll_loss": 0.919921875,
|
| 1126 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1127 |
+
"rewards/chosen": 0.28046876192092896,
|
| 1128 |
+
"rewards/margins": 1.9656250476837158,
|
| 1129 |
+
"rewards/rejected": -1.6857421398162842,
|
| 1130 |
+
"step": 690
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.5347593582887701,
|
| 1134 |
+
"grad_norm": 125.30928463570962,
|
| 1135 |
+
"learning_rate": 2.589134125636672e-07,
|
| 1136 |
+
"logits/chosen": -0.24868163466453552,
|
| 1137 |
+
"logits/rejected": -0.21337890625,
|
| 1138 |
+
"logps/chosen": -669.7999877929688,
|
| 1139 |
+
"logps/rejected": -607.0,
|
| 1140 |
+
"loss": 0.5241,
|
| 1141 |
+
"nll_loss": 0.987109363079071,
|
| 1142 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1143 |
+
"rewards/chosen": 0.698010265827179,
|
| 1144 |
+
"rewards/margins": 2.317578077316284,
|
| 1145 |
+
"rewards/rejected": -1.62060546875,
|
| 1146 |
+
"step": 700
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 0.5423987776928954,
|
| 1150 |
+
"grad_norm": 86.47824062520769,
|
| 1151 |
+
"learning_rate": 2.546689303904924e-07,
|
| 1152 |
+
"logits/chosen": -0.15998534858226776,
|
| 1153 |
+
"logits/rejected": -0.13544921576976776,
|
| 1154 |
+
"logps/chosen": -527.0,
|
| 1155 |
+
"logps/rejected": -507.3999938964844,
|
| 1156 |
+
"loss": 0.4433,
|
| 1157 |
+
"nll_loss": 0.8832031488418579,
|
| 1158 |
+
"rewards/accuracies": 0.75,
|
| 1159 |
+
"rewards/chosen": 0.833325207233429,
|
| 1160 |
+
"rewards/margins": 1.917578101158142,
|
| 1161 |
+
"rewards/rejected": -1.0823242664337158,
|
| 1162 |
+
"step": 710
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 0.5500381970970206,
|
| 1166 |
+
"grad_norm": 241.9764128861329,
|
| 1167 |
+
"learning_rate": 2.5042444821731745e-07,
|
| 1168 |
+
"logits/chosen": -0.23203125596046448,
|
| 1169 |
+
"logits/rejected": -0.15019531548023224,
|
| 1170 |
+
"logps/chosen": -623.2000122070312,
|
| 1171 |
+
"logps/rejected": -519.4000244140625,
|
| 1172 |
+
"loss": 0.4256,
|
| 1173 |
+
"nll_loss": 0.934374988079071,
|
| 1174 |
+
"rewards/accuracies": 0.75,
|
| 1175 |
+
"rewards/chosen": 0.8759765625,
|
| 1176 |
+
"rewards/margins": 2.4292969703674316,
|
| 1177 |
+
"rewards/rejected": -1.5539062023162842,
|
| 1178 |
+
"step": 720
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.5576776165011459,
|
| 1182 |
+
"grad_norm": 199.54510567975194,
|
| 1183 |
+
"learning_rate": 2.4617996604414257e-07,
|
| 1184 |
+
"logits/chosen": -0.20654296875,
|
| 1185 |
+
"logits/rejected": -0.18228760361671448,
|
| 1186 |
+
"logps/chosen": -569.5999755859375,
|
| 1187 |
+
"logps/rejected": -539.2000122070312,
|
| 1188 |
+
"loss": 0.44,
|
| 1189 |
+
"nll_loss": 0.9097656011581421,
|
| 1190 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1191 |
+
"rewards/chosen": 0.22563476860523224,
|
| 1192 |
+
"rewards/margins": 2.035937547683716,
|
| 1193 |
+
"rewards/rejected": -1.8076171875,
|
| 1194 |
+
"step": 730
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.5653170359052712,
|
| 1198 |
+
"grad_norm": 57.705062448013685,
|
| 1199 |
+
"learning_rate": 2.4193548387096775e-07,
|
| 1200 |
+
"logits/chosen": -0.29963380098342896,
|
| 1201 |
+
"logits/rejected": -0.2674804627895355,
|
| 1202 |
+
"logps/chosen": -658.4000244140625,
|
| 1203 |
+
"logps/rejected": -642.4000244140625,
|
| 1204 |
+
"loss": 0.395,
|
| 1205 |
+
"nll_loss": 0.998046875,
|
| 1206 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1207 |
+
"rewards/chosen": -0.02570800855755806,
|
| 1208 |
+
"rewards/margins": 2.359375,
|
| 1209 |
+
"rewards/rejected": -2.383593797683716,
|
| 1210 |
+
"step": 740
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 0.5729564553093965,
|
| 1214 |
+
"grad_norm": 90.1800130721227,
|
| 1215 |
+
"learning_rate": 2.3769100169779285e-07,
|
| 1216 |
+
"logits/chosen": -0.17238768935203552,
|
| 1217 |
+
"logits/rejected": -0.21173095703125,
|
| 1218 |
+
"logps/chosen": -690.4000244140625,
|
| 1219 |
+
"logps/rejected": -671.0,
|
| 1220 |
+
"loss": 0.4136,
|
| 1221 |
+
"nll_loss": 0.9566406011581421,
|
| 1222 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1223 |
+
"rewards/chosen": 0.26005858182907104,
|
| 1224 |
+
"rewards/margins": 2.177929639816284,
|
| 1225 |
+
"rewards/rejected": -1.918359398841858,
|
| 1226 |
+
"step": 750
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 0.5805958747135218,
|
| 1230 |
+
"grad_norm": 128.16902765016945,
|
| 1231 |
+
"learning_rate": 2.33446519524618e-07,
|
| 1232 |
+
"logits/chosen": -0.18836669623851776,
|
| 1233 |
+
"logits/rejected": -0.13487549126148224,
|
| 1234 |
+
"logps/chosen": -570.0,
|
| 1235 |
+
"logps/rejected": -552.4000244140625,
|
| 1236 |
+
"loss": 0.4234,
|
| 1237 |
+
"nll_loss": 0.950390636920929,
|
| 1238 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1239 |
+
"rewards/chosen": 0.518481433391571,
|
| 1240 |
+
"rewards/margins": 2.1351561546325684,
|
| 1241 |
+
"rewards/rejected": -1.615625023841858,
|
| 1242 |
+
"step": 760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.5882352941176471,
|
| 1246 |
+
"grad_norm": 179.91151487829103,
|
| 1247 |
+
"learning_rate": 2.2920203735144312e-07,
|
| 1248 |
+
"logits/chosen": -0.15025635063648224,
|
| 1249 |
+
"logits/rejected": -0.20454101264476776,
|
| 1250 |
+
"logps/chosen": -590.5999755859375,
|
| 1251 |
+
"logps/rejected": -570.7999877929688,
|
| 1252 |
+
"loss": 0.3452,
|
| 1253 |
+
