full-state checkpoint 10-percent (step 75)
Browse files
ckpt-10-percent/adapter_config.json
CHANGED
|
@@ -33,13 +33,13 @@
|
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
| 36 |
-
"v_proj",
|
| 37 |
"up_proj",
|
| 38 |
-
"q_proj",
|
| 39 |
-
"k_proj",
|
| 40 |
"o_proj",
|
| 41 |
"gate_proj",
|
| 42 |
-
"down_proj"
|
|
|
|
|
|
|
|
|
|
| 43 |
],
|
| 44 |
"target_parameters": null,
|
| 45 |
"task_type": "CAUSAL_LM",
|
|
|
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
|
|
|
| 36 |
"up_proj",
|
|
|
|
|
|
|
| 37 |
"o_proj",
|
| 38 |
"gate_proj",
|
| 39 |
+
"down_proj",
|
| 40 |
+
"v_proj",
|
| 41 |
+
"k_proj",
|
| 42 |
+
"q_proj"
|
| 43 |
],
|
| 44 |
"target_parameters": null,
|
| 45 |
"task_type": "CAUSAL_LM",
|
ckpt-10-percent/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 528550256
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0f59456b130c136a1e74d092ac27b7e24d7bd6a28968f8e4d30ecf1dd2f2a6af
|
| 3 |
size 528550256
|
ckpt-10-percent/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:70f986b0e013e0c8a3a70c9c0465167f804fd40e54dc5bcfca3169f0237c29a6
|
| 3 |
+
size 268963141
|
ckpt-10-percent/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3384fda721d87cf13c70812a8bea527841773ed994afaaf23bda164b902877b
|
| 3 |
+
size 14709
|
ckpt-10-percent/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:620e365c24703d8eb8dcba3ff0a9ceabbb8471df3a45aaf4beb33a3f78143275
|
| 3 |
+
size 1465
|
ckpt-10-percent/trainer_state.json
ADDED
|
@@ -0,0 +1,1662 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.2,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 75,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"completion_length": 390.8525085449219,
|
| 14 |
+
"completions/clipped_ratio": 0.022499999031424522,
|
| 15 |
+
"completions/max_length": 2048.0,
|
| 16 |
+
"completions/max_terminated_length": 1494.0,
|
| 17 |
+
"completions/mean_length": 391.8299865722656,
|
| 18 |
+
"completions/mean_terminated_length": 353.70843505859375,
|
| 19 |
+
"completions/min_length": 36.0,
|
| 20 |
+
"completions/min_terminated_length": 36.0,
|
| 21 |
+
"epoch": 0.0026666666666666666,
|
| 22 |
+
"frac_reward_zero_std": 0.0,
|
| 23 |
+
"grad_norm": 0.23882409930229187,
|
| 24 |
+
"kl": 0.0,
|
| 25 |
+
"learning_rate": 1e-05,
|
| 26 |
+
"loss": -0.0,
|
| 27 |
+
"num_tokens": 274622.0,
|
| 28 |
+
"reward": -0.3036116659641266,
|
| 29 |
+
"reward_std": 1.1926226615905762,
|
| 30 |
+
"rewards/multidomain_reward_func/mean": -0.3036116659641266,
|
| 31 |
+
"rewards/multidomain_reward_func/std": 1.5280684232711792,
|
| 32 |
+
"step": 1
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"completion_length": 429.7850082397461,
|
| 36 |
+
"completions/clipped_ratio": 0.042500000447034836,
|
| 37 |
+
"completions/max_length": 2048.0,
|
| 38 |
+
"completions/max_terminated_length": 1361.0,
|
| 39 |
+
"completions/mean_length": 430.74249267578125,
|
| 40 |
+
"completions/mean_terminated_length": 358.9582214355469,
|
| 41 |
+
"completions/min_length": 18.0,
|
| 42 |
+
"completions/min_terminated_length": 18.0,
|
| 43 |
+
"epoch": 0.005333333333333333,
|
| 44 |
+
"frac_reward_zero_std": 0.0,
|
| 45 |
+
"grad_norm": 0.2470335215330124,
|
| 46 |
+
"kl": 0.0,
|
| 47 |
+
"learning_rate": 9.986666666666667e-06,
|
| 48 |
+
"loss": 0.0,
|
| 49 |
+
"num_tokens": 571689.0,
|
| 50 |
+
"reward": -0.1849958449602127,
|
| 51 |
+
"reward_std": 1.1945927143096924,
|
| 52 |
+
"rewards/multidomain_reward_func/mean": -0.1849958449602127,
|
| 53 |
+
"rewards/multidomain_reward_func/std": 1.7641037702560425,
|
| 54 |
+
"step": 2
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"completion_length": 469.2000045776367,
|
| 58 |
+
"completions/clipped_ratio": 0.0625,
|
| 59 |
+
"completions/max_length": 2048.0,
|
| 60 |
+
"completions/max_terminated_length": 1267.0,
|
| 61 |
+
"completions/mean_length": 470.1374816894531,
|
| 62 |
+
"completions/mean_terminated_length": 364.9466552734375,
|
| 63 |
+
"completions/min_length": 79.0,
|
| 64 |
+
"completions/min_terminated_length": 79.0,
|
| 65 |
+
"epoch": 0.008,
|
| 66 |
+
"frac_reward_zero_std": 0.09999999403953552,
|
| 67 |
+
"grad_norm": 0.2603771984577179,
|
| 68 |
+
"kl": 0.0,
|
| 69 |
+
"learning_rate": 9.973333333333333e-06,
|
| 70 |
+
"loss": -0.0,
|
| 71 |
+
"num_tokens": 898834.0,
|
| 72 |
+
"reward": -0.8332051634788513,
|
| 73 |
+
"reward_std": 0.9539631605148315,
|
| 74 |
+
"rewards/multidomain_reward_func/mean": -0.8332051038742065,
|
| 75 |
+
"rewards/multidomain_reward_func/std": 2.1196253299713135,
|
| 76 |
+
"step": 3
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"completion_length": 391.4550064086914,
|
| 80 |
+
"completions/clipped_ratio": 0.004999999888241291,
|
| 81 |
+
"completions/max_length": 2048.0,
|
| 82 |
+
"completions/max_terminated_length": 1078.0,
|
| 83 |
+
"completions/mean_length": 392.4499816894531,
|
| 84 |
+
"completions/mean_terminated_length": 384.1306457519531,
|
| 85 |
+
"completions/min_length": 63.0,
|
| 86 |
+
"completions/min_terminated_length": 63.0,
|
| 87 |
+
"epoch": 0.010666666666666666,
|
| 88 |
+
"frac_reward_zero_std": 0.0,
|
| 89 |
+
"grad_norm": 0.30947721004486084,
|
| 90 |
+
"kl": 0.0,
|
| 91 |
+
"learning_rate": 9.960000000000001e-06,
|
| 92 |
+
"loss": -0.0,
|
| 93 |
+
"num_tokens": 1170514.0,
|
| 94 |
+
"reward": 0.25277450680732727,
|
| 95 |
+
"reward_std": 0.9387752413749695,
|
| 96 |
+
"rewards/multidomain_reward_func/mean": 0.25277450680732727,
|
| 97 |
+
"rewards/multidomain_reward_func/std": 1.2881438732147217,
|
| 98 |
+
"step": 4
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"completion_length": 414.002507019043,
|
| 102 |
+
"completions/clipped_ratio": 0.01249999925494194,
|
| 103 |
+
"completions/max_length": 2048.0,
|
| 104 |
+
"completions/max_terminated_length": 1733.0,
|
| 105 |
+
"completions/mean_length": 414.989990234375,
|
| 106 |
+
"completions/mean_terminated_length": 394.3190002441406,
|
| 107 |
+
"completions/min_length": 11.0,
|
| 108 |
+
"completions/min_terminated_length": 11.0,
|
| 109 |
+
"epoch": 0.013333333333333334,
|
| 110 |
+
"frac_reward_zero_std": 0.0,
|
| 111 |
+
"grad_norm": 0.23343618214130402,
|
| 112 |
+
"kl": 0.0,
|
| 113 |
+
"learning_rate": 9.946666666666667e-06,
|
| 114 |
+
"loss": -0.0,
|
| 115 |
+
"num_tokens": 1454280.0,
|
| 116 |
+
"reward": 0.15532104671001434,
|
| 117 |
+
"reward_std": 0.8334585428237915,
|
| 118 |
+
"rewards/multidomain_reward_func/mean": 0.15532104671001434,
|
| 119 |
+
"rewards/multidomain_reward_func/std": 1.405861258506775,
|
| 120 |
+
"step": 5
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"completion_length": 417.9025054931641,
|
| 124 |
+
"completions/clipped_ratio": 0.014999999664723873,
|
| 125 |
+
"completions/max_length": 2048.0,
|
| 126 |
+
"completions/max_terminated_length": 1255.0,
|
| 127 |
+
"completions/mean_length": 418.8874816894531,
|
| 128 |
+
"completions/mean_terminated_length": 394.07867431640625,
|
| 129 |
+
"completions/min_length": 36.0,
|
| 130 |
+
"completions/min_terminated_length": 36.0,
|
| 131 |
+
"epoch": 0.016,
|
| 132 |
+
"frac_reward_zero_std": 0.0,
|
| 133 |
+
"grad_norm": 0.15190428495407104,
|
| 134 |
+
"kl": 0.0,
|
| 135 |
+
"learning_rate": 9.933333333333334e-06,
|
| 136 |
+
"loss": -0.0,
|
| 137 |
+
"num_tokens": 1732945.0,
|
| 138 |
+
"reward": 0.1933300942182541,
|
| 139 |
+
"reward_std": 1.0449475049972534,
|
| 140 |
+
"rewards/multidomain_reward_func/mean": 0.19333010911941528,
|
| 141 |
+
"rewards/multidomain_reward_func/std": 1.4342669248580933,
|
| 142 |
+
"step": 6
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"completion_length": 421.90250396728516,
|
| 146 |
+
"completions/clipped_ratio": 0.007499999832361937,
|
| 147 |
+
"completions/max_length": 2048.0,
|
| 148 |
+
"completions/max_terminated_length": 1837.0,
|
| 149 |
+
"completions/mean_length": 422.8949890136719,
|
| 150 |
+
"completions/mean_terminated_length": 410.6145935058594,
|
| 151 |
+
"completions/min_length": 18.0,
|
| 152 |
+
"completions/min_terminated_length": 18.0,
|
| 153 |
+
"epoch": 0.018666666666666668,
|
| 154 |
+
"frac_reward_zero_std": 0.0,
|
| 155 |
+
"grad_norm": 0.1601637899875641,
|
| 156 |
+
"kl": 0.0,
|
| 157 |
+
"learning_rate": 9.920000000000002e-06,
|
| 158 |
+
"loss": 0.0,
|
| 159 |
+
"num_tokens": 2018863.0,
|
| 160 |
+
"reward": 0.14378485083580017,
|
| 161 |
+
"reward_std": 0.9768227934837341,
|
| 162 |
+
"rewards/multidomain_reward_func/mean": 0.14378486573696136,
|
| 163 |
+
"rewards/multidomain_reward_func/std": 1.4048492908477783,
|
| 164 |
+
"step": 7
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"completion_length": 391.8425033569336,
|
| 168 |
+
"completions/clipped_ratio": 0.004999999888241291,
|
| 169 |
+
"completions/max_length": 2048.0,
|
| 170 |
+
"completions/max_terminated_length": 1118.0,
|
| 171 |
+
"completions/mean_length": 392.8374938964844,
|
| 172 |
+
"completions/mean_terminated_length": 384.5201110839844,
|
| 173 |
+
"completions/min_length": 33.0,
|
| 174 |
+
"completions/min_terminated_length": 33.0,
|
| 175 |
+
"epoch": 0.021333333333333333,
|
| 176 |
+
"frac_reward_zero_std": 0.0,
|
| 177 |
+
"grad_norm": 0.31883543729782104,
|
| 178 |
+
"kl": 0.0,
|
| 179 |
+
"learning_rate": 9.906666666666668e-06,
|
| 180 |
+
"loss": 0.0,
|
| 181 |
+
"num_tokens": 2290948.0,
|
| 182 |
+
"reward": 0.26540714502334595,
|
| 183 |
+
"reward_std": 0.9163522720336914,
|
| 184 |
+
"rewards/multidomain_reward_func/mean": 0.26540714502334595,
|
| 185 |
+
"rewards/multidomain_reward_func/std": 1.2766904830932617,
|
| 186 |
+
"step": 8
