Run 8. Outer Step 133. Inner Step 12.

config.json

@@ -264,16 +264,16 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6447866,
-    6447885,
-    6447899
+    6447914,
+    6447938,
+    6447956
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 4,
+  "inner_step": 12,
   "intermediate_size": 5632,
   "last_allreduce_block": 6447825,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:722d54324e9cbc50c000f6ddf10d45601a95130b8bd0fde0b4e88ee312a4c884
+oid sha256:80a36edaecafa47f2fc13fcdb91246bd09d377435463c3fbca95609d9ec57380
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d750b199795cda65e0a63356c34cd069bf2378bb262a7487ce17436f2e5e80c3
+oid sha256:bc4b93776c8bf4367c717fc9f77c6820151024534c02721f52dcece8765e97b0
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:07ada625a58b58b79f9b1476c6dc2d77c68095aca4866a7bd884f790139e2e8c
+oid sha256:9a1d60e8aeb36e801d4dba67857976894cdec7190f218d964b50c18ff49a5b37
 size 4400216536