AresMicro-10k checkpoint
Browse files- README.md +2 -2
- checkpoint.pkl +1 -1
- config.json +2 -2
README.md
CHANGED
|
@@ -14,8 +14,8 @@ tags:
|
|
| 14 |
|
| 15 |
- Context: 32
|
| 16 |
- Vocab: 21
|
| 17 |
-
- PT Batch:
|
| 18 |
-
- RL Batch:
|
| 19 |
- Shared block loops: 50
|
| 20 |
- Hidden: 32
|
| 21 |
- Heads: 4
|
|
|
|
| 14 |
|
| 15 |
- Context: 32
|
| 16 |
- Vocab: 21
|
| 17 |
+
- PT Batch: 16384
|
| 18 |
+
- RL Batch: 16384
|
| 19 |
- Shared block loops: 50
|
| 20 |
- Hidden: 32
|
| 21 |
- Heads: 4
|
checkpoint.pkl
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 134052
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a08488c99bd7d6fdac0d984f1682569cc8709b78b214abbd91dec8147b9e017e
|
| 3 |
size 134052
|
config.json
CHANGED
|
@@ -18,8 +18,8 @@
|
|
| 18 |
"mlp_type": "swiglu",
|
| 19 |
"norm_type": "rmsnorm",
|
| 20 |
"rl_temperature": 0.7,
|
| 21 |
-
"pt_batch_size":
|
| 22 |
-
"rl_batch_size":
|
| 23 |
"pt_target_tokens": 5000000000,
|
| 24 |
"rl_target_tokens": 5000000000,
|
| 25 |
"benchmark_file": "purearith-1-answer-only.jsonl"
|
|
|
|
| 18 |
"mlp_type": "swiglu",
|
| 19 |
"norm_type": "rmsnorm",
|
| 20 |
"rl_temperature": 0.7,
|
| 21 |
+
"pt_batch_size": 16384,
|
| 22 |
+
"rl_batch_size": 16384,
|
| 23 |
"pt_target_tokens": 5000000000,
|
| 24 |
"rl_target_tokens": 5000000000,
|
| 25 |
"benchmark_file": "purearith-1-answer-only.jsonl"
|