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"completion_length": 426.43250579833983,
|
| 190 |
+
"completions/clipped_ratio": 0.007499999832361937,
|
| 191 |
+
"completions/max_length": 2048.0,
|
| 192 |
+
"completions/max_terminated_length": 1471.0,
|
| 193 |
+
"completions/mean_length": 427.42498779296875,
|
| 194 |
+
"completions/mean_terminated_length": 415.1788330078125,
|
| 195 |
+
"completions/min_length": 36.0,
|
| 196 |
+
"completions/min_terminated_length": 36.0,
|
| 197 |
+
"epoch": 0.024,
|
| 198 |
+
"frac_reward_zero_std": 0.0,
|
| 199 |
+
"grad_norm": 0.1413382887840271,
|
| 200 |
+
"kl": 0.0,
|
| 201 |
+
"learning_rate": 9.893333333333334e-06,
|
| 202 |
+
"loss": 0.0,
|
| 203 |
+
"num_tokens": 2583858.0,
|
| 204 |
+
"reward": 0.23617644608020782,
|
| 205 |
+
"reward_std": 0.9365644454956055,
|
| 206 |
+
"rewards/multidomain_reward_func/mean": 0.23617644608020782,
|
| 207 |
+
"rewards/multidomain_reward_func/std": 1.3539695739746094,
|
| 208 |
+
"step": 9
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"completion_length": 467.14500427246094,
|
| 212 |
+
"completions/clipped_ratio": 0.009999999776482582,
|
| 213 |
+
"completions/max_length": 2048.0,
|
| 214 |
+
"completions/max_terminated_length": 1490.0,
|
| 215 |
+
"completions/mean_length": 468.1349792480469,
|
| 216 |
+
"completions/mean_terminated_length": 452.1767578125,
|
| 217 |
+
"completions/min_length": 97.0,
|
| 218 |
+
"completions/min_terminated_length": 97.0,
|
| 219 |
+
"epoch": 0.02666666666666667,
|
| 220 |
+
"frac_reward_zero_std": 0.0,
|
| 221 |
+
"grad_norm": 0.17736880481243134,
|
| 222 |
+
"kl": 0.0,
|
| 223 |
+
"learning_rate": 9.88e-06,
|
| 224 |
+
"loss": -0.0,
|
| 225 |
+
"num_tokens": 2908252.0,
|
| 226 |
+
"reward": 0.10222619771957397,
|
| 227 |
+
"reward_std": 0.9887304306030273,
|
| 228 |
+
"rewards/multidomain_reward_func/mean": 0.10222619771957397,
|
| 229 |
+
"rewards/multidomain_reward_func/std": 1.4667627811431885,
|
| 230 |
+
"step": 10
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"completion_length": 421.3575073242188,
|
| 234 |
+
"completions/clipped_ratio": 0.007499999832361937,
|
| 235 |
+
"completions/max_length": 2048.0,
|
| 236 |
+
"completions/max_terminated_length": 1658.0,
|
| 237 |
+
"completions/mean_length": 422.3499755859375,
|
| 238 |
+
"completions/mean_terminated_length": 410.0654602050781,
|
| 239 |
+
"completions/min_length": 97.0,
|
| 240 |
+
"completions/min_terminated_length": 97.0,
|
| 241 |
+
"epoch": 0.029333333333333333,
|
| 242 |
+
"frac_reward_zero_std": 0.0,
|
| 243 |
+
"grad_norm": 0.1340264528989792,
|
| 244 |
+
"kl": 0.0,
|
| 245 |
+
"learning_rate": 9.866666666666668e-06,
|
| 246 |
+
"loss": 0.0,
|
| 247 |
+
"num_tokens": 3198092.0,
|
| 248 |
+
"reward": 0.30930107831954956,
|
| 249 |
+
"reward_std": 0.7165582180023193,
|
| 250 |
+
"rewards/multidomain_reward_func/mean": 0.3093010187149048,
|
| 251 |
+
"rewards/multidomain_reward_func/std": 1.2975008487701416,
|
| 252 |
+
"step": 11
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"completion_length": 418.21250610351564,
|
| 256 |
+
"completions/clipped_ratio": 0.0,
|
| 257 |
+
"completions/max_length": 1392.0,
|
| 258 |
+
"completions/max_terminated_length": 1392.0,
|
| 259 |
+
"completions/mean_length": 419.2124938964844,
|
| 260 |
+
"completions/mean_terminated_length": 419.2124938964844,
|
| 261 |
+
"completions/min_length": 115.0,
|
| 262 |
+
"completions/min_terminated_length": 115.0,
|
| 263 |
+
"epoch": 0.032,
|
| 264 |
+
"frac_reward_zero_std": 0.0,
|
| 265 |
+
"grad_norm": 0.16614454984664917,
|
| 266 |
+
"kl": 0.0,
|
| 267 |
+
"learning_rate": 9.853333333333334e-06,
|
| 268 |
+
"loss": 0.0,
|
| 269 |
+
"num_tokens": 3487357.0,
|
| 270 |
+
"reward": 0.550111711025238,
|
| 271 |
+
"reward_std": 0.7317887544631958,
|
| 272 |
+
"rewards/multidomain_reward_func/mean": 0.550111711025238,
|
| 273 |
+
"rewards/multidomain_reward_func/std": 1.0946627855300903,
|
| 274 |
+
"step": 12
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"completion_length": 419.3975082397461,
|
| 278 |
+
"completions/clipped_ratio": 0.0,
|
| 279 |
+
"completions/max_length": 1651.0,
|
| 280 |
+
"completions/max_terminated_length": 1651.0,
|
| 281 |
+
"completions/mean_length": 420.3974914550781,
|
| 282 |
+
"completions/mean_terminated_length": 420.3974914550781,
|
| 283 |
+
"completions/min_length": 73.0,
|
| 284 |
+
"completions/min_terminated_length": 73.0,
|
| 285 |
+
"epoch": 0.034666666666666665,
|
| 286 |
+
"frac_reward_zero_std": 0.0,
|
| 287 |
+
"grad_norm": 0.1116686686873436,
|
| 288 |
+
"kl": 0.0,
|
| 289 |
+
"learning_rate": 9.84e-06,
|
| 290 |
+
"loss": 0.0,
|
| 291 |
+
"num_tokens": 3783606.0,
|
| 292 |
+
"reward": 0.5550642609596252,
|
| 293 |
+
"reward_std": 0.6016712188720703,
|
| 294 |
+
"rewards/multidomain_reward_func/mean": 0.5550642609596252,
|
| 295 |
+
"rewards/multidomain_reward_func/std": 1.0916095972061157,
|
| 296 |
+
"step": 13
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"completion_length": 426.627507019043,
|
| 300 |
+
"completions/clipped_ratio": 0.007499999832361937,
|
| 301 |
+
"completions/max_length": 2048.0,
|
| 302 |
+
"completions/max_terminated_length": 1428.0,
|
| 303 |
+
"completions/mean_length": 427.6199951171875,
|
| 304 |
+
"completions/mean_terminated_length": 415.37530517578125,
|
| 305 |
+
"completions/min_length": 104.0,
|
| 306 |
+
"completions/min_terminated_length": 104.0,
|
| 307 |
+
"epoch": 0.037333333333333336,
|
| 308 |
+
"frac_reward_zero_std": 0.0,
|
| 309 |
+
"grad_norm": 0.13436946272850037,
|
| 310 |
+
"kl": 0.0,
|
| 311 |
+
"learning_rate": 9.826666666666667e-06,
|
| 312 |
+
"loss": 0.0,
|
| 313 |
+
"num_tokens": 4083764.0,
|
| 314 |
+
"reward": 0.466512531042099,
|
| 315 |
+
"reward_std": 0.7918379306793213,
|
| 316 |
+
"rewards/multidomain_reward_func/mean": 0.466512531042099,
|
| 317 |
+
"rewards/multidomain_reward_func/std": 1.2242188453674316,
|
| 318 |
+
"step": 14
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"completion_length": 378.4425079345703,
|
| 322 |
+
"completions/clipped_ratio": 0.0024999999441206455,
|
| 323 |
+
"completions/max_length": 2048.0,
|
| 324 |
+
"completions/max_terminated_length": 938.0,
|
| 325 |
+
"completions/mean_length": 379.44000244140625,
|
| 326 |
+
"completions/mean_terminated_length": 375.2581481933594,
|
| 327 |
+
"completions/min_length": 73.0,
|
| 328 |
+
"completions/min_terminated_length": 73.0,
|
| 329 |
+
"epoch": 0.04,
|
| 330 |
+
"frac_reward_zero_std": 0.0,
|
| 331 |
+
"grad_norm": 0.12867581844329834,
|
| 332 |
+
"kl": 0.0,
|
| 333 |
+
"learning_rate": 9.813333333333333e-06,
|
| 334 |
+
"loss": 0.0,
|
| 335 |
+
"num_tokens": 4345610.0,
|
| 336 |
+
"reward": 0.5736742615699768,
|
| 337 |
+
"reward_std": 0.6882736086845398,
|
| 338 |
+
"rewards/multidomain_reward_func/mean": 0.5736742615699768,
|
| 339 |
+
"rewards/multidomain_reward_func/std": 1.0166397094726562,
|
| 340 |
+
"step": 15
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"completion_length": 409.86500854492186,
|
| 344 |
+
"completions/clipped_ratio": 0.0,
|
| 345 |
+
"completions/max_length": 1269.0,
|
| 346 |
+
"completions/max_terminated_length": 1269.0,
|
| 347 |
+
"completions/mean_length": 410.864990234375,
|
| 348 |
+
"completions/mean_terminated_length": 410.864990234375,
|
| 349 |
+
"completions/min_length": 98.0,
|
| 350 |
+
"completions/min_terminated_length": 98.0,
|
| 351 |
+
"epoch": 0.042666666666666665,
|
| 352 |
+
"frac_reward_zero_std": 0.0,
|
| 353 |
+
"grad_norm": 0.1019008606672287,
|
| 354 |
+
"kl": 0.0,
|
| 355 |
+
"learning_rate": 9.800000000000001e-06,
|
| 356 |
+
"loss": -0.0,
|
| 357 |
+
"num_tokens": 4628316.0,
|
| 358 |
+
"reward": 0.7713325619697571,
|
| 359 |
+
"reward_std": 0.6617187857627869,
|
| 360 |
+
"rewards/multidomain_reward_func/mean": 0.7713325023651123,
|
| 361 |
+
"rewards/multidomain_reward_func/std": 1.1467927694320679,
|
| 362 |
+
"step": 16
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"completion_length": 425.4700073242187,
|
| 366 |
+
"completions/clipped_ratio": 0.0024999999441206455,
|
| 367 |
+
"completions/max_length": 2048.0,
|
| 368 |
+
"completions/max_terminated_length": 1551.0,
|
| 369 |
+
"completions/mean_length": 426.4674987792969,
|
| 370 |
+
"completions/mean_terminated_length": 422.40350341796875,
|
| 371 |
+
"completions/min_length": 93.0,
|
| 372 |
+
"completions/min_terminated_length": 93.0,
|
| 373 |
+
"epoch": 0.04533333333333334,
|
| 374 |
+
"frac_reward_zero_std": 0.0,
|
| 375 |
+
"grad_norm": 0.106789730489254,
|
| 376 |
+
"kl": 0.0,
|
| 377 |
+
"learning_rate": 9.786666666666667e-06,
|
| 378 |
+
"loss": 0.0,
|
| 379 |
+
"num_tokens": 4928883.0,
|
| 380 |
+
"reward": 0.5637399554252625,
|
| 381 |
+
"reward_std": 0.7021534442901611,
|
| 382 |
+
"rewards/multidomain_reward_func/mean": 0.5637399554252625,
|
| 383 |
+
"rewards/multidomain_reward_func/std": 1.0422096252441406,
|
| 384 |
+
"step": 17
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"completion_length": 431.8600051879883,
|
| 388 |
+
"completions/clipped_ratio": 0.0024999999441206455,
|
| 389 |
+
"completions/max_length": 2048.0,
|
| 390 |
+
"completions/max_terminated_length": 1521.0,
|
| 391 |
+
"completions/mean_length": 432.85748291015625,
|
| 392 |
+
"completions/mean_terminated_length": 428.80950927734375,
|
| 393 |
+
"completions/min_length": 105.0,
|
| 394 |
+
"completions/min_terminated_length": 105.0,
|
| 395 |
+
"epoch": 0.048,
|
| 396 |
+
"frac_reward_zero_std": 0.0,
|
| 397 |
+
"grad_norm": 0.12454589456319809,
|
| 398 |