"nll_loss": 0.905078113079071,
|
| 1254 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1255 |
+
"rewards/chosen": 1.132714867591858,
|
| 1256 |
+
"rewards/margins": 2.6820311546325684,
|
| 1257 |
+
"rewards/rejected": -1.5529296398162842,
|
| 1258 |
+
"step": 770
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 0.5958747135217723,
|
| 1262 |
+
"grad_norm": 147.28111171420568,
|
| 1263 |
+
"learning_rate": 2.2495755517826824e-07,
|
| 1264 |
+
"logits/chosen": -0.2244003266096115,
|
| 1265 |
+
"logits/rejected": -0.20163574814796448,
|
| 1266 |
+
"logps/chosen": -580.0,
|
| 1267 |
+
"logps/rejected": -517.0,
|
| 1268 |
+
"loss": 0.4638,
|
| 1269 |
+
"nll_loss": 0.9203125238418579,
|
| 1270 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1271 |
+
"rewards/chosen": 1.217919945716858,
|
| 1272 |
+
"rewards/margins": 2.2210936546325684,
|
| 1273 |
+
"rewards/rejected": -1.00244140625,
|
| 1274 |
+
"step": 780
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 0.6035141329258976,
|
| 1278 |
+
"grad_norm": 148.2346332858827,
|
| 1279 |
+
"learning_rate": 2.2071307300509337e-07,
|
| 1280 |
+
"logits/chosen": -0.2839111387729645,
|
| 1281 |
+
"logits/rejected": -0.26567381620407104,
|
| 1282 |
+
"logps/chosen": -589.5999755859375,
|
| 1283 |
+
"logps/rejected": -557.0,
|
| 1284 |
+
"loss": 0.3842,
|
| 1285 |
+
"nll_loss": 0.866406261920929,
|
| 1286 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1287 |
+
"rewards/chosen": 0.83984375,
|
| 1288 |
+
"rewards/margins": 2.340625047683716,
|
| 1289 |
+
"rewards/rejected": -1.5037109851837158,
|
| 1290 |
+
"step": 790
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.6111535523300229,
|
| 1294 |
+
"grad_norm": 88.75284081371973,
|
| 1295 |
+
"learning_rate": 2.164685908319185e-07,
|
| 1296 |
+
"logits/chosen": -0.23496094346046448,
|
| 1297 |
+
"logits/rejected": -0.19589844346046448,
|
| 1298 |
+
"logps/chosen": -686.0,
|
| 1299 |
+
"logps/rejected": -661.2000122070312,
|
| 1300 |
+
"loss": 0.4166,
|
| 1301 |
+
"nll_loss": 1.068359375,
|
| 1302 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1303 |
+
"rewards/chosen": 1.013671875,
|
| 1304 |
+
"rewards/margins": 2.171875,
|
| 1305 |
+
"rewards/rejected": -1.1590087413787842,
|
| 1306 |
+
"step": 800
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.6187929717341482,
|
| 1310 |
+
"grad_norm": 145.23948778484626,
|
| 1311 |
+
"learning_rate": 2.1222410865874364e-07,
|
| 1312 |
+
"logits/chosen": -0.20212402939796448,
|
| 1313 |
+
"logits/rejected": -0.20252685248851776,
|
| 1314 |
+
"logps/chosen": -539.0,
|
| 1315 |
+
"logps/rejected": -484.6000061035156,
|
| 1316 |
+
"loss": 0.3318,
|
| 1317 |
+
"nll_loss": 0.8890625238418579,
|
| 1318 |
+
"rewards/accuracies": 0.8125,
|
| 1319 |
+
"rewards/chosen": 0.648242175579071,
|
| 1320 |
+
"rewards/margins": 2.7249999046325684,
|
| 1321 |
+
"rewards/rejected": -2.0816407203674316,
|
| 1322 |
+
"step": 810
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 0.6264323911382735,
|
| 1326 |
+
"grad_norm": 128.52828674770328,
|
| 1327 |
+
"learning_rate": 2.0797962648556874e-07,
|
| 1328 |
+
"logits/chosen": -0.11715392768383026,
|
| 1329 |
+
"logits/rejected": -0.13743896782398224,
|
| 1330 |
+
"logps/chosen": -547.0,
|
| 1331 |
+
"logps/rejected": -538.2000122070312,
|
| 1332 |
+
"loss": 0.4387,
|
| 1333 |
+
"nll_loss": 0.813671886920929,
|
| 1334 |
+
"rewards/accuracies": 0.75,
|
| 1335 |
+
"rewards/chosen": 1.023046851158142,
|
| 1336 |
+
"rewards/margins": 2.3441405296325684,
|
| 1337 |
+
"rewards/rejected": -1.322265625,
|
| 1338 |
+
"step": 820
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 0.6340718105423988,
|
| 1342 |
+
"grad_norm": 144.47861170444614,
|
| 1343 |
+
"learning_rate": 2.037351443123939e-07,
|
| 1344 |
+
"logits/chosen": -0.06186523288488388,
|
| 1345 |
+
"logits/rejected": -0.10273437201976776,
|
| 1346 |
+
"logps/chosen": -533.4000244140625,
|
| 1347 |
+
"logps/rejected": -495.0,
|
| 1348 |
+
"loss": 0.3348,
|
| 1349 |
+
"nll_loss": 0.8785156011581421,
|
| 1350 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1351 |
+
"rewards/chosen": 0.628613293170929,
|
| 1352 |
+
"rewards/margins": 2.4515624046325684,
|
| 1353 |
+
"rewards/rejected": -1.8227050304412842,
|
| 1354 |
+
"step": 830
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.6417112299465241,
|
| 1358 |
+
"grad_norm": 231.8284171525199,
|
| 1359 |
+
"learning_rate": 1.99490662139219e-07,
|
| 1360 |
+
"logits/chosen": -0.2966064512729645,
|
| 1361 |
+
"logits/rejected": -0.3492675721645355,
|
| 1362 |
+
"logps/chosen": -563.2000122070312,
|
| 1363 |
+
"logps/rejected": -554.5999755859375,
|
| 1364 |
+
"loss": 0.4784,
|
| 1365 |
+
"nll_loss": 1.038671851158142,
|
| 1366 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1367 |
+
"rewards/chosen": 0.3266967833042145,
|
| 1368 |
+
"rewards/margins": 2.3394532203674316,
|
| 1369 |
+
"rewards/rejected": -2.016064405441284,
|
| 1370 |
+
"step": 840
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 0.6493506493506493,
|
| 1374 |
+
"grad_norm": 95.18920210483807,
|
| 1375 |
+
"learning_rate": 1.9524617996604413e-07,
|
| 1376 |
+
"logits/chosen": -0.2720703184604645,
|
| 1377 |
+
"logits/rejected": -0.21528320014476776,
|
| 1378 |
+
"logps/chosen": -639.4000244140625,
|
| 1379 |
+
"logps/rejected": -599.4000244140625,
|
| 1380 |
+
"loss": 0.4277,
|
| 1381 |
+
"nll_loss": 0.913281261920929,
|
| 1382 |
+