+
"kl": 0.0,
|
| 399 |
+
"learning_rate": 9.773333333333335e-06,
|
| 400 |
+
"loss": 0.0,
|
| 401 |
+
"num_tokens": 5226216.0,
|
| 402 |
+
"reward": 0.5953689217567444,
|
| 403 |
+
"reward_std": 0.5794697403907776,
|
| 404 |
+
"rewards/multidomain_reward_func/mean": 0.5953689217567444,
|
| 405 |
+
"rewards/multidomain_reward_func/std": 1.0605465173721313,
|
| 406 |
+
"step": 18
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"completion_length": 412.7725082397461,
|
| 410 |
+
"completions/clipped_ratio": 0.004999999888241291,
|
| 411 |
+
"completions/max_length": 2048.0,
|
| 412 |
+
"completions/max_terminated_length": 1357.0,
|
| 413 |
+
"completions/mean_length": 413.7674865722656,
|
| 414 |
+
"completions/mean_terminated_length": 405.5552673339844,
|
| 415 |
+
"completions/min_length": 92.0,
|
| 416 |
+
"completions/min_terminated_length": 92.0,
|
| 417 |
+
"epoch": 0.050666666666666665,
|
| 418 |
+
"frac_reward_zero_std": 0.0,
|
| 419 |
+
"grad_norm": 0.13599112629890442,
|
| 420 |
+
"kl": 0.0,
|
| 421 |
+
"learning_rate": 9.760000000000001e-06,
|
| 422 |
+
"loss": -0.0,
|
| 423 |
+
"num_tokens": 5512113.0,
|
| 424 |
+
"reward": 0.39229491353034973,
|
| 425 |
+
"reward_std": 0.8084317445755005,
|
| 426 |
+
"rewards/multidomain_reward_func/mean": 0.39229488372802734,
|
| 427 |
+
"rewards/multidomain_reward_func/std": 1.3224574327468872,
|
| 428 |
+
"step": 19
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"completion_length": 387.01250610351565,
|
| 432 |
+
"completions/clipped_ratio": 0.0,
|
| 433 |
+
"completions/max_length": 1417.0,
|
| 434 |
+
"completions/max_terminated_length": 1417.0,
|
| 435 |
+
"completions/mean_length": 388.0124816894531,
|
| 436 |
+
"completions/mean_terminated_length": 388.0124816894531,
|
| 437 |
+
"completions/min_length": 109.0,
|
| 438 |
+
"completions/min_terminated_length": 109.0,
|
| 439 |
+
"epoch": 0.05333333333333334,
|
| 440 |
+
"frac_reward_zero_std": 0.0,
|
| 441 |
+
"grad_norm": 0.12738779187202454,
|
| 442 |
+
"kl": 0.0,
|
| 443 |
+
"learning_rate": 9.746666666666668e-06,
|
| 444 |
+
"loss": 0.0,
|
| 445 |
+
"num_tokens": 5778268.0,
|
| 446 |
+
"reward": 0.5316053628921509,
|
| 447 |
+
"reward_std": 0.7337101697921753,
|
| 448 |
+
"rewards/multidomain_reward_func/mean": 0.5316053628921509,
|
| 449 |
+
"rewards/multidomain_reward_func/std": 1.270464539527893,
|
| 450 |
+
"step": 20
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"completion_length": 413.6300018310547,
|
| 454 |
+
"completions/clipped_ratio": 0.0,
|
| 455 |
+
"completions/max_length": 1348.0,
|
| 456 |
+
"completions/max_terminated_length": 1348.0,
|
| 457 |
+
"completions/mean_length": 414.6300048828125,
|
| 458 |
+
"completions/mean_terminated_length": 414.6300048828125,
|
| 459 |
+
"completions/min_length": 108.0,
|
| 460 |
+
"completions/min_terminated_length": 108.0,
|
| 461 |
+
"epoch": 0.056,
|
| 462 |
+
"frac_reward_zero_std": 0.0,
|
| 463 |
+
"grad_norm": 0.11424646526575089,
|
| 464 |
+
"kl": 0.0,
|
| 465 |
+
"learning_rate": 9.733333333333334e-06,
|
| 466 |
+
"loss": 0.0,
|
| 467 |
+
"num_tokens": 6069330.0,
|
| 468 |
+
"reward": 0.5331506133079529,
|
| 469 |
+
"reward_std": 0.7416218519210815,
|
| 470 |
+
"rewards/multidomain_reward_func/mean": 0.5331505537033081,
|
| 471 |
+
"rewards/multidomain_reward_func/std": 1.2239094972610474,
|
| 472 |
+
"step": 21
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"completion_length": 369.3925064086914,
|
| 476 |
+
"completions/clipped_ratio": 0.0,
|
| 477 |
+
"completions/max_length": 1257.0,
|
| 478 |
+
"completions/max_terminated_length": 1257.0,
|
| 479 |
+
"completions/mean_length": 370.3924865722656,
|
| 480 |
+
"completions/mean_terminated_length": 370.3924865722656,
|
| 481 |
+
"completions/min_length": 93.0,
|
| 482 |
+
"completions/min_terminated_length": 93.0,
|
| 483 |
+
"epoch": 0.058666666666666666,
|
| 484 |
+
"frac_reward_zero_std": 0.0,
|
| 485 |
+
"grad_norm": 0.12145662307739258,
|
| 486 |
+
"kl": 0.0,
|
| 487 |
+
"learning_rate": 9.72e-06,
|
| 488 |
+
"loss": 0.0,
|
| 489 |
+
"num_tokens": 6327207.0,
|
| 490 |
+
"reward": 0.6458982229232788,
|
| 491 |
+
"reward_std": 0.6354836821556091,
|
| 492 |
+
"rewards/multidomain_reward_func/mean": 0.6458981037139893,
|
| 493 |
+
"rewards/multidomain_reward_func/std": 0.9722529053688049,
|
| 494 |
+
"step": 22
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"completion_length": 427.4825103759766,
|
| 498 |
+
"completions/clipped_ratio": 0.0024999999441206455,
|
| 499 |
+
"completions/max_length": 2048.0,
|
| 500 |
+
"completions/max_terminated_length": 1758.0,
|
| 501 |
+
"completions/mean_length": 428.47998046875,
|
| 502 |
+
"completions/mean_terminated_length": 424.4210510253906,
|
| 503 |
+
"completions/min_length": 114.0,
|
| 504 |
+
"completions/min_terminated_length": 114.0,
|
| 505 |
+
"epoch": 0.06133333333333333,
|
| 506 |
+
"frac_reward_zero_std": 0.0,
|
| 507 |
+
"grad_norm": 0.11638902127742767,
|
| 508 |
+
"kl": 0.0,
|
| 509 |
+
"learning_rate": 9.706666666666668e-06,
|
| 510 |
+
"loss": 0.0,
|
| 511 |
+
"num_tokens": 6623079.0,
|
| 512 |
+
"reward": 0.678668737411499,
|
| 513 |
+
"reward_std": 0.6608802676200867,
|
| 514 |
+
"rewards/multidomain_reward_func/mean": 0.678668737411499,
|
| 515 |
+
"rewards/multidomain_reward_func/std": 1.0120640993118286,
|
| 516 |
+
"step": 23
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"completion_length": 450.40000915527344,
|
| 520 |
+
"completions/clipped_ratio": 0.017500000074505806,
|
| 521 |
+
"completions/max_length": 2048.0,
|
| 522 |
+
"completions/max_terminated_length": 1979.0,
|
| 523 |
+
"completions/mean_length": 451.3824768066406,
|
| 524 |
+
"completions/mean_terminated_length": 422.94403076171875,
|
| 525 |
+
"completions/min_length": 53.0,
|
| 526 |
+
"completions/min_terminated_length": 53.0,
|
| 527 |
+
"epoch": 0.064,
|
| 528 |
+
"frac_reward_zero_std": 0.09999999403953552,
|
| 529 |
+
"grad_norm": 0.11742237955331802,
|
| 530 |
+
"kl": 0.0,
|
| 531 |
+
"learning_rate": 9.693333333333334e-06,
|
| 532 |
+
"loss": 0.0,
|
| 533 |
+
"num_tokens": 6931172.0,
|
| 534 |
+
"reward": 0.08583255857229233,
|
| 535 |
+
"reward_std": 0.5651226043701172,
|
| 536 |
+
"rewards/multidomain_reward_func/mean": 0.08583252131938934,
|
| 537 |
+
"rewards/multidomain_reward_func/std": 1.92720365524292,
|
| 538 |
+
"step": 24
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"completion_length": 414.2875061035156,
|
| 542 |
+
"completions/clipped_ratio": 0.0,
|
| 543 |
+
"completions/max_length": 1183.0,
|
| 544 |
+
"completions/max_terminated_length": 1183.0,
|
| 545 |
+
"completions/mean_length": 415.2874755859375,
|
| 546 |
+
"completions/mean_terminated_length": 415.2874755859375,
|
| 547 |
+
"completions/min_length": 96.0,
|
| 548 |
+
"completions/min_terminated_length": 96.0,
|
| 549 |
+
"epoch": 0.06666666666666667,
|
| 550 |
+
"frac_reward_zero_std": 0.0,
|
| 551 |
+
"grad_norm": 0.11120733618736267,
|
| 552 |
+
"kl": 0.0,
|
| 553 |
+
"learning_rate": 9.68e-06,
|
| 554 |
+
"loss": -0.0,
|
| 555 |
+
"num_tokens": 7209577.0,
|
| 556 |
+
"reward": 0.6717811226844788,
|
| 557 |
+
"reward_std": 0.5638630390167236,
|
| 558 |
+
"rewards/multidomain_reward_func/mean": 0.6717811822891235,
|
| 559 |
+
"rewards/multidomain_reward_func/std": 1.0027689933776855,
|
| 560 |
+
"step": 25
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"completion_length": 445.37500762939453,
|
| 564 |
+
"completions/clipped_ratio": 0.0,
|
| 565 |
+
"completions/max_length": 1532.0,
|
| 566 |
+
"completions/max_terminated_length": 1532.0,
|
| 567 |
+
"completions/mean_length": 446.375,
|
| 568 |
+
"completions/mean_terminated_length": 446.375,
|
| 569 |
+
"completions/min_length": 147.0,
|
| 570 |
+
"completions/min_terminated_length": 147.0,
|
| 571 |
+
"epoch": 0.06933333333333333,
|
| 572 |
+
"frac_reward_zero_std": 0.0,
|
| 573 |
+
"grad_norm": 0.10715946555137634,
|
| 574 |
+
"kl": 0.0,
|
| 575 |
+
"learning_rate": 9.666666666666667e-06,
|
| 576 |
+
"loss": 0.0,
|
| 577 |
+
"num_tokens": 7514747.0,
|
| 578 |
+
"reward": 0.6113654971122742,
|
| 579 |
+
"reward_std": 0.6842783093452454,
|
| 580 |
+
"rewards/multidomain_reward_func/mean": 0.6113654971122742,
|
| 581 |
+
"rewards/multidomain_reward_func/std": 1.0898476839065552,
|
| 582 |
+
"step": 26
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"completion_length": 404.92500762939454,
|
| 586 |
+
"completions/clipped_ratio": 0.0,
|
| 587 |
+
"completions/max_length": 1058.0,
|
| 588 |
+
"completions/max_terminated_length": 1058.0,
|
| 589 |
+
"completions/mean_length": 405.92498779296875,
|
| 590 |
+
"completions/mean_terminated_length": 405.92498779296875,
|
| 591 |
+
"completions/min_length": 115.0,
|
| 592 |
+
"completions/min_terminated_length": 115.0,
|
| 593 |
+
"epoch": 0.072,
|
| 594 |
+
"frac_reward_zero_std": 0.0,
|
| 595 |
+
"grad_norm": 0.10682351142168045,
|
| 596 |
+
"kl": 0.0,
|
| 597 |
+
"learning_rate": 9.653333333333335e-06,
|
| 598 |
+
"loss": 0.0,
|
| 599 |
+
"num_tokens": 7779577.0,
|
| 600 |
+
"reward": 0.521935760974884,
|
| 601 |
+
"reward_std": 0.5503350496292114,
|
| 602 |
+
"rewards/multidomain_reward_func/mean": 0.5219358205795288,
|
| 603 |
+
"rewards/multidomain_reward_func/std": 1.1004360914230347,
|
| 604 |
+
"step": 27
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"completion_length": 462.3850036621094,
|
| 608 |
+
"completions/clipped_ratio": 0.0,
|
| 609 |
+
"completions/max_length": 1845.0,
|
| 610 |
+
"completions/max_terminated_length": 1845.0,
|
| 611 |
+