"rewards/accuracies": 0.75,
|
| 1383 |
+
"rewards/chosen": 1.0625,
|
| 1384 |
+
"rewards/margins": 2.268261671066284,
|
| 1385 |
+
"rewards/rejected": -1.204687476158142,
|
| 1386 |
+
"step": 850
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 0.6569900687547746,
|
| 1390 |
+
"grad_norm": 179.17446116841967,
|
| 1391 |
+
"learning_rate": 1.9100169779286926e-07,
|
| 1392 |
+
"logits/chosen": -0.28715819120407104,
|
| 1393 |
+
"logits/rejected": -0.2583984434604645,
|
| 1394 |
+
"logps/chosen": -548.5999755859375,
|
| 1395 |
+
"logps/rejected": -489.3999938964844,
|
| 1396 |
+
"loss": 0.4577,
|
| 1397 |
+
"nll_loss": 0.846875011920929,
|
| 1398 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1399 |
+
"rewards/chosen": 1.1994140148162842,
|
| 1400 |
+
"rewards/margins": 2.366015672683716,
|
| 1401 |
+
"rewards/rejected": -1.167578101158142,
|
| 1402 |
+
"step": 860
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.6646294881588999,
|
| 1406 |
+
"grad_norm": 176.32245150389784,
|
| 1407 |
+
"learning_rate": 1.867572156196944e-07,
|
| 1408 |
+
"logits/chosen": -0.2516418397426605,
|
| 1409 |
+
"logits/rejected": -0.24804076552391052,
|
| 1410 |
+
"logps/chosen": -462.0,
|
| 1411 |
+
"logps/rejected": -476.0,
|
| 1412 |
+
"loss": 0.5009,
|
| 1413 |
+
"nll_loss": 0.921875,
|
| 1414 |
+
"rewards/accuracies": 0.6875,
|
| 1415 |
+
"rewards/chosen": 0.588671863079071,
|
| 1416 |
+
"rewards/margins": 2.198046922683716,
|
| 1417 |
+
"rewards/rejected": -1.612036108970642,
|
| 1418 |
+
"step": 870
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 0.6722689075630253,
|
| 1422 |
+
"grad_norm": 154.3222829630188,
|
| 1423 |
+
"learning_rate": 1.825127334465195e-07,
|
| 1424 |
+
"logits/chosen": -0.10764160007238388,
|
| 1425 |
+
"logits/rejected": -0.10710449516773224,
|
| 1426 |
+
"logps/chosen": -545.4000244140625,
|
| 1427 |
+
"logps/rejected": -537.0,
|
| 1428 |
+
"loss": 0.4212,
|
| 1429 |
+
"nll_loss": 0.8804687261581421,
|
| 1430 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1431 |
+
"rewards/chosen": 0.8570312261581421,
|
| 1432 |
+
"rewards/margins": 2.375781297683716,
|
| 1433 |
+
"rewards/rejected": -1.517333984375,
|
| 1434 |
+
"step": 880
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 0.6799083269671505,
|
| 1438 |
+
"grad_norm": 90.0921653237205,
|
| 1439 |
+
"learning_rate": 1.7826825127334463e-07,
|
| 1440 |
+
"logits/chosen": -0.2685546875,
|
| 1441 |
+
"logits/rejected": -0.24018554389476776,
|
| 1442 |
+
"logps/chosen": -573.0,
|
| 1443 |
+
"logps/rejected": -550.0,
|
| 1444 |
+
"loss": 0.3369,
|
| 1445 |
+
"nll_loss": 0.922656238079071,
|
| 1446 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1447 |
+
"rewards/chosen": 0.4517578184604645,
|
| 1448 |
+
"rewards/margins": 2.914843797683716,
|
| 1449 |
+
"rewards/rejected": -2.46484375,
|
| 1450 |
+
"step": 890
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 0.6875477463712758,
|
| 1454 |
+
"grad_norm": 90.60087330499096,
|
| 1455 |
+
"learning_rate": 1.7402376910016978e-07,
|
| 1456 |
+
"logits/chosen": -0.17197266221046448,
|
| 1457 |
+
"logits/rejected": -0.16635742783546448,
|
| 1458 |
+
"logps/chosen": -646.5999755859375,
|
| 1459 |
+
"logps/rejected": -654.0,
|
| 1460 |
+
"loss": 0.5631,
|
| 1461 |
+
"nll_loss": 1.0390625,
|
| 1462 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1463 |
+
"rewards/chosen": 0.09462890774011612,
|
| 1464 |
+
"rewards/margins": 2.5414061546325684,
|
| 1465 |
+
"rewards/rejected": -2.447265625,
|
| 1466 |
+
"step": 900
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.6951871657754011,
|
| 1470 |
+
"grad_norm": 411.9292153674767,
|
| 1471 |
+
"learning_rate": 1.697792869269949e-07,
|
| 1472 |
+
"logits/chosen": -0.21950682997703552,
|
| 1473 |
+
"logits/rejected": -0.14067383110523224,
|
| 1474 |
+
"logps/chosen": -782.4000244140625,
|
| 1475 |
+
"logps/rejected": -746.5999755859375,
|
| 1476 |
+
"loss": 0.5727,
|
| 1477 |
+
"nll_loss": 0.967968761920929,
|
| 1478 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1479 |
+
"rewards/chosen": 0.3875488340854645,
|
| 1480 |
+
"rewards/margins": 1.941015601158142,
|
| 1481 |
+
"rewards/rejected": -1.554296851158142,
|
| 1482 |
+
"step": 910
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 0.7028265851795263,
|
| 1486 |
+
"grad_norm": 96.59177093218652,
|
| 1487 |
+
"learning_rate": 1.6553480475382003e-07,
|
| 1488 |
+
"logits/chosen": -0.17082519829273224,
|
| 1489 |
+
"logits/rejected": -0.24100342392921448,
|
| 1490 |
+
"logps/chosen": -523.4000244140625,
|
| 1491 |
+
"logps/rejected": -496.79998779296875,
|
| 1492 |
+
"loss": 0.4115,
|
| 1493 |
+
"nll_loss": 0.9515625238418579,
|
| 1494 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1495 |
+
"rewards/chosen": 0.02167968824505806,
|
| 1496 |
+
"rewards/margins": 2.335156202316284,
|
| 1497 |
+
"rewards/rejected": -2.3140625953674316,
|
| 1498 |
+
"step": 920
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 0.7104660045836516,
|
| 1502 |
+
"grad_norm": 114.00603134519599,
|
| 1503 |
+
"learning_rate": 1.6129032258064515e-07,
|
| 1504 |
+
"logits/chosen": -0.0623779296875,
|
| 1505 |
+
"logits/rejected": -0.03818359225988388,
|
| 1506 |
+
"logps/chosen": -604.7999877929688,
|
| 1507 |
+
"logps/rejected": -563.4000244140625,
|
| 1508 |
+
"loss": 0.374,
|
| 1509 |
+
"nll_loss": 0.859375,
|
| 1510 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1511 |