"completions/mean_length": 463.3849792480469,
|
| 612 |
+
"completions/mean_terminated_length": 463.3849792480469,
|
| 613 |
+
"completions/min_length": 120.0,
|
| 614 |
+
"completions/min_terminated_length": 120.0,
|
| 615 |
+
"epoch": 0.07466666666666667,
|
| 616 |
+
"frac_reward_zero_std": 0.0,
|
| 617 |
+
"grad_norm": 0.1338387280702591,
|
| 618 |
+
"kl": 0.0,
|
| 619 |
+
"learning_rate": 9.640000000000001e-06,
|
| 620 |
+
"loss": 0.0,
|
| 621 |
+
"num_tokens": 8087041.0,
|
| 622 |
+
"reward": 0.6461222767829895,
|
| 623 |
+
"reward_std": 0.7280974984169006,
|
| 624 |
+
"rewards/multidomain_reward_func/mean": 0.6461222767829895,
|
| 625 |
+
"rewards/multidomain_reward_func/std": 1.1095894575119019,
|
| 626 |
+
"step": 28
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"completion_length": 466.19500732421875,
|
| 630 |
+
"completions/clipped_ratio": 0.0024999999441206455,
|
| 631 |
+
"completions/max_length": 2048.0,
|
| 632 |
+
"completions/max_terminated_length": 1823.0,
|
| 633 |
+
"completions/mean_length": 467.1924743652344,
|
| 634 |
+
"completions/mean_terminated_length": 463.2305603027344,
|
| 635 |
+
"completions/min_length": 125.0,
|
| 636 |
+
"completions/min_terminated_length": 125.0,
|
| 637 |
+
"epoch": 0.07733333333333334,
|
| 638 |
+
"frac_reward_zero_std": 0.0,
|
| 639 |
+
"grad_norm": 0.11648301780223846,
|
| 640 |
+
"kl": 0.0,
|
| 641 |
+
"learning_rate": 9.626666666666667e-06,
|
| 642 |
+
"loss": 0.0,
|
| 643 |
+
"num_tokens": 8389638.0,
|
| 644 |
+
"reward": 0.763523519039154,
|
| 645 |
+
"reward_std": 0.6302433013916016,
|
| 646 |
+
"rewards/multidomain_reward_func/mean": 0.7635235786437988,
|
| 647 |
+
"rewards/multidomain_reward_func/std": 0.9619850516319275,
|
| 648 |
+
"step": 29
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"completion_length": 463.572509765625,
|
| 652 |
+
"completions/clipped_ratio": 0.0,
|
| 653 |
+
"completions/max_length": 1148.0,
|
| 654 |
+
"completions/max_terminated_length": 1148.0,
|
| 655 |
+
"completions/mean_length": 464.5724792480469,
|
| 656 |
+
"completions/mean_terminated_length": 464.5724792480469,
|
| 657 |
+
"completions/min_length": 118.0,
|
| 658 |
+
"completions/min_terminated_length": 118.0,
|
| 659 |
+
"epoch": 0.08,
|
| 660 |
+
"frac_reward_zero_std": 0.0,
|
| 661 |
+
"grad_norm": 0.10752178728580475,
|
| 662 |
+
"kl": 0.0,
|
| 663 |
+
"learning_rate": 9.613333333333335e-06,
|
| 664 |
+
"loss": -0.0,
|
| 665 |
+
"num_tokens": 8699697.0,
|
| 666 |
+
"reward": 0.6276810169219971,
|
| 667 |
+
"reward_std": 0.6993708610534668,
|
| 668 |
+
"rewards/multidomain_reward_func/mean": 0.6276810169219971,
|
| 669 |
+
"rewards/multidomain_reward_func/std": 1.024697184562683,
|
| 670 |
+
"step": 30
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"completion_length": 433.10250701904295,
|
| 674 |
+
"completions/clipped_ratio": 0.0,
|
| 675 |
+
"completions/max_length": 1404.0,
|
| 676 |
+
"completions/max_terminated_length": 1404.0,
|
| 677 |
+
"completions/mean_length": 434.10247802734375,
|
| 678 |
+
"completions/mean_terminated_length": 434.10247802734375,
|
| 679 |
+
"completions/min_length": 152.0,
|
| 680 |
+
"completions/min_terminated_length": 152.0,
|
| 681 |
+
"epoch": 0.08266666666666667,
|
| 682 |
+
"frac_reward_zero_std": 0.0,
|
| 683 |
+
"grad_norm": 0.1253821700811386,
|
| 684 |
+
"kl": 0.0,
|
| 685 |
+
"learning_rate": 9.600000000000001e-06,
|
| 686 |
+
"loss": 0.0,
|
| 687 |
+
"num_tokens": 8978008.0,
|
| 688 |
+
"reward": 0.7020823359489441,
|
| 689 |
+
"reward_std": 0.6984261274337769,
|
| 690 |
+
"rewards/multidomain_reward_func/mean": 0.7020823955535889,
|
| 691 |
+
"rewards/multidomain_reward_func/std": 1.1664484739303589,
|
| 692 |
+
"step": 31
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"completion_length": 493.20751037597654,
|
| 696 |
+
"completions/clipped_ratio": 0.0,
|
| 697 |
+
"completions/max_length": 1631.0,
|
| 698 |
+
"completions/max_terminated_length": 1631.0,
|
| 699 |
+
"completions/mean_length": 494.2074890136719,
|
| 700 |
+
"completions/mean_terminated_length": 494.2074890136719,
|
| 701 |
+
"completions/min_length": 132.0,
|
| 702 |
+
"completions/min_terminated_length": 132.0,
|
| 703 |
+
"epoch": 0.08533333333333333,
|
| 704 |
+
"frac_reward_zero_std": 0.0,
|
| 705 |
+
"grad_norm": 0.28045377135276794,
|
| 706 |
+
"kl": 0.0,
|
| 707 |
+
"learning_rate": 9.586666666666667e-06,
|
| 708 |
+
"loss": 0.0,
|
| 709 |
+
"num_tokens": 9303251.0,
|
| 710 |
+
"reward": 0.36521032452583313,
|
| 711 |
+
"reward_std": 0.7486643195152283,
|
| 712 |
+
"rewards/multidomain_reward_func/mean": 0.3652103543281555,
|
| 713 |
+
"rewards/multidomain_reward_func/std": 1.215452790260315,
|
| 714 |
+
"step": 32
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"completion_length": 459.21251068115237,
|
| 718 |
+
"completions/clipped_ratio": 0.0,
|
| 719 |
+
"completions/max_length": 1346.0,
|
| 720 |
+
"completions/max_terminated_length": 1346.0,
|
| 721 |
+
"completions/mean_length": 460.2124938964844,
|
| 722 |
+
"completions/mean_terminated_length": 460.2124938964844,
|
| 723 |
+
"completions/min_length": 114.0,
|
| 724 |
+
"completions/min_terminated_length": 114.0,
|
| 725 |
+
"epoch": 0.088,
|
| 726 |
+
"frac_reward_zero_std": 0.0,
|
| 727 |
+
"grad_norm": 0.09916597604751587,
|
| 728 |
+
"kl": 0.0,
|
| 729 |
+
"learning_rate": 9.573333333333334e-06,
|
| 730 |
+
"loss": 0.0,
|
| 731 |
+
"num_tokens": 9600216.0,
|
| 732 |
+
"reward": 0.6524021625518799,
|
| 733 |
+
"reward_std": 0.6229043006896973,
|
| 734 |
+
"rewards/multidomain_reward_func/mean": 0.6524021625518799,
|
| 735 |
+
"rewards/multidomain_reward_func/std": 1.1347594261169434,
|
| 736 |
+
"step": 33
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"completion_length": 478.38000640869143,
|
| 740 |
+
"completions/clipped_ratio": 0.0024999999441206455,
|
| 741 |
+
"completions/max_length": 2048.0,
|
| 742 |
+
"completions/max_terminated_length": 1426.0,
|
| 743 |
+
"completions/mean_length": 479.37750244140625,
|
| 744 |
+
"completions/mean_terminated_length": 475.44610595703125,
|
| 745 |
+
"completions/min_length": 116.0,
|
| 746 |
+
"completions/min_terminated_length": 116.0,
|
| 747 |
+
"epoch": 0.09066666666666667,
|
| 748 |
+
"frac_reward_zero_std": 0.0,
|
| 749 |
+
"grad_norm": 0.09763700515031815,
|
| 750 |
+
"kl": 0.0,
|
| 751 |
+
"learning_rate": 9.56e-06,
|
| 752 |
+
"loss": 0.0,
|
| 753 |
+
"num_tokens": 9909247.0,
|
| 754 |
+
"reward": 0.5732457041740417,
|
| 755 |
+
"reward_std": 0.6212210655212402,
|
| 756 |
+
"rewards/multidomain_reward_func/mean": 0.5732457041740417,
|
| 757 |
+
"rewards/multidomain_reward_func/std": 1.1442022323608398,
|
| 758 |
+
"step": 34
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"completion_length": 530.3800079345704,
|
| 762 |
+
"completions/clipped_ratio": 0.004999999888241291,
|
| 763 |
+
"completions/max_length": 2048.0,
|
| 764 |
+
"completions/max_terminated_length": 2014.0,
|
| 765 |
+
"completions/mean_length": 531.375,
|
| 766 |
+
"completions/mean_terminated_length": 523.7537841796875,
|
| 767 |
+
"completions/min_length": 115.0,
|
| 768 |
+
"completions/min_terminated_length": 115.0,
|
| 769 |
+
"epoch": 0.09333333333333334,
|
| 770 |
+
"frac_reward_zero_std": 0.0,
|
| 771 |
+
"grad_norm": 0.12349400669336319,
|
| 772 |
+
"kl": 0.0,
|
| 773 |
+
"learning_rate": 9.546666666666668e-06,
|
| 774 |
+
"loss": 0.0,
|
| 775 |
+
"num_tokens": 10265587.0,
|
| 776 |
+
"reward": 0.45227575302124023,
|
| 777 |
+
"reward_std": 0.7804867625236511,
|
| 778 |
+
"rewards/multidomain_reward_func/mean": 0.4522757828235626,
|
| 779 |
+
"rewards/multidomain_reward_func/std": 1.2044216394424438,
|
| 780 |
+
"step": 35
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"completion_length": 488.8800079345703,
|
| 784 |
+
"completions/clipped_ratio": 0.0,
|
| 785 |
+
"completions/max_length": 2034.0,
|
| 786 |
+
"completions/max_terminated_length": 2034.0,
|
| 787 |
+
"completions/mean_length": 489.8799743652344,
|
| 788 |
+
"completions/mean_terminated_length": 489.8799743652344,
|
| 789 |
+
"completions/min_length": 132.0,
|
| 790 |
+
"completions/min_terminated_length": 132.0,
|
| 791 |
+
"epoch": 0.096,
|
| 792 |
+
"frac_reward_zero_std": 0.0,
|
| 793 |
+
"grad_norm": 0.13603246212005615,
|
| 794 |
+
"kl": 0.0,
|
| 795 |
+
"learning_rate": 9.533333333333334e-06,
|
| 796 |
+
"loss": 0.0,
|
| 797 |
+
"num_tokens": 10596189.0,
|
| 798 |
+
"reward": 0.6199676990509033,
|
| 799 |
+
"reward_std": 0.7394188046455383,
|
| 800 |
+
"rewards/multidomain_reward_func/mean": 0.6199676990509033,
|
| 801 |
+
"rewards/multidomain_reward_func/std": 1.151370644569397,
|
| 802 |
+
"step": 36
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"completion_length": 469.7675079345703,
|
| 806 |
+
"completions/clipped_ratio": 0.0,
|
| 807 |
+
"completions/max_length": 1208.0,
|
| 808 |
+
"completions/max_terminated_length": 1208.0,
|
| 809 |
+
"completions/mean_length": 470.7674865722656,
|
| 810 |
+
"completions/mean_terminated_length": 470.7674865722656,
|
| 811 |
+
"completions/min_length": 130.0,
|
| 812 |
+
"completions/min_terminated_length": 130.0,
|
| 813 |
+
"epoch": 0.09866666666666667,
|
| 814 |
+
"frac_reward_zero_std": 0.0,
|
| 815 |
+
"grad_norm": 0.09747374057769775,
|
| 816 |
+
"kl": 0.0,
|
| 817 |
+
"learning_rate": 9.52e-06,
|
| 818 |
+
"loss": 0.0,
|
| 819 |
+
"num_tokens": 10896396.0,
|
| 820 |
+
"reward": 0.935648500919342,
|
| 821 |
+
"reward_std": 0.5971238017082214,
|
| 822 |
+
"rewards/multidomain_reward_func/mean": 0.935648500919342,
|
| 823 |
+