+
"rewards/chosen": 0.599719226360321,
|
| 1512 |
+
"rewards/margins": 2.612499952316284,
|
| 1513 |
+
"rewards/rejected": -2.0126953125,
|
| 1514 |
+
"step": 930
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.7181054239877769,
|
| 1518 |
+
"grad_norm": 46.449877624622495,
|
| 1519 |
+
"learning_rate": 1.570458404074703e-07,
|
| 1520 |
+
"logits/chosen": -0.20942382514476776,
|
| 1521 |
+
"logits/rejected": -0.23570556938648224,
|
| 1522 |
+
"logps/chosen": -527.4000244140625,
|
| 1523 |
+
"logps/rejected": -528.4000244140625,
|
| 1524 |
+
"loss": 0.3851,
|
| 1525 |
+
"nll_loss": 0.9398437738418579,
|
| 1526 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1527 |
+
"rewards/chosen": 0.19345703721046448,
|
| 1528 |
+
"rewards/margins": 2.53125,
|
| 1529 |
+
"rewards/rejected": -2.3375000953674316,
|
| 1530 |
+
"step": 940
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 0.7257448433919023,
|
| 1534 |
+
"grad_norm": 154.8870158247193,
|
| 1535 |
+
"learning_rate": 1.528013582342954e-07,
|
| 1536 |
+
"logits/chosen": -0.17775878310203552,
|
| 1537 |
+
"logits/rejected": -0.15126952528953552,
|
| 1538 |
+
"logps/chosen": -644.0,
|
| 1539 |
+
"logps/rejected": -595.4000244140625,
|
| 1540 |
+
"loss": 0.3593,
|
| 1541 |
+
"nll_loss": 0.948046863079071,
|
| 1542 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1543 |
+
"rewards/chosen": 0.614453136920929,
|
| 1544 |
+
"rewards/margins": 2.4710936546325684,
|
| 1545 |
+
"rewards/rejected": -1.8585937023162842,
|
| 1546 |
+
"step": 950
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 0.7333842627960275,
|
| 1550 |
+
"grad_norm": 60.309689191140706,
|
| 1551 |
+
"learning_rate": 1.4855687606112052e-07,
|
| 1552 |
+
"logits/chosen": -0.20181885361671448,
|
| 1553 |
+
"logits/rejected": -0.17143554985523224,
|
| 1554 |
+
"logps/chosen": -614.2000122070312,
|
| 1555 |
+
"logps/rejected": -502.20001220703125,
|
| 1556 |
+
"loss": 0.2511,
|
| 1557 |
+
"nll_loss": 0.9703124761581421,
|
| 1558 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1559 |
+
"rewards/chosen": 0.37763673067092896,
|
| 1560 |
+
"rewards/margins": 2.9703125953674316,
|
| 1561 |
+
"rewards/rejected": -2.592968702316284,
|
| 1562 |
+
"step": 960
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 0.7410236822001528,
|
| 1566 |
+
"grad_norm": 93.06951737576422,
|
| 1567 |
+
"learning_rate": 1.4431239388794567e-07,
|
| 1568 |
+
"logits/chosen": -0.11916504055261612,
|
| 1569 |
+
"logits/rejected": -0.12851563096046448,
|
| 1570 |
+
"logps/chosen": -522.0,
|
| 1571 |
+
"logps/rejected": -478.20001220703125,
|
| 1572 |
+
"loss": 0.3418,
|
| 1573 |
+
"nll_loss": 0.948046863079071,
|
| 1574 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1575 |
+
"rewards/chosen": 0.33341675996780396,
|
| 1576 |
+
"rewards/margins": 2.641406297683716,
|
| 1577 |
+
"rewards/rejected": -2.311718702316284,
|
| 1578 |
+
"step": 970
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.7486631016042781,
|
| 1582 |
+
"grad_norm": 86.2530129397837,
|
| 1583 |
+
"learning_rate": 1.400679117147708e-07,
|
| 1584 |
+
"logits/chosen": -0.26170653104782104,
|
| 1585 |
+
"logits/rejected": -0.27277833223342896,
|
| 1586 |
+
"logps/chosen": -557.7999877929688,
|
| 1587 |
+
"logps/rejected": -540.5999755859375,
|
| 1588 |
+
"loss": 0.3527,
|
| 1589 |
+
"nll_loss": 0.98046875,
|
| 1590 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1591 |
+
"rewards/chosen": 0.4949707090854645,
|
| 1592 |
+
"rewards/margins": 2.564453125,
|
| 1593 |
+
"rewards/rejected": -2.0708985328674316,
|
| 1594 |
+
"step": 980
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 0.7563025210084033,
|
| 1598 |
+
"grad_norm": 87.12906010791761,
|
| 1599 |
+
"learning_rate": 1.3582342954159592e-07,
|
| 1600 |
+
"logits/chosen": -0.26884764432907104,
|
| 1601 |
+
"logits/rejected": -0.22690430283546448,
|
| 1602 |
+
"logps/chosen": -634.4000244140625,
|
| 1603 |
+
"logps/rejected": -615.5999755859375,
|
| 1604 |
+
"loss": 0.3518,
|
| 1605 |
+
"nll_loss": 1.00390625,
|
| 1606 |
+
"rewards/accuracies": 0.75,
|
| 1607 |
+
"rewards/chosen": 0.05859375,
|
| 1608 |
+
"rewards/margins": 3.135937452316284,
|
| 1609 |
+
"rewards/rejected": -3.0796875953674316,
|
| 1610 |
+
"step": 990
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 0.7639419404125286,
|
| 1614 |
+
"grad_norm": 245.3625476453072,
|
| 1615 |
+
"learning_rate": 1.3157894736842104e-07,
|
| 1616 |
+
"logits/chosen": -0.20460815727710724,
|
| 1617 |
+
"logits/rejected": -0.16091307997703552,
|
| 1618 |
+
"logps/chosen": -667.5999755859375,
|
| 1619 |
+
"logps/rejected": -630.4000244140625,
|
| 1620 |
+
"loss": 0.5093,
|
| 1621 |
+
"nll_loss": 0.916015625,
|
| 1622 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1623 |
+
"rewards/chosen": -0.29121094942092896,
|
| 1624 |
+
"rewards/margins": 2.0160155296325684,
|
| 1625 |
+
"rewards/rejected": -2.307812452316284,
|
| 1626 |
+
"step": 1000
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.771581359816654,
|
| 1630 |
+
"grad_norm": 124.28190362733702,
|
| 1631 |
+
"learning_rate": 1.273344651952462e-07,
|
| 1632 |
+
"logits/chosen": -0.11392822116613388,
|
| 1633 |
+
"logits/rejected": -0.11372070014476776,
|
| 1634 |
+
"logps/chosen": -574.0,
|
| 1635 |
+
"logps/rejected": -589.5999755859375,
|
| 1636 |
+
"loss": 0.2951,
|
| 1637 |
+
"nll_loss": 0.9234374761581421,
|
| 1638 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1639 |
+
"rewards/chosen": -0.04301757737994194,
|
| 1640 |
+
"rewards/margins": 2.9124999046325684,
|
| 1641 |
+
"rewards/rejected": -2.962890625,
|
| 1642 |
+
"step": 1010
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 0.7792207792207793,
|
| 1646 |
+
"grad_norm": 100.0627026970796,
|
| 1647 |
+
"learning_rate": 1.2308998302207129e-07,
|
| 1648 |
+
"logits/chosen": -0.3117919862270355,
|
| 1649 |
+
"logits/rejected": -0.27192384004592896,
|
| 1650 |
+
"logps/chosen": -643.2000122070312,
|
| 1651 |
+
"logps/rejected": -626.4000244140625,
|
| 1652 |
+
"loss": 0.4462,
|
| 1653 |
+
"nll_loss": 0.983203113079071,
|
| 1654 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1655 |
+
"rewards/chosen": 0.3858398497104645,
|
| 1656 |
+
"rewards/margins": 2.7281250953674316,
|
| 1657 |
+
"rewards/rejected": -2.3414063453674316,
|
| 1658 |
+
"step": 1020
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 0.7868601986249045,
|
| 1662 |
+
"grad_norm": 99.33843951990231,
|
| 1663 |
+
"learning_rate": 1.1884550084889642e-07,
|
| 1664 |
+
"logits/chosen": -0.20637206733226776,
|
| 1665 |
+
"logits/rejected": -0.17661133408546448,
|
| 1666 |
+
"logps/chosen": -621.0,
|
| 1667 |
+
"logps/rejected": -586.4000244140625,
|
| 1668 |
+
"loss": 0.4063,
|
| 1669 |
+
"nll_loss": 0.924609363079071,
|
| 1670 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1671 |
+
"rewards/chosen": 0.1715087890625,
|
| 1672 |
+
"rewards/margins": 2.515625,
|
| 1673 |
+
"rewards/rejected": -2.344531297683716,
|
| 1674 |
+
"step": 1030
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 0.7944996180290298,
|
| 1678 |
+
"grad_norm": 173.70184429546075,
|
| 1679 |
+
"learning_rate": 1.1460101867572156e-07,
|
| 1680 |
+
"logits/chosen": -0.15251465141773224,
|
| 1681 |
+
"logits/rejected": -0.12940673530101776,
|
| 1682 |
+
"logps/chosen": -624.5999755859375,
|
| 1683 |
+
"logps/rejected": -595.0,
|
| 1684 |
+
"loss": 0.294,
|
| 1685 |
+
"nll_loss": 0.928906261920929,
|
| 1686 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1687 |
+
"rewards/chosen": 0.10312499850988388,
|
| 1688 |
+
"rewards/margins": 2.552734375,
|
| 1689 |
+
"rewards/rejected": -2.452343702316284,
|
| 1690 |
+
"step": 1040
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.8021390374331551,
|
| 1694 |
+
"grad_norm": 112.58127449933384,
|
| 1695 |
+
"learning_rate": 1.1035653650254668e-07,
|
| 1696 |
+
"logits/chosen": -0.202392578125,
|
| 1697 |
+
"logits/rejected": -0.167236328125,
|
| 1698 |
+
"logps/chosen": -592.7999877929688,
|
| 1699 |
+
"logps/rejected": -562.0,
|
| 1700 |
+
"loss": 0.4084,
|
| 1701 |
+
"nll_loss": 0.884765625,
|
| 1702 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1703 |
+
"rewards/chosen": 0.49921876192092896,
|
| 1704 |
+
"rewards/margins": 2.3228516578674316,
|
| 1705 |
+
"rewards/rejected": -1.8244140148162842,
|
| 1706 |
+
"step": 1050
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 0.8097784568372803,
|
| 1710 |
+
"grad_norm": 13.477126429230648,
|
| 1711 |
+
"learning_rate": 1.0611205432937182e-07,
|
| 1712 |
+
"logits/chosen": -0.06467285007238388,
|
| 1713 |
+
"logits/rejected": -0.10791015625,
|
| 1714 |
+
"logps/chosen": -637.5999755859375,
|
| 1715 |
+
"logps/rejected": -594.4000244140625,
|
| 1716 |
+
"loss": 0.4124,
|
| 1717 |
+
"nll_loss": 0.9261718988418579,
|
| 1718 |
+
"rewards/accuracies": 0.6875,
|
| 1719 |
+
"rewards/chosen": 0.5888671875,
|
| 1720 |
+
"rewards/margins": 2.6421875953674316,
|
| 1721 |
+
"rewards/rejected": -2.053906202316284,
|
| 1722 |
+
"step": 1060
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 0.8174178762414056,
|
| 1726 |
+
"grad_norm": 146.54916411561834,
|
| 1727 |
+
"learning_rate": 1.0186757215619694e-07,
|
| 1728 |
+
"logits/chosen": -0.02625732496380806,
|
| 1729 |
+
"logits/rejected": -0.01333007775247097,
|
| 1730 |
+
"logps/chosen": -607.2000122070312,
|
| 1731 |
+
"logps/rejected": -569.0,
|
| 1732 |
+
"loss": 0.3095,
|
| 1733 |
+
"nll_loss": 0.8355468511581421,
|
| 1734 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1735 |
+
"rewards/chosen": 0.664257824420929,
|
| 1736 |
+
"rewards/margins": 2.471874952316284,
|
| 1737 |
+
"rewards/rejected": -1.8097655773162842,
|
| 1738 |
+
"step": 1070
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.825057295645531,
|
| 1742 |
+
"grad_norm": 27.03231972763934,
|
| 1743 |
+
"learning_rate": 9.762308998302207e-08,
|
| 1744 |
+
"logits/chosen": -0.2533203065395355,
|
| 1745 |
+
"logits/rejected": -0.19873046875,
|
| 1746 |
+
"logps/chosen": -530.2000122070312,
|
| 1747 |
+
"logps/rejected": -508.79998779296875,
|
| 1748 |
+
"loss": 0.363,
|
| 1749 |
+
"nll_loss": 0.925000011920929,
|
| 1750 |
+
"rewards/accuracies": 0.8125,
|
| 1751 |
+
"rewards/chosen": 0.32301026582717896,
|
| 1752 |
+
"rewards/margins": 2.681640625,
|
| 1753 |
+
"rewards/rejected": -2.3609375953674316,
|
| 1754 |
+
"step": 1080
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 0.8326967150496563,
|
| 1758 |
+
"grad_norm": 95.08593550485008,
|
| 1759 |
+
"learning_rate": 9.33786078098472e-08,
|
| 1760 |
+
"logits/chosen": -0.113616943359375,
|
| 1761 |
+
"logits/rejected": -0.18590088188648224,
|
| 1762 |
+
"logps/chosen": -522.0,
|
| 1763 |
+
"logps/rejected": -463.0,
|
| 1764 |
+
"loss": 0.3409,
|
| 1765 |
+
"nll_loss": 0.901171863079071,
|
| 1766 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1767 |