"rewards/multidomain_reward_func/std": 0.9697657823562622,
|
| 824 |
+
"step": 37
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"completion_length": 493.3000015258789,
|
| 828 |
+
"completions/clipped_ratio": 0.0024999999441206455,
|
| 829 |
+
"completions/max_length": 2048.0,
|
| 830 |
+
"completions/max_terminated_length": 1513.0,
|
| 831 |
+
"completions/mean_length": 494.2974853515625,
|
| 832 |
+
"completions/mean_terminated_length": 490.40350341796875,
|
| 833 |
+
"completions/min_length": 125.0,
|
| 834 |
+
"completions/min_terminated_length": 125.0,
|
| 835 |
+
"epoch": 0.10133333333333333,
|
| 836 |
+
"frac_reward_zero_std": 0.0,
|
| 837 |
+
"grad_norm": 0.0883842334151268,
|
| 838 |
+
"kl": 0.0,
|
| 839 |
+
"learning_rate": 9.506666666666667e-06,
|
| 840 |
+
"loss": 0.0,
|
| 841 |
+
"num_tokens": 11215205.0,
|
| 842 |
+
"reward": 0.7769816517829895,
|
| 843 |
+
"reward_std": 0.5231021642684937,
|
| 844 |
+
"rewards/multidomain_reward_func/mean": 0.7769816517829895,
|
| 845 |
+
"rewards/multidomain_reward_func/std": 0.9222862720489502,
|
| 846 |
+
"step": 38
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"completion_length": 486.83250885009767,
|
| 850 |
+
"completions/clipped_ratio": 0.0,
|
| 851 |
+
"completions/max_length": 1406.0,
|
| 852 |
+
"completions/max_terminated_length": 1406.0,
|
| 853 |
+
"completions/mean_length": 487.8324890136719,
|
| 854 |
+
"completions/mean_terminated_length": 487.8324890136719,
|
| 855 |
+
"completions/min_length": 133.0,
|
| 856 |
+
"completions/min_terminated_length": 133.0,
|
| 857 |
+
"epoch": 0.104,
|
| 858 |
+
"frac_reward_zero_std": 0.0,
|
| 859 |
+
"grad_norm": 0.10908982902765274,
|
| 860 |
+
"kl": 0.0,
|
| 861 |
+
"learning_rate": 9.493333333333334e-06,
|
| 862 |
+
"loss": 0.0,
|
| 863 |
+
"num_tokens": 11540558.0,
|
| 864 |
+
"reward": 0.6732150912284851,
|
| 865 |
+
"reward_std": 0.6396495699882507,
|
| 866 |
+
"rewards/multidomain_reward_func/mean": 0.6732151508331299,
|
| 867 |
+
"rewards/multidomain_reward_func/std": 1.1686171293258667,
|
| 868 |
+
"step": 39
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"completion_length": 487.9100112915039,
|
| 872 |
+
"completions/clipped_ratio": 0.0,
|
| 873 |
+
"completions/max_length": 1473.0,
|
| 874 |
+
"completions/max_terminated_length": 1473.0,
|
| 875 |
+
"completions/mean_length": 488.9100036621094,
|
| 876 |
+
"completions/mean_terminated_length": 488.9100036621094,
|
| 877 |
+
"completions/min_length": 129.0,
|
| 878 |
+
"completions/min_terminated_length": 129.0,
|
| 879 |
+
"epoch": 0.10666666666666667,
|
| 880 |
+
"frac_reward_zero_std": 0.0,
|
| 881 |
+
"grad_norm": 0.10951139777898788,
|
| 882 |
+
"kl": 0.0,
|
| 883 |
+
"learning_rate": 9.48e-06,
|
| 884 |
+
"loss": 0.0,
|
| 885 |
+
"num_tokens": 11863322.0,
|
| 886 |
+
"reward": 0.6731404662132263,
|
| 887 |
+
"reward_std": 0.6585783362388611,
|
| 888 |
+
"rewards/multidomain_reward_func/mean": 0.6731404662132263,
|
| 889 |
+
"rewards/multidomain_reward_func/std": 1.0647237300872803,
|
| 890 |
+
"step": 40
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"completion_length": 502.2775161743164,
|
| 894 |
+
"completions/clipped_ratio": 0.0,
|
| 895 |
+
"completions/max_length": 1538.0,
|
| 896 |
+
"completions/max_terminated_length": 1538.0,
|
| 897 |
+
"completions/mean_length": 503.2774963378906,
|
| 898 |
+
"completions/mean_terminated_length": 503.2774963378906,
|
| 899 |
+
"completions/min_length": 145.0,
|
| 900 |
+
"completions/min_terminated_length": 145.0,
|
| 901 |
+
"epoch": 0.10933333333333334,
|
| 902 |
+
"frac_reward_zero_std": 0.0,
|
| 903 |
+
"grad_norm": 1.849650263786316,
|
| 904 |
+
"kl": 0.0,
|
| 905 |
+
"learning_rate": 9.466666666666667e-06,
|
| 906 |
+
"loss": 0.0,
|
| 907 |
+
"num_tokens": 12192863.0,
|
| 908 |
+
"reward": 0.7937699556350708,
|
| 909 |
+
"reward_std": 0.6187800168991089,
|
| 910 |
+
"rewards/multidomain_reward_func/mean": 0.7937700152397156,
|
| 911 |
+
"rewards/multidomain_reward_func/std": 1.0126259326934814,
|
| 912 |
+
"step": 41
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"completion_length": 478.65250396728516,
|
| 916 |
+
"completions/clipped_ratio": 0.0,
|
| 917 |
+
"completions/max_length": 1362.0,
|
| 918 |
+
"completions/max_terminated_length": 1362.0,
|
| 919 |
+
"completions/mean_length": 479.6524963378906,
|
| 920 |
+
"completions/mean_terminated_length": 479.6524963378906,
|
| 921 |
+
"completions/min_length": 130.0,
|
| 922 |
+
"completions/min_terminated_length": 130.0,
|
| 923 |
+
"epoch": 0.112,
|
| 924 |
+
"frac_reward_zero_std": 0.0,
|
| 925 |
+
"grad_norm": 0.12482614070177078,
|
| 926 |
+
"kl": 0.0,
|
| 927 |
+
"learning_rate": 9.453333333333335e-06,
|
| 928 |
+
"loss": 0.0,
|
| 929 |
+
"num_tokens": 12500484.0,
|
| 930 |
+
"reward": 0.6001678109169006,
|
| 931 |
+
"reward_std": 0.6798524260520935,
|
| 932 |
+
"rewards/multidomain_reward_func/mean": 0.6001678705215454,
|
| 933 |
+
"rewards/multidomain_reward_func/std": 1.120275616645813,
|
| 934 |
+
"step": 42
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"completion_length": 517.6400100708008,
|
| 938 |
+
"completions/clipped_ratio": 0.0024999999441206455,
|
| 939 |
+
"completions/max_length": 2048.0,
|
| 940 |
+
"completions/max_terminated_length": 1579.0,
|
| 941 |
+
"completions/mean_length": 518.6375122070312,
|
| 942 |
+
"completions/mean_terminated_length": 514.8045043945312,
|
| 943 |
+
"completions/min_length": 144.0,
|
| 944 |
+
"completions/min_terminated_length": 144.0,
|
| 945 |
+
"epoch": 0.11466666666666667,
|
| 946 |
+
"frac_reward_zero_std": 0.0,
|
| 947 |
+
"grad_norm": 0.09683530032634735,
|
| 948 |
+
"kl": 0.0,
|
| 949 |
+
"learning_rate": 9.440000000000001e-06,
|
| 950 |
+
"loss": 0.0,
|
| 951 |
+
"num_tokens": 12841819.0,
|
| 952 |
+
"reward": 0.753699541091919,
|
| 953 |
+
"reward_std": 0.6156747341156006,
|
| 954 |
+
"rewards/multidomain_reward_func/mean": 0.753699541091919,
|
| 955 |
+
"rewards/multidomain_reward_func/std": 0.9609552025794983,
|
| 956 |
+
"step": 43
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"completion_length": 505.51000823974607,
|
| 960 |
+
"completions/clipped_ratio": 0.0,
|
| 961 |
+
"completions/max_length": 1757.0,
|
| 962 |
+
"completions/max_terminated_length": 1757.0,
|
| 963 |
+
"completions/mean_length": 506.5099792480469,
|
| 964 |
+
"completions/mean_terminated_length": 506.5099792480469,
|
| 965 |
+
"completions/min_length": 129.0,
|
| 966 |
+
"completions/min_terminated_length": 129.0,
|
| 967 |
+
"epoch": 0.11733333333333333,
|
| 968 |
+
"frac_reward_zero_std": 0.0,
|
| 969 |
+
"grad_norm": 0.07461382448673248,
|
| 970 |
+
"kl": 0.0,
|
| 971 |
+
"learning_rate": 9.426666666666667e-06,
|
| 972 |
+
"loss": 0.0,
|
| 973 |
+
"num_tokens": 13168013.0,
|
| 974 |
+
"reward": 0.8316676616668701,
|
| 975 |
+
"reward_std": 0.47557753324508667,
|
| 976 |
+
"rewards/multidomain_reward_func/mean": 0.8316677212715149,
|
| 977 |
+
"rewards/multidomain_reward_func/std": 0.9142977595329285,
|
| 978 |
+
"step": 44
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"completion_length": 1016.6600128173828,
|
| 982 |
+
"completions/clipped_ratio": 0.3125,
|
| 983 |
+
"completions/max_length": 2048.0,
|
| 984 |
+
"completions/max_terminated_length": 1516.0,
|
| 985 |
+
"completions/mean_length": 1017.3474731445312,
|
| 986 |
+
"completions/mean_terminated_length": 548.8690795898438,
|
| 987 |
+
"completions/min_length": 6.0,
|
| 988 |
+
"completions/min_terminated_length": 6.0,
|
| 989 |
+
"epoch": 0.12,
|
| 990 |
+
"frac_reward_zero_std": 0.3499999940395355,
|
| 991 |
+
"grad_norm": 0.1530371755361557,
|
| 992 |
+
"kl": 0.0,
|
| 993 |
+
"learning_rate": 9.413333333333334e-06,
|
| 994 |
+
"loss": 0.0,
|
| 995 |
+
"num_tokens": 13716622.0,
|
| 996 |
+
"reward": -1.7727051973342896,
|
| 997 |
+
"reward_std": 0.515305757522583,
|
| 998 |
+
"rewards/multidomain_reward_func/mean": -1.772705078125,
|
| 999 |
+
"rewards/multidomain_reward_func/std": 2.787855386734009,
|
| 1000 |
+
"step": 45
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"completion_length": 450.5750076293945,
|
| 1004 |
+
"completions/clipped_ratio": 0.0,
|
| 1005 |
+
"completions/max_length": 1174.0,
|
| 1006 |
+
"completions/max_terminated_length": 1174.0,
|
| 1007 |
+
"completions/mean_length": 451.5749816894531,
|
| 1008 |
+
"completions/mean_terminated_length": 451.5749816894531,
|
| 1009 |
+
"completions/min_length": 138.0,
|
| 1010 |
+
"completions/min_terminated_length": 138.0,
|
| 1011 |
+
"epoch": 0.12266666666666666,
|
| 1012 |
+
"frac_reward_zero_std": 0.0,
|
| 1013 |
+
"grad_norm": 0.11144713312387466,
|
| 1014 |
+
"kl": 0.0,
|
| 1015 |
+
"learning_rate": 9.4e-06,
|
| 1016 |
+
"loss": 0.0,
|
| 1017 |
+
"num_tokens": 14013782.0,
|
| 1018 |
+
"reward": 0.7144942283630371,
|
| 1019 |
+
"reward_std": 0.6038557887077332,
|
| 1020 |
+
"rewards/multidomain_reward_func/mean": 0.7144941687583923,
|
| 1021 |
+
"rewards/multidomain_reward_func/std": 1.0141149759292603,
|
| 1022 |
+
"step": 46
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"completion_length": 462.4975051879883,
|
| 1026 |
+
"completions/clipped_ratio": 0.0024999999441206455,
|
| 1027 |
+
"completions/max_length": 2048.0,
|
| 1028 |
+
"completions/max_terminated_length": 1033.0,
|
| 1029 |
+
"completions/mean_length": 463.4949951171875,
|
| 1030 |
+
"completions/mean_terminated_length": 459.5238037109375,
|
| 1031 |
+
"completions/min_length": 124.0,
|
| 1032 |
+
"completions/min_terminated_length": 124.0,
|
| 1033 |
+
"epoch": 0.12533333333333332,
|
| 1034 |
+