+
"rewards/chosen": 0.597607433795929,
|
| 1768 |
+
"rewards/margins": 2.7945313453674316,
|
| 1769 |
+
"rewards/rejected": -2.1976561546325684,
|
| 1770 |
+
"step": 1090
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 0.8403361344537815,
|
| 1774 |
+
"grad_norm": 43.42987904751351,
|
| 1775 |
+
"learning_rate": 8.913412563667231e-08,
|
| 1776 |
+
"logits/chosen": -0.26826173067092896,
|
| 1777 |
+
"logits/rejected": -0.2728515565395355,
|
| 1778 |
+
"logps/chosen": -591.4000244140625,
|
| 1779 |
+
"logps/rejected": -585.5999755859375,
|
| 1780 |
+
"loss": 0.4074,
|
| 1781 |
+
"nll_loss": 0.938281238079071,
|
| 1782 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1783 |
+
"rewards/chosen": 0.4159179627895355,
|
| 1784 |
+
"rewards/margins": 2.444531202316284,
|
| 1785 |
+
"rewards/rejected": -2.027539014816284,
|
| 1786 |
+
"step": 1100
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 0.8479755538579068,
|
| 1790 |
+
"grad_norm": 130.2620800651245,
|
| 1791 |
+
"learning_rate": 8.488964346349745e-08,
|
| 1792 |
+
"logits/chosen": -0.11006774753332138,
|
| 1793 |
+
"logits/rejected": -0.13276366889476776,
|
| 1794 |
+
"logps/chosen": -505.20001220703125,
|
| 1795 |
+
"logps/rejected": -493.0,
|
| 1796 |
+
"loss": 0.3638,
|
| 1797 |
+
"nll_loss": 0.8988281488418579,
|
| 1798 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1799 |
+
"rewards/chosen": 0.3427734375,
|
| 1800 |
+
"rewards/margins": 2.474609375,
|
| 1801 |
+
"rewards/rejected": -2.1343750953674316,
|
| 1802 |
+
"step": 1110
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.8556149732620321,
|
| 1806 |
+
"grad_norm": 258.0177244688908,
|
| 1807 |
+
"learning_rate": 8.064516129032257e-08,
|
| 1808 |
+
"logits/chosen": -0.17624512314796448,
|
| 1809 |
+
"logits/rejected": -0.15255126357078552,
|
| 1810 |
+
"logps/chosen": -585.5999755859375,
|
| 1811 |
+
"logps/rejected": -592.5999755859375,
|
| 1812 |
+
"loss": 0.4054,
|
| 1813 |
+
"nll_loss": 0.9449218511581421,
|
| 1814 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1815 |
+
"rewards/chosen": 0.28657227754592896,
|
| 1816 |
+
"rewards/margins": 2.617968797683716,
|
| 1817 |
+
"rewards/rejected": -2.334765672683716,
|
| 1818 |
+
"step": 1120
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 0.8632543926661573,
|
| 1822 |
+
"grad_norm": 167.06471829769077,
|
| 1823 |
+
"learning_rate": 7.64006791171477e-08,
|
| 1824 |
+
"logits/chosen": -0.01499023474752903,
|
| 1825 |
+
"logits/rejected": -0.0026123046409338713,
|
| 1826 |
+
"logps/chosen": -526.7999877929688,
|
| 1827 |
+
"logps/rejected": -518.7999877929688,
|
| 1828 |
+
"loss": 0.394,
|
| 1829 |
+
"nll_loss": 0.834765613079071,
|
| 1830 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1831 |
+
"rewards/chosen": -0.09824218600988388,
|
| 1832 |
+
"rewards/margins": 2.207812547683716,
|
| 1833 |
+
"rewards/rejected": -2.3070311546325684,
|
| 1834 |
+
"step": 1130
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 0.8708938120702827,
|
| 1838 |
+
"grad_norm": 203.27949384009779,
|
| 1839 |
+
"learning_rate": 7.215619694397283e-08,
|
| 1840 |
+
"logits/chosen": -0.04530029371380806,
|
| 1841 |
+
"logits/rejected": -0.010485840030014515,
|
| 1842 |
+
"logps/chosen": -570.4000244140625,
|
| 1843 |
+
"logps/rejected": -554.0,
|
| 1844 |
+
"loss": 0.3578,
|
| 1845 |
+
"nll_loss": 0.8656250238418579,
|
| 1846 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1847 |
+
"rewards/chosen": 0.8838866949081421,
|
| 1848 |
+
"rewards/margins": 2.2896485328674316,
|
| 1849 |
+
"rewards/rejected": -1.40283203125,
|
| 1850 |
+
"step": 1140
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.878533231474408,
|
| 1854 |
+
"grad_norm": 91.28683779041214,
|
| 1855 |
+
"learning_rate": 6.791171477079796e-08,
|
| 1856 |
+
"logits/chosen": -0.158447265625,
|
| 1857 |
+
"logits/rejected": -0.15849609673023224,
|
| 1858 |
+
"logps/chosen": -659.4000244140625,
|
| 1859 |
+
"logps/rejected": -620.0,
|
| 1860 |
+
"loss": 0.3552,
|
| 1861 |
+
"nll_loss": 1.006250023841858,
|
| 1862 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1863 |
+
"rewards/chosen": 1.120996117591858,
|
| 1864 |
+
"rewards/margins": 2.5785155296325684,
|
| 1865 |
+
"rewards/rejected": -1.453710913658142,
|
| 1866 |
+
"step": 1150
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 0.8861726508785333,
|
| 1870 |
+
"grad_norm": 165.0639361803876,
|
| 1871 |
+
"learning_rate": 6.36672325976231e-08,
|
| 1872 |
+
"logits/chosen": -0.17514649033546448,
|
| 1873 |
+
"logits/rejected": -0.22218628227710724,
|
| 1874 |
+
"logps/chosen": -545.5999755859375,
|
| 1875 |
+
"logps/rejected": -525.5999755859375,
|
| 1876 |
+
"loss": 0.3801,
|
| 1877 |
+
"nll_loss": 0.984375,
|
| 1878 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1879 |
+
"rewards/chosen": 0.7798827886581421,
|
| 1880 |
+
"rewards/margins": 2.7671875953674316,
|
| 1881 |
+
"rewards/rejected": -1.98876953125,
|
| 1882 |
+
"step": 1160
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 0.8938120702826585,
|
| 1886 |
+
"grad_norm": 70.36211868043284,
|
| 1887 |
+
"learning_rate": 5.942275042444821e-08,
|
| 1888 |
+
"logits/chosen": -0.07485046237707138,
|
| 1889 |
+
"logits/rejected": -0.07871093600988388,
|
| 1890 |
+
"logps/chosen": -517.7999877929688,
|
| 1891 |
+
"logps/rejected": -501.20001220703125,
|
| 1892 |
+
"loss": 0.4365,
|
| 1893 |
+
"nll_loss": 0.889453113079071,
|
| 1894 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1895 |
+
"rewards/chosen": 0.19501952826976776,
|
| 1896 |
+
"rewards/margins": 2.075390577316284,
|
| 1897 |
+
"rewards/rejected": -1.877343773841858,
|
| 1898 |
+
"step": 1170
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 0.9014514896867838,
|
| 1902 |
+
"grad_norm": 46.76470844821065,
|
| 1903 |
+
"learning_rate": 5.517826825127334e-08,
|
| 1904 |
+
"logits/chosen": -0.21860352158546448,
|
| 1905 |
+
"logits/rejected": -0.21206054091453552,
|
| 1906 |
+
"logps/chosen": -566.5999755859375,
|
| 1907 |
+
"logps/rejected": -548.0,
|
| 1908 |
+
"loss": 0.452,
|
| 1909 |
+
"nll_loss": 0.876953125,
|
| 1910 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1911 |
+
"rewards/chosen": 0.6045898199081421,
|
| 1912 |
+
"rewards/margins": 1.9511229991912842,
|
| 1913 |
+
"rewards/rejected": -1.346093773841858,
|
| 1914 |
+
"step": 1180
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 0.9090909090909091,
|
| 1918 |
+
"grad_norm": 132.73200075615364,
|
| 1919 |
+
"learning_rate": 5.093378607809847e-08,
|
| 1920 |
+
"logits/chosen": -0.13608399033546448,
|
| 1921 |
+
"logits/rejected": -0.16939696669578552,
|
| 1922 |
+
"logps/chosen": -535.0,
|
| 1923 |
+
"logps/rejected": -506.20001220703125,
|
| 1924 |
+
"loss": 0.3551,
|
| 1925 |
+
"nll_loss": 0.946093738079071,
|
| 1926 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1927 |
+
"rewards/chosen": 0.880932629108429,
|
| 1928 |
+
"rewards/margins": 2.707812547683716,
|
| 1929 |
+
"rewards/rejected": -1.8278319835662842,
|
| 1930 |
+
"step": 1190
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 0.9167303284950343,
|
| 1934 |
+
"grad_norm": 141.55461267715054,
|
| 1935 |
+
"learning_rate": 4.66893039049236e-08,
|
| 1936 |
+
"logits/chosen": -0.20478515326976776,
|
| 1937 |
+
"logits/rejected": -0.16940918564796448,
|
| 1938 |
+
"logps/chosen": -531.7999877929688,
|
| 1939 |
+
"logps/rejected": -499.6000061035156,
|
| 1940 |
+
"loss": 0.3211,
|
| 1941 |
+
"nll_loss": 0.9398437738418579,
|
| 1942 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1943 |
+
"rewards/chosen": 0.8980468511581421,
|
| 1944 |
+
"rewards/margins": 2.780468702316284,
|
| 1945 |
+
"rewards/rejected": -1.885156273841858,
|
| 1946 |
+
"step": 1200
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 0.9243697478991597,
|
| 1950 |
+
"grad_norm": 188.7068399957901,
|
| 1951 |
+
"learning_rate": 4.2444821731748725e-08,
|
| 1952 |
+
"logits/chosen": -0.2447509765625,
|
| 1953 |
+
"logits/rejected": -0.19533690810203552,
|
| 1954 |
+
"logps/chosen": -581.2000122070312,
|
| 1955 |
+
"logps/rejected": -552.2000122070312,
|
| 1956 |
+
"loss": 0.3269,
|
| 1957 |
+
"nll_loss": 0.961718738079071,
|
| 1958 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1959 |
+
"rewards/chosen": 0.3602050840854645,
|
| 1960 |
+
"rewards/margins": 2.5765624046325684,
|
| 1961 |
+
"rewards/rejected": -2.2173829078674316,
|
| 1962 |
+
"step": 1210
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 0.932009167303285,
|
| 1966 |
+
"grad_norm": 104.70717258392624,
|
| 1967 |
+
"learning_rate": 3.820033955857385e-08,
|
| 1968 |
+
"logits/chosen": -0.21748046576976776,
|
| 1969 |
+
"logits/rejected": -0.20539550483226776,
|
| 1970 |
+
"logps/chosen": -736.5999755859375,
|
| 1971 |
+
"logps/rejected": -709.7999877929688,
|
| 1972 |
+
"loss": 0.4171,
|
| 1973 |
+
"nll_loss": 1.0066406726837158,
|
| 1974 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1975 |
+
"rewards/chosen": 0.4925781190395355,
|
| 1976 |
+
"rewards/margins": 2.092578172683716,
|
| 1977 |
+
"rewards/rejected": -1.601171851158142,
|
| 1978 |
+
"step": 1220
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"epoch": 0.9396485867074102,
|
| 1982 |
+
"grad_norm": 120.69266491201337,
|
| 1983 |
+
"learning_rate": 3.395585738539898e-08,
|
| 1984 |
+
"logits/chosen": -0.16705322265625,
|
| 1985 |
+
"logits/rejected": -0.20624999701976776,
|
| 1986 |
+
"logps/chosen": -546.4000244140625,
|
| 1987 |
+
"logps/rejected": -513.5999755859375,
|
| 1988 |
+
"loss": 0.5148,
|
| 1989 |
+
"nll_loss": 0.892578125,
|
| 1990 |
+
"rewards/accuracies": 0.625,
|
| 1991 |
+
"rewards/chosen": 0.23613281548023224,
|
| 1992 |
+
"rewards/margins": 2.397656202316284,
|
| 1993 |
+
"rewards/rejected": -2.1617188453674316,
|
| 1994 |
+
"step": 1230
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 0.9472880061115355,
|
| 1998 |
+
"grad_norm": 89.80554106184103,
|
| 1999 |
+
"learning_rate": 2.9711375212224106e-08,
|
| 2000 |
+
"logits/chosen": -0.22031250596046448,
|
| 2001 |
+
"logits/rejected": -0.26923829317092896,
|
| 2002 |
+
"logps/chosen": -567.2000122070312,
|
| 2003 |
+
"logps/rejected": -513.0,
|
| 2004 |
+
"loss": 0.3652,
|
| 2005 |
+
"nll_loss": 0.949999988079071,
|
| 2006 |
+
"rewards/accuracies": 0.800000011920929,
|
| 2007 |
+
"rewards/chosen": 0.590991199016571,
|
| 2008 |
+
"rewards/margins": 2.520312547683716,
|
| 2009 |
+
"rewards/rejected": -1.9285156726837158,
|
| 2010 |
+
"step": 1240
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 0.9549274255156608,
|
| 2014 |
+
"grad_norm": 68.44803041900859,
|
| 2015 |
+
"learning_rate": 2.5466893039049236e-08,
|
| 2016 |
+
"logits/chosen": -0.10751952975988388,
|
| 2017 |
+
"logits/rejected": -0.10664062201976776,
|
| 2018 |
+
"logps/chosen": -557.4000244140625,
|
| 2019 |
+
"logps/rejected": -524.0,
|
| 2020 |
+
"loss": 0.3961,