"frac_reward_zero_std": 0.0,
|
| 1035 |
+
"grad_norm": 0.10090559720993042,
|
| 1036 |
+
"kl": 0.0,
|
| 1037 |
+
"learning_rate": 9.386666666666668e-06,
|
| 1038 |
+
"loss": 0.0,
|
| 1039 |
+
"num_tokens": 14317510.0,
|
| 1040 |
+
"reward": 0.6212114095687866,
|
| 1041 |
+
"reward_std": 0.6358470916748047,
|
| 1042 |
+
"rewards/multidomain_reward_func/mean": 0.6212114095687866,
|
| 1043 |
+
"rewards/multidomain_reward_func/std": 1.0588219165802002,
|
| 1044 |
+
"step": 47
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"completion_length": 467.2750045776367,
|
| 1048 |
+
"completions/clipped_ratio": 0.0,
|
| 1049 |
+
"completions/max_length": 1720.0,
|
| 1050 |
+
"completions/max_terminated_length": 1720.0,
|
| 1051 |
+
"completions/mean_length": 468.2749938964844,
|
| 1052 |
+
"completions/mean_terminated_length": 468.2749938964844,
|
| 1053 |
+
"completions/min_length": 119.0,
|
| 1054 |
+
"completions/min_terminated_length": 119.0,
|
| 1055 |
+
"epoch": 0.128,
|
| 1056 |
+
"frac_reward_zero_std": 0.0,
|
| 1057 |
+
"grad_norm": 0.09495209157466888,
|
| 1058 |
+
"kl": 0.0,
|
| 1059 |
+
"learning_rate": 9.373333333333334e-06,
|
| 1060 |
+
"loss": 0.0,
|
| 1061 |
+
"num_tokens": 14621120.0,
|
| 1062 |
+
"reward": 0.9720970988273621,
|
| 1063 |
+
"reward_std": 0.6470828652381897,
|
| 1064 |
+
"rewards/multidomain_reward_func/mean": 0.9720970392227173,
|
| 1065 |
+
"rewards/multidomain_reward_func/std": 1.0593924522399902,
|
| 1066 |
+
"step": 48
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"completion_length": 490.20000610351565,
|
| 1070 |
+
"completions/clipped_ratio": 0.0,
|
| 1071 |
+
"completions/max_length": 1525.0,
|
| 1072 |
+
"completions/max_terminated_length": 1525.0,
|
| 1073 |
+
"completions/mean_length": 491.1999816894531,
|
| 1074 |
+
"completions/mean_terminated_length": 491.1999816894531,
|
| 1075 |
+
"completions/min_length": 140.0,
|
| 1076 |
+
"completions/min_terminated_length": 140.0,
|
| 1077 |
+
"epoch": 0.13066666666666665,
|
| 1078 |
+
"frac_reward_zero_std": 0.0,
|
| 1079 |
+
"grad_norm": 0.09339933842420578,
|
| 1080 |
+
"kl": 0.0,
|
| 1081 |
+
"learning_rate": 9.360000000000002e-06,
|
| 1082 |
+
"loss": 0.0,
|
| 1083 |
+
"num_tokens": 14945280.0,
|
| 1084 |
+
"reward": 0.7716130018234253,
|
| 1085 |
+
"reward_std": 0.5648047924041748,
|
| 1086 |
+
"rewards/multidomain_reward_func/mean": 0.7716130018234253,
|
| 1087 |
+
"rewards/multidomain_reward_func/std": 1.0510609149932861,
|
| 1088 |
+
"step": 49
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"completion_length": 470.83250885009767,
|
| 1092 |
+
"completions/clipped_ratio": 0.0,
|
| 1093 |
+
"completions/max_length": 1374.0,
|
| 1094 |
+
"completions/max_terminated_length": 1374.0,
|
| 1095 |
+
"completions/mean_length": 471.8324890136719,
|
| 1096 |
+
"completions/mean_terminated_length": 471.8324890136719,
|
| 1097 |
+
"completions/min_length": 123.0,
|
| 1098 |
+
"completions/min_terminated_length": 123.0,
|
| 1099 |
+
"epoch": 0.13333333333333333,
|
| 1100 |
+
"frac_reward_zero_std": 0.0,
|
| 1101 |
+
"grad_norm": 0.10589201003313065,
|
| 1102 |
+
"kl": 0.0,
|
| 1103 |
+
"learning_rate": 9.346666666666666e-06,
|
| 1104 |
+
"loss": 0.0,
|
| 1105 |
+
"num_tokens": 15252173.0,
|
| 1106 |
+
"reward": 0.8854246139526367,
|
| 1107 |
+
"reward_std": 0.5977504849433899,
|
| 1108 |
+
"rewards/multidomain_reward_func/mean": 0.8854245543479919,
|
| 1109 |
+
"rewards/multidomain_reward_func/std": 1.0996488332748413,
|
| 1110 |
+
"step": 50
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"completion_length": 487.5975112915039,
|
| 1114 |
+
"completions/clipped_ratio": 0.004999999888241291,
|
| 1115 |
+
"completions/max_length": 2048.0,
|
| 1116 |
+
"completions/max_terminated_length": 1913.0,
|
| 1117 |
+
"completions/mean_length": 488.5924987792969,
|
| 1118 |
+
"completions/mean_terminated_length": 480.75628662109375,
|
| 1119 |
+
"completions/min_length": 177.0,
|
| 1120 |
+
"completions/min_terminated_length": 177.0,
|
| 1121 |
+
"epoch": 0.136,
|
| 1122 |
+
"frac_reward_zero_std": 0.0,
|
| 1123 |
+
"grad_norm": 0.12346775829792023,
|
| 1124 |
+
"kl": 0.0,
|
| 1125 |
+
"learning_rate": 9.333333333333334e-06,
|
| 1126 |
+
"loss": 0.0,
|
| 1127 |
+
"num_tokens": 15564510.0,
|
| 1128 |
+
"reward": 0.48230254650115967,
|
| 1129 |
+
"reward_std": 0.6803807020187378,
|
| 1130 |
+
"rewards/multidomain_reward_func/mean": 0.4823025166988373,
|
| 1131 |
+
"rewards/multidomain_reward_func/std": 1.2003233432769775,
|
| 1132 |
+
"step": 51
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"completion_length": 459.94250640869143,
|
| 1136 |
+
"completions/clipped_ratio": 0.0,
|
| 1137 |
+
"completions/max_length": 1186.0,
|
| 1138 |
+
"completions/max_terminated_length": 1186.0,
|
| 1139 |
+
"completions/mean_length": 460.9425048828125,
|
| 1140 |
+
"completions/mean_terminated_length": 460.9425048828125,
|
| 1141 |
+
"completions/min_length": 124.0,
|
| 1142 |
+
"completions/min_terminated_length": 124.0,
|
| 1143 |
+
"epoch": 0.13866666666666666,
|
| 1144 |
+
"frac_reward_zero_std": 0.0,
|
| 1145 |
+
"grad_norm": 0.13084262609481812,
|
| 1146 |
+
"kl": 0.0,
|
| 1147 |
+
"learning_rate": 9.32e-06,
|
| 1148 |
+
"loss": 0.0,
|
| 1149 |
+
"num_tokens": 15868027.0,
|
| 1150 |
+
"reward": 0.796763002872467,
|
| 1151 |
+
"reward_std": 0.7062838077545166,
|
| 1152 |
+
"rewards/multidomain_reward_func/mean": 0.7967629432678223,
|
| 1153 |
+
"rewards/multidomain_reward_func/std": 1.2271337509155273,
|
| 1154 |
+
"step": 52
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"completion_length": 453.93250427246096,
|
| 1158 |
+
"completions/clipped_ratio": 0.0024999999441206455,
|
| 1159 |
+
"completions/max_length": 2048.0,
|
| 1160 |
+
"completions/max_terminated_length": 1102.0,
|
| 1161 |
+
"completions/mean_length": 454.92999267578125,
|
| 1162 |
+
"completions/mean_terminated_length": 450.9373474121094,
|
| 1163 |
+
"completions/min_length": 169.0,
|
| 1164 |
+
"completions/min_terminated_length": 169.0,
|
| 1165 |
+
"epoch": 0.14133333333333334,
|
| 1166 |
+
"frac_reward_zero_std": 0.0,
|
| 1167 |
+
"grad_norm": 0.14952056109905243,
|
| 1168 |
+
"kl": 0.0,
|
| 1169 |
+
"learning_rate": 9.306666666666667e-06,
|
| 1170 |
+
"loss": 0.0,
|
| 1171 |
+
"num_tokens": 16159159.0,
|
| 1172 |
+
"reward": 0.6241033673286438,
|
| 1173 |
+
"reward_std": 0.6670031547546387,
|
| 1174 |
+
"rewards/multidomain_reward_func/mean": 0.6241033673286438,
|
| 1175 |
+
"rewards/multidomain_reward_func/std": 1.1514482498168945,
|
| 1176 |
+
"step": 53
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"completion_length": 452.2375061035156,
|
| 1180 |
+
"completions/clipped_ratio": 0.0,
|
| 1181 |
+
"completions/max_length": 1412.0,
|
| 1182 |
+
"completions/max_terminated_length": 1412.0,
|
| 1183 |
+
"completions/mean_length": 453.23748779296875,
|
| 1184 |
+
"completions/mean_terminated_length": 453.23748779296875,
|
| 1185 |
+
"completions/min_length": 144.0,
|
| 1186 |
+
"completions/min_terminated_length": 144.0,
|
| 1187 |
+
"epoch": 0.144,
|
| 1188 |
+
"frac_reward_zero_std": 0.0,
|
| 1189 |
+
"grad_norm": 0.09604249149560928,
|
| 1190 |
+
"kl": 0.0,
|
| 1191 |
+
"learning_rate": 9.293333333333335e-06,
|
| 1192 |
+
"loss": -0.0,
|
| 1193 |
+
"num_tokens": 16452664.0,
|
| 1194 |
+
"reward": 0.4965752363204956,
|
| 1195 |
+
"reward_std": 0.5294793844223022,
|
| 1196 |
+
"rewards/multidomain_reward_func/mean": 0.4965752363204956,
|
| 1197 |
+
"rewards/multidomain_reward_func/std": 1.1207164525985718,
|
| 1198 |
+
"step": 54
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"completion_length": 465.0425033569336,
|
| 1202 |
+
"completions/clipped_ratio": 0.0,
|
| 1203 |
+
"completions/max_length": 1350.0,
|
| 1204 |
+
"completions/max_terminated_length": 1350.0,
|
| 1205 |
+
"completions/mean_length": 466.04248046875,
|
| 1206 |
+
"completions/mean_terminated_length": 466.04248046875,
|
| 1207 |
+
"completions/min_length": 132.0,
|
| 1208 |
+
"completions/min_terminated_length": 132.0,
|
| 1209 |
+
"epoch": 0.14666666666666667,
|
| 1210 |
+
"frac_reward_zero_std": 0.0,
|
| 1211 |
+
"grad_norm": 0.11836592108011246,
|
| 1212 |
+
"kl": 0.0,
|
| 1213 |
+
"learning_rate": 9.280000000000001e-06,
|
| 1214 |
+
"loss": 0.0,
|
| 1215 |
+
"num_tokens": 16757261.0,
|
| 1216 |
+
"reward": 0.6661089658737183,
|
| 1217 |
+
"reward_std": 0.5464302897453308,
|
| 1218 |
+
"rewards/multidomain_reward_func/mean": 0.666109025478363,
|
| 1219 |
+
"rewards/multidomain_reward_func/std": 0.9100214242935181,
|
| 1220 |
+
"step": 55
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"completion_length": 491.40500793457034,
|
| 1224 |
+
"completions/clipped_ratio": 0.0,
|
| 1225 |
+
"completions/max_length": 1505.0,
|
| 1226 |
+
"completions/max_terminated_length": 1505.0,
|
| 1227 |
+
"completions/mean_length": 492.4049987792969,
|
| 1228 |
+
"completions/mean_terminated_length": 492.4049987792969,
|
| 1229 |
+
"completions/min_length": 145.0,
|
| 1230 |
+
"completions/min_terminated_length": 145.0,
|
| 1231 |
+
"epoch": 0.14933333333333335,
|
| 1232 |
+
"frac_reward_zero_std": 0.0,
|
| 1233 |
+
"grad_norm": 0.0901034027338028,
|
| 1234 |
+
"kl": 0.0,
|
| 1235 |
+
"learning_rate": 9.266666666666667e-06,
|
| 1236 |
+
"loss": 0.0,
|
| 1237 |
+
"num_tokens": 17086573.0,
|
| 1238 |
+
"reward": 0.8763098120689392,
|
| 1239 |
+
"reward_std": 0.5632340908050537,
|
| 1240 |
+
"rewards/multidomain_reward_func/mean": 0.876309871673584,
|
| 1241 |
+
"rewards/multidomain_reward_func/std": 0.8210961818695068,
|
| 1242 |
+
"step": 56