|
| 2021 |
+
"nll_loss": 0.864062488079071,
|
| 2022 |
+
"rewards/accuracies": 0.699999988079071,
|
| 2023 |
+
"rewards/chosen": 0.5328124761581421,
|
| 2024 |
+
"rewards/margins": 2.385009765625,
|
| 2025 |
+
"rewards/rejected": -1.853906273841858,
|
| 2026 |
+
"step": 1250
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.9625668449197861,
|
| 2030 |
+
"grad_norm": 77.86404690618204,
|
| 2031 |
+
"learning_rate": 2.1222410865874363e-08,
|
| 2032 |
+
"logits/chosen": -0.23062744736671448,
|
| 2033 |
+
"logits/rejected": -0.19792480766773224,
|
| 2034 |
+
"logps/chosen": -579.0,
|
| 2035 |
+
"logps/rejected": -549.4000244140625,
|
| 2036 |
+
"loss": 0.3363,
|
| 2037 |
+
"nll_loss": 0.908984363079071,
|
| 2038 |
+
"rewards/accuracies": 0.800000011920929,
|
| 2039 |
+
"rewards/chosen": 0.6390625238418579,
|
| 2040 |
+
"rewards/margins": 2.806640625,
|
| 2041 |
+
"rewards/rejected": -2.1689453125,
|
| 2042 |
+
"step": 1260
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 0.9702062643239114,
|
| 2046 |
+
"grad_norm": 92.2508879537955,
|
| 2047 |
+
"learning_rate": 1.697792869269949e-08,
|
| 2048 |
+
"logits/chosen": -0.243896484375,
|
| 2049 |
+
"logits/rejected": -0.2039794921875,
|
| 2050 |
+
"logps/chosen": -596.0,
|
| 2051 |
+
"logps/rejected": -552.0,
|
| 2052 |
+
"loss": 0.442,
|
| 2053 |
+
"nll_loss": 0.9007812738418579,
|
| 2054 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 2055 |
+
"rewards/chosen": 0.3742431700229645,
|
| 2056 |
+
"rewards/margins": 2.436718702316284,
|
| 2057 |
+
"rewards/rejected": -2.05859375,
|
| 2058 |
+
"step": 1270
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 0.9778456837280367,
|
| 2062 |
+
"grad_norm": 99.79262236865266,
|
| 2063 |
+
"learning_rate": 1.2733446519524618e-08,
|
| 2064 |
+
"logits/chosen": -0.23527832329273224,
|
| 2065 |
+
"logits/rejected": -0.218994140625,
|
| 2066 |
+
"logps/chosen": -552.2000122070312,
|
| 2067 |
+
"logps/rejected": -540.0,
|
| 2068 |
+
"loss": 0.4244,
|
| 2069 |
+
"nll_loss": 0.8421875238418579,
|
| 2070 |
+
"rewards/accuracies": 0.699999988079071,
|
| 2071 |
+
"rewards/chosen": 0.4585937559604645,
|
| 2072 |
+
"rewards/margins": 2.216992139816284,
|
| 2073 |
+
"rewards/rejected": -1.75732421875,
|
| 2074 |
+
"step": 1280
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 0.985485103132162,
|
| 2078 |
+
"grad_norm": 140.72925309996154,
|
| 2079 |
+
"learning_rate": 8.488964346349745e-09,
|
| 2080 |
+
"logits/chosen": -0.2117919921875,
|
| 2081 |
+
"logits/rejected": -0.18214111030101776,
|
| 2082 |
+
"logps/chosen": -638.5999755859375,
|
| 2083 |
+
"logps/rejected": -619.7999877929688,
|
| 2084 |
+
"loss": 0.4388,
|
| 2085 |
+
"nll_loss": 0.943359375,
|
| 2086 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 2087 |
+
"rewards/chosen": 0.4850097596645355,
|
| 2088 |
+
"rewards/margins": 2.5679688453674316,
|
| 2089 |
+
"rewards/rejected": -2.085156202316284,
|
| 2090 |
+
"step": 1290
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 0.9931245225362872,
|
| 2094 |
+
"grad_norm": 49.408062951712566,
|
| 2095 |
+
"learning_rate": 4.244482173174872e-09,
|
| 2096 |
+
"logits/chosen": -0.09760741889476776,
|
| 2097 |
+
"logits/rejected": -0.031402587890625,
|
| 2098 |
+
"logps/chosen": -593.2000122070312,
|
| 2099 |
+
"logps/rejected": -587.0,
|
| 2100 |
+
"loss": 0.4112,
|
| 2101 |
+
"nll_loss": 0.869140625,
|
| 2102 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 2103 |
+
"rewards/chosen": 0.6004394292831421,
|
| 2104 |
+
"rewards/margins": 2.287890672683716,
|
| 2105 |
+
"rewards/rejected": -1.6857421398162842,
|
| 2106 |
+
"step": 1300
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 1.0,
|
| 2110 |
+
"eval_logits/chosen": -0.10636549443006516,
|
| 2111 |
+
"eval_logits/rejected": -0.1420053392648697,
|
| 2112 |
+
"eval_logps/chosen": -509.3846130371094,
|
| 2113 |
+
"eval_logps/rejected": -480.6153869628906,
|
| 2114 |
+
"eval_loss": 0.3973870873451233,
|
| 2115 |
+
"eval_nll_loss": 0.9269831776618958,
|
| 2116 |
+
"eval_rewards/accuracies": 0.7211538553237915,
|
| 2117 |
+
"eval_rewards/chosen": 0.2996356785297394,
|
| 2118 |
+
"eval_rewards/margins": 2.412559986114502,
|
| 2119 |
+
"eval_rewards/rejected": -2.112868070602417,
|
| 2120 |
+
"eval_runtime": 8.7861,
|
| 2121 |
+
"eval_samples_per_second": 11.382,
|
| 2122 |
+
"eval_steps_per_second": 1.48,
|
| 2123 |
+
"step": 1309
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 1.0,
|
| 2127 |
+
"step": 1309,
|
| 2128 |
+
"total_flos": 0.0,
|
| 2129 |
+
"train_loss": 0.4408511233930828,
|
| 2130 |
+
"train_runtime": 2418.9161,
|
| 2131 |
+
"train_samples_per_second": 4.327,
|
| 2132 |
+
"train_steps_per_second": 0.541
|
| 2133 |
+
}
|
| 2134 |
+
],
|
| 2135 |
+
"logging_steps": 10,
|
| 2136 |
+
"max_steps": 1309,
|
| 2137 |
+
"num_input_tokens_seen": 0,
|
| 2138 |
+
"num_train_epochs": 1,
|
| 2139 |
+
"save_steps": 500,
|
| 2140 |
+
"stateful_callbacks": {
|
| 2141 |
+
"TrainerControl": {
|
| 2142 |
+
"args": {
|
| 2143 |
+
"should_epoch_stop": false,
|
| 2144 |
+
"should_evaluate": false,
|
| 2145 |
+
"should_log": false,
|
| 2146 |
+
"should_save": true,
|
| 2147 |
+
"should_training_stop": true
|
| 2148 |
+
},
|
| 2149 |
+
"attributes": {}
|
| 2150 |
+
}
|
| 2151 |
+
},
|
| 2152 |
+
"total_flos": 0.0,
|
| 2153 |
+
"train_batch_size": 2,
|
| 2154 |
+
"trial_name": null,
|
| 2155 |
+
"trial_params": null
|
| 2156 |
+
}
|