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"completion_length": 495.40500793457034,
|
| 1246 |
+
"completions/clipped_ratio": 0.0,
|
| 1247 |
+
"completions/max_length": 1396.0,
|
| 1248 |
+
"completions/max_terminated_length": 1396.0,
|
| 1249 |
+
"completions/mean_length": 496.4049987792969,
|
| 1250 |
+
"completions/mean_terminated_length": 496.4049987792969,
|
| 1251 |
+
"completions/min_length": 168.0,
|
| 1252 |
+
"completions/min_terminated_length": 168.0,
|
| 1253 |
+
"epoch": 0.152,
|
| 1254 |
+
"frac_reward_zero_std": 0.0,
|
| 1255 |
+
"grad_norm": 0.09418133646249771,
|
| 1256 |
+
"kl": 0.0,
|
| 1257 |
+
"learning_rate": 9.253333333333333e-06,
|
| 1258 |
+
"loss": 0.0,
|
| 1259 |
+
"num_tokens": 17411285.0,
|
| 1260 |
+
"reward": 0.7451098561286926,
|
| 1261 |
+
"reward_std": 0.5770621299743652,
|
| 1262 |
+
"rewards/multidomain_reward_func/mean": 0.7451097965240479,
|
| 1263 |
+
"rewards/multidomain_reward_func/std": 0.9386699795722961,
|
| 1264 |
+
"step": 57
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"completion_length": 1361.317529296875,
|
| 1268 |
+
"completions/clipped_ratio": 0.5299999713897705,
|
| 1269 |
+
"completions/max_length": 2048.0,
|
| 1270 |
+
"completions/max_terminated_length": 1974.0,
|
| 1271 |
+
"completions/mean_length": 1361.7874755859375,
|
| 1272 |
+
"completions/mean_terminated_length": 587.973388671875,
|
| 1273 |
+
"completions/min_length": 5.0,
|
| 1274 |
+
"completions/min_terminated_length": 5.0,
|
| 1275 |
+
"epoch": 0.15466666666666667,
|
| 1276 |
+
"frac_reward_zero_std": 0.6499999761581421,
|
| 1277 |
+
"grad_norm": 0.1710672676563263,
|
| 1278 |
+
"kl": 0.0,
|
| 1279 |
+
"learning_rate": 9.240000000000001e-06,
|
| 1280 |
+
"loss": 0.0,
|
| 1281 |
+
"num_tokens": 18114280.0,
|
| 1282 |
+
"reward": -3.2304539680480957,
|
| 1283 |
+
"reward_std": 0.3466276526451111,
|
| 1284 |
+
"rewards/multidomain_reward_func/mean": -3.230454206466675,
|
| 1285 |
+
"rewards/multidomain_reward_func/std": 2.6598012447357178,
|
| 1286 |
+
"step": 58
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"completion_length": 459.1775100708008,
|
| 1290 |
+
"completions/clipped_ratio": 0.0,
|
| 1291 |
+
"completions/max_length": 1182.0,
|
| 1292 |
+
"completions/max_terminated_length": 1182.0,
|
| 1293 |
+
"completions/mean_length": 460.177490234375,
|
| 1294 |
+
"completions/mean_terminated_length": 460.177490234375,
|
| 1295 |
+
"completions/min_length": 172.0,
|
| 1296 |
+
"completions/min_terminated_length": 172.0,
|
| 1297 |
+
"epoch": 0.15733333333333333,
|
| 1298 |
+
"frac_reward_zero_std": 0.0,
|
| 1299 |
+
"grad_norm": 0.18426430225372314,
|
| 1300 |
+
"kl": 0.0,
|
| 1301 |
+
"learning_rate": 9.226666666666668e-06,
|
| 1302 |
+
"loss": 0.0,
|
| 1303 |
+
"num_tokens": 18410481.0,
|
| 1304 |
+
"reward": 0.7067541480064392,
|
| 1305 |
+
"reward_std": 0.5836679935455322,
|
| 1306 |
+
"rewards/multidomain_reward_func/mean": 0.706754207611084,
|
| 1307 |
+
"rewards/multidomain_reward_func/std": 0.9405755996704102,
|
| 1308 |
+
"step": 59
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"completion_length": 518.2050018310547,
|
| 1312 |
+
"completions/clipped_ratio": 0.0,
|
| 1313 |
+
"completions/max_length": 1761.0,
|
| 1314 |
+
"completions/max_terminated_length": 1761.0,
|
| 1315 |
+
"completions/mean_length": 519.2050170898438,
|
| 1316 |
+
"completions/mean_terminated_length": 519.2050170898438,
|
| 1317 |
+
"completions/min_length": 150.0,
|
| 1318 |
+
"completions/min_terminated_length": 150.0,
|
| 1319 |
+
"epoch": 0.16,
|
| 1320 |
+
"frac_reward_zero_std": 0.0,
|
| 1321 |
+
"grad_norm": 0.10580391436815262,
|
| 1322 |
+
"kl": 0.0,
|
| 1323 |
+
"learning_rate": 9.213333333333334e-06,
|
| 1324 |
+
"loss": 0.0,
|
| 1325 |
+
"num_tokens": 18777053.0,
|
| 1326 |
+
"reward": 0.6891916990280151,
|
| 1327 |
+
"reward_std": 0.6400912404060364,
|
| 1328 |
+
"rewards/multidomain_reward_func/mean": 0.6891917586326599,
|
| 1329 |
+
"rewards/multidomain_reward_func/std": 1.0435802936553955,
|
| 1330 |
+
"step": 60
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"completion_length": 488.422509765625,
|
| 1334 |
+
"completions/clipped_ratio": 0.0,
|
| 1335 |
+
"completions/max_length": 1770.0,
|
| 1336 |
+
"completions/max_terminated_length": 1770.0,
|
| 1337 |
+
"completions/mean_length": 489.4224853515625,
|
| 1338 |
+
"completions/mean_terminated_length": 489.4224853515625,
|
| 1339 |
+
"completions/min_length": 149.0,
|
| 1340 |
+
"completions/min_terminated_length": 149.0,
|
| 1341 |
+
"epoch": 0.16266666666666665,
|
| 1342 |
+
"frac_reward_zero_std": 0.0,
|
| 1343 |
+
"grad_norm": 0.11052470654249191,
|
| 1344 |
+
"kl": 0.0,
|
| 1345 |
+
"learning_rate": 9.200000000000002e-06,
|
| 1346 |
+
"loss": 0.0,
|
| 1347 |
+
"num_tokens": 19105212.0,
|
| 1348 |
+
"reward": 0.7272980809211731,
|
| 1349 |
+
"reward_std": 0.6313333511352539,
|
| 1350 |
+
"rewards/multidomain_reward_func/mean": 0.7272981405258179,
|
| 1351 |
+
"rewards/multidomain_reward_func/std": 0.9895555973052979,
|
| 1352 |
+
"step": 61
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"completion_length": 499.8575073242188,
|
| 1356 |
+
"completions/clipped_ratio": 0.0024999999441206455,
|
| 1357 |
+
"completions/max_length": 2048.0,
|
| 1358 |
+
"completions/max_terminated_length": 1807.0,
|
| 1359 |
+
"completions/mean_length": 500.85498046875,
|
| 1360 |
+
"completions/mean_terminated_length": 496.9774475097656,
|
| 1361 |
+
"completions/min_length": 142.0,
|
| 1362 |
+
"completions/min_terminated_length": 142.0,
|
| 1363 |
+
"epoch": 0.16533333333333333,
|
| 1364 |
+
"frac_reward_zero_std": 0.0,
|
| 1365 |
+
"grad_norm": 0.09797242283821106,
|
| 1366 |
+
"kl": 0.0,
|
| 1367 |
+
"learning_rate": 9.186666666666666e-06,
|
| 1368 |
+
"loss": -0.0,
|
| 1369 |
+
"num_tokens": 19439224.0,
|
| 1370 |
+
"reward": 0.7694770693778992,
|
| 1371 |
+
"reward_std": 0.6147860288619995,
|
| 1372 |
+
"rewards/multidomain_reward_func/mean": 0.7694770693778992,
|
| 1373 |
+
"rewards/multidomain_reward_func/std": 0.9149509072303772,
|
| 1374 |
+
"step": 62
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"completion_length": 470.8975067138672,
|
| 1378 |
+
"completions/clipped_ratio": 0.0,
|
| 1379 |
+
"completions/max_length": 982.0,
|
| 1380 |
+
"completions/max_terminated_length": 982.0,
|
| 1381 |
+
"completions/mean_length": 471.8974914550781,
|
| 1382 |
+
"completions/mean_terminated_length": 471.8974914550781,
|
| 1383 |
+
"completions/min_length": 117.0,
|
| 1384 |
+
"completions/min_terminated_length": 117.0,
|
| 1385 |
+
"epoch": 0.168,
|
| 1386 |
+
"frac_reward_zero_std": 0.0,
|
| 1387 |
+
"grad_norm": 0.12676118314266205,
|
| 1388 |
+
"kl": 0.0,
|
| 1389 |
+
"learning_rate": 9.173333333333334e-06,
|
| 1390 |
+
"loss": 0.0,
|
| 1391 |
+
"num_tokens": 19736443.0,
|
| 1392 |
+
"reward": 0.8108384609222412,
|
| 1393 |
+
"reward_std": 0.6640507578849792,
|
| 1394 |
+
"rewards/multidomain_reward_func/mean": 0.8108384609222412,
|
| 1395 |
+
"rewards/multidomain_reward_func/std": 1.193241834640503,
|
| 1396 |
+
"step": 63
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"completion_length": 498.67500762939454,
|
| 1400 |
+
"completions/clipped_ratio": 0.007499999832361937,
|
| 1401 |
+
"completions/max_length": 2048.0,
|
| 1402 |
+
"completions/max_terminated_length": 1361.0,
|
| 1403 |
+
"completions/mean_length": 499.66748046875,
|
| 1404 |
+
"completions/mean_terminated_length": 487.96722412109375,
|
| 1405 |
+
"completions/min_length": 166.0,
|
| 1406 |
+
"completions/min_terminated_length": 166.0,
|
| 1407 |
+
"epoch": 0.17066666666666666,
|
| 1408 |
+
"frac_reward_zero_std": 0.0,
|
| 1409 |
+
"grad_norm": 0.13014905154705048,
|
| 1410 |
+
"kl": 0.0,
|
| 1411 |
+
"learning_rate": 9.16e-06,
|
| 1412 |
+
"loss": -0.0,
|
| 1413 |
+
"num_tokens": 20057610.0,
|
| 1414 |
+
"reward": 0.6247091293334961,
|
| 1415 |
+
"reward_std": 0.7679011225700378,
|
| 1416 |
+
"rewards/multidomain_reward_func/mean": 0.6247091889381409,
|
| 1417 |
+
"rewards/multidomain_reward_func/std": 1.1993564367294312,
|
| 1418 |
+
"step": 64
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"completion_length": 458.45250549316404,
|
| 1422 |
+
"completions/clipped_ratio": 0.0,
|
| 1423 |
+
"completions/max_length": 1030.0,
|
| 1424 |
+
"completions/max_terminated_length": 1030.0,
|
| 1425 |
+
"completions/mean_length": 459.4524841308594,
|
| 1426 |
+
"completions/mean_terminated_length": 459.4524841308594,
|
| 1427 |
+
"completions/min_length": 139.0,
|
| 1428 |
+
"completions/min_terminated_length": 139.0,
|
| 1429 |
+
"epoch": 0.17333333333333334,
|
| 1430 |
+
"frac_reward_zero_std": 0.0,
|
| 1431 |
+
"grad_norm": 0.13847854733467102,
|
| 1432 |
+
"kl": 0.0,
|
| 1433 |
+
"learning_rate": 9.146666666666667e-06,
|
| 1434 |
+
"loss": 0.0,
|
| 1435 |
+
"num_tokens": 20372321.0,
|
| 1436 |
+
"reward": 0.7129409909248352,
|
| 1437 |
+
"reward_std": 0.617203950881958,
|
| 1438 |
+
"rewards/multidomain_reward_func/mean": 0.7129409909248352,
|
| 1439 |
+
"rewards/multidomain_reward_func/std": 1.0829230546951294,
|
| 1440 |
+
"step": 65
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"completion_length": 496.3400085449219,
|
| 1444 |
+
"completions/clipped_ratio": 0.0,
|
| 1445 |
+
"completions/max_length": 1668.0,
|
| 1446 |
+
"completions/max_terminated_length": 1668.0,
|
| 1447 |
+
"completions/mean_length": 497.3399963378906,
|
| 1448 |
+
"completions/mean_terminated_length": 497.3399963378906,
|
| 1449 |
+
"completions/min_length": 153.0,
|
| 1450 |
+
"completions/min_terminated_length": 153.0,
|
| 1451 |
+
"epoch": 0.176,
|
| 1452 |
+
"frac_reward_zero_std": 0.0,
|
| 1453 |
+
"grad_norm": 0.10193447768688202,
|
| 1454 |
+
"kl": 0.0,
|
| 1455 |
+
"learning_rate": 9.133333333333335e-06,
|
| 1456 |
+
"loss": 0.0,
|
| 1457 |
+
"num_tokens": 20709467.0,
|
| 1458 |
+
"reward": 0.6699061393737793,
|
| 1459 |
+
"reward_std": 0.6168271899223328,
|
| 1460 |
+
"rewards/multidomain_reward_func/mean": 0.6699060797691345,
|
| 1461 |
+
"rewards/multidomain_reward_func/std": 1.0752849578857422,
|
| 1462 |
+
"step": 66
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"completion_length": 481.9150100708008,
|
| 1466 |
+
"completions/clipped_ratio": 0.0,
|
| 1467 |
+
"completions/max_length": 1616.0,
|
| 1468 |
+
"completions/max_terminated_length": 1616.0,
|
| 1469 |
+
"completions/mean_length": 482.91497802734375,
|
| 1470 |
+
"completions/mean_terminated_length": 482.91497802734375,
|
| 1471 |
+
"completions/min_length": 139.0,
|
| 1472 |
+
"completions/min_terminated_length": 139.0,
|
| 1473 |
+
"epoch": 0.17866666666666667,
|
| 1474 |
+
"frac_reward_zero_std": 0.0,
|
| 1475 |
+
"grad_norm": 0.11991851031780243,
|
| 1476 |
+
"kl": 0.0,
|
| 1477 |
+
"learning_rate": 9.12e-06,
|
| 1478 |
+
"loss": 0.0,
|
| 1479 |
+
"num_tokens": 21044563.0,
|
| 1480 |
+
"reward": 0.5581293106079102,
|
| 1481 |
+
"reward_std": 0.6784390807151794,
|
| 1482 |
+
"rewards/multidomain_reward_func/mean": 0.5581292510032654,
|
| 1483 |
+
"rewards/multidomain_reward_func/std": 1.1184920072555542,
|
| 1484 |
+
"step": 67
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"completion_length": 483.61750946044924,
|
| 1488 |
+
"completions/clipped_ratio": 0.004999999888241291,
|
| 1489 |
+
"completions/max_length": 2048.0,
|
| 1490 |
+
"completions/max_terminated_length": 1506.0,
|
| 1491 |
+
"completions/mean_length": 484.61248779296875,
|
| 1492 |
+
"completions/mean_terminated_length": 476.75628662109375,
|
| 1493 |
+
"completions/min_length": 143.0,
|
| 1494 |
+
"completions/min_terminated_length": 143.0,
|
| 1495 |
+
"epoch": 0.18133333333333335,
|
| 1496 |
+
"frac_reward_zero_std": 0.0,
|
| 1497 |
+
"grad_norm": 0.10080371052026749,
|
| 1498 |
+
"kl": 0.0,
|
| 1499 |
+
"learning_rate": 9.106666666666667e-06,
|
| 1500 |
+
"loss": 0.0,
|
| 1501 |
+
"num_tokens": 21359318.0,
|
| 1502 |
+
"reward": 0.5767732262611389,
|
| 1503 |
+
"reward_std": 0.6775339841842651,
|
| 1504 |
+
"rewards/multidomain_reward_func/mean": 0.5767732262611389,
|
| 1505 |
+
"rewards/multidomain_reward_func/std": 1.1861456632614136,
|
| 1506 |
+
"step": 68
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"completion_length": 451.9850082397461,
|
| 1510 |
+
"completions/clipped_ratio": 0.0,
|
| 1511 |
+
"completions/max_length": 1193.0,
|
| 1512 |
+
"completions/max_terminated_length": 1193.0,
|
| 1513 |
+
"completions/mean_length": 452.9849853515625,
|
| 1514 |
+
"completions/mean_terminated_length": 452.9849853515625,
|
| 1515 |
+
"completions/min_length": 157.0,
|
| 1516 |
+
"completions/min_terminated_length": 157.0,
|
| 1517 |
+
"epoch": 0.184,
|
| 1518 |
+
"frac_reward_zero_std": 0.0,
|
| 1519 |
+
"grad_norm": 0.1563381850719452,
|
| 1520 |
+
"kl": 0.0,
|
| 1521 |
+
"learning_rate": 9.093333333333333e-06,
|
| 1522 |
+
"loss": 0.0,
|
| 1523 |
+
"num_tokens": 21654302.0,
|
| 1524 |
+
"reward": 0.8275067210197449,
|
| 1525 |
+
"reward_std": 0.6079853177070618,
|
| 1526 |
+
"rewards/multidomain_reward_func/mean": 0.8275066614151001,
|
| 1527 |
+
"rewards/multidomain_reward_func/std": 0.9333595633506775,
|
| 1528 |
+
"step": 69
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"completion_length": 499.7975128173828,
|
| 1532 |
+
"completions/clipped_ratio": 0.0,
|
| 1533 |
+
"completions/max_length": 1561.0,
|
| 1534 |
+
"completions/max_terminated_length": 1561.0,
|
| 1535 |
+
"completions/mean_length": 500.7974853515625,
|
| 1536 |
+
"completions/mean_terminated_length": 500.7974853515625,
|
| 1537 |
+
"completions/min_length": 158.0,
|
| 1538 |
+
"completions/min_terminated_length": 158.0,
|
| 1539 |
+
"epoch": 0.18666666666666668,
|
| 1540 |
+
"frac_reward_zero_std": 0.0,
|
| 1541 |
+
"grad_norm": 0.11193910241127014,
|
| 1542 |
+
"kl": 0.0,
|
| 1543 |
+
"learning_rate": 9.080000000000001e-06,
|
| 1544 |
+
"loss": 0.0,
|
| 1545 |
+
"num_tokens": 21985271.0,
|
| 1546 |
+
"reward": 0.6707971096038818,
|
| 1547 |
+
"reward_std": 0.6554468870162964,
|
| 1548 |
+
"rewards/multidomain_reward_func/mean": 0.6707971096038818,
|
| 1549 |
+
"rewards/multidomain_reward_func/std": 0.9665917158126831,
|
| 1550 |
+
"step": 70
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"completion_length": 463.9925048828125,
|
| 1554 |
+
"completions/clipped_ratio": 0.0,
|
| 1555 |
+
"completions/max_length": 1094.0,
|
| 1556 |
+
"completions/max_terminated_length": 1094.0,
|
| 1557 |
+
"completions/mean_length": 464.99249267578125,
|
| 1558 |
+
"completions/mean_terminated_length": 464.99249267578125,
|
| 1559 |
+
"completions/min_length": 135.0,
|
| 1560 |
+
"completions/min_terminated_length": 135.0,
|
| 1561 |
+
"epoch": 0.18933333333333333,
|
| 1562 |
+
"frac_reward_zero_std": 0.0,
|
| 1563 |
+
"grad_norm": 0.12109918892383575,
|
| 1564 |
+
"kl": 0.0,
|
| 1565 |
+
"learning_rate": 9.066666666666667e-06,
|
| 1566 |
+
"loss": 0.0,
|
| 1567 |
+
"num_tokens": 22281108.0,
|
| 1568 |
+
"reward": 0.7637037634849548,
|
| 1569 |
+
"reward_std": 0.7281582951545715,
|
| 1570 |
+
"rewards/multidomain_reward_func/mean": 0.7637037634849548,
|
| 1571 |
+
"rewards/multidomain_reward_func/std": 1.2205768823623657,
|
| 1572 |
+
"step": 71
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"completion_length": 464.7775054931641,
|
| 1576 |
+
"completions/clipped_ratio": 0.0,
|
| 1577 |
+
"completions/max_length": 1411.0,
|
| 1578 |
+
"completions/max_terminated_length": 1411.0,
|
| 1579 |
+
"completions/mean_length": 465.7774963378906,
|
| 1580 |
+
"completions/mean_terminated_length": 465.7774963378906,
|
| 1581 |
+
"completions/min_length": 119.0,
|
| 1582 |
+
"completions/min_terminated_length": 119.0,
|
| 1583 |
+
"epoch": 0.192,
|
| 1584 |
+
"frac_reward_zero_std": 0.0,
|
| 1585 |
+
"grad_norm": 0.11413338035345078,
|
| 1586 |
+
"kl": 0.0,
|
| 1587 |
+
"learning_rate": 9.053333333333334e-06,
|
| 1588 |
+
"loss": -0.0,
|
| 1589 |
+
"num_tokens": 22588769.0,
|
| 1590 |
+
"reward": 0.7388538122177124,
|
| 1591 |
+
"reward_std": 0.647964596748352,
|
| 1592 |
+
"rewards/multidomain_reward_func/mean": 0.7388537526130676,
|
| 1593 |
+
"rewards/multidomain_reward_func/std": 1.1589411497116089,
|
| 1594 |
+
"step": 72
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"completion_length": 451.3100051879883,
|
| 1598 |
+
"completions/clipped_ratio": 0.0,
|
| 1599 |
+
"completions/max_length": 1049.0,
|
| 1600 |
+
"completions/max_terminated_length": 1049.0,
|
| 1601 |
+
"completions/mean_length": 452.30999755859375,
|
| 1602 |
+
"completions/mean_terminated_length": 452.30999755859375,
|
| 1603 |
+
"completions/min_length": 136.0,
|
| 1604 |
+
"completions/min_terminated_length": 136.0,
|
| 1605 |
+
"epoch": 0.19466666666666665,
|
| 1606 |
+
"frac_reward_zero_std": 0.0,
|
| 1607 |
+
"grad_norm": 0.10486124455928802,
|
| 1608 |
+
"kl": 0.0,
|
| 1609 |
+
"learning_rate": 9.040000000000002e-06,
|
| 1610 |
+
"loss": 0.0,
|
| 1611 |
+
"num_tokens": 22881323.0,
|
| 1612 |
+
"reward": 0.825734555721283,
|
| 1613 |
+
"reward_std": 0.5979833602905273,
|
| 1614 |
+
"rewards/multidomain_reward_func/mean": 0.8257344961166382,
|
| 1615 |
+
"rewards/multidomain_reward_func/std": 1.0813428163528442,
|
| 1616 |
+
"step": 73
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"completion_length": 460.61500854492186,
|
| 1620 |
+
"completions/clipped_ratio": 0.0,
|
| 1621 |
+
"completions/max_length": 889.0,
|
| 1622 |
+
"completions/max_terminated_length": 889.0,
|
| 1623 |
+
"completions/mean_length": 461.614990234375,
|
| 1624 |
+
"completions/mean_terminated_length": 461.614990234375,
|
| 1625 |
+
"completions/min_length": 153.0,
|
| 1626 |
+
"completions/min_terminated_length": 153.0,
|
| 1627 |
+
"epoch": 0.19733333333333333,
|
| 1628 |
+
"frac_reward_zero_std": 0.0,
|
| 1629 |
+
"grad_norm": 0.1169808879494667,
|
| 1630 |
+
"kl": 0.0,
|
| 1631 |
+
"learning_rate": 9.026666666666666e-06,
|
| 1632 |
+
"loss": 0.0,
|
| 1633 |
+
"num_tokens": 23181119.0,
|
| 1634 |
+
"reward": 0.5477174520492554,
|
| 1635 |
+
"reward_std": 0.670683741569519,
|
| 1636 |
+
"rewards/multidomain_reward_func/mean": 0.5477175116539001,
|
| 1637 |
+
"rewards/multidomain_reward_func/std": 1.1359970569610596,
|
| 1638 |
+
"step": 74
|
| 1639 |
+
}
|
| 1640 |
+
],
|
| 1641 |
+
"logging_steps": 1,
|
| 1642 |
+
"max_steps": 750,
|
| 1643 |
+
"num_input_tokens_seen": 23521351,
|
| 1644 |
+
"num_train_epochs": 2,
|
| 1645 |
+
"save_steps": 250,
|
| 1646 |
+
"stateful_callbacks": {
|
| 1647 |
+
"TrainerControl": {
|
| 1648 |
+
"args": {
|
| 1649 |
+
"should_epoch_stop": false,
|
| 1650 |
+
"should_evaluate": false,
|
| 1651 |
+
"should_log": false,
|
| 1652 |
+
"should_save": false,
|
| 1653 |
+
"should_training_stop": false
|
| 1654 |
+
},
|
| 1655 |
+
"attributes": {}
|
| 1656 |
+
}
|
| 1657 |
+
},
|
| 1658 |
+
"total_flos": 0.0,
|
| 1659 |
+
"train_batch_size": 20,
|
| 1660 |
+
"trial_name": null,
|
| 1661 |
+
"trial_params": null
|
| 1662 |
+
}
|
ckpt-10-percent/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7505
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:08a7756db7d337356b03f79cfc247284ee5c8ed3985ced618c069718c8faeac6
|
| 3 |
size 7505
|