Run 11. Outer Step 200. Inner Step 23.

config.json

@@ -264,8 +264,8 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6659110,
-    6659138
+    6659204,
+    6659235
   ],
   "bos_token_id": 1,
   "dtype": "float32",
@@ -274,7 +274,7 @@
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 9,
+  "inner_step": 22,
   "intermediate_size": 5632,
   "last_allreduce_block": 6659044,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2edd53ea14f90342c3bff2d4b540cc7a8e78d07c38cf2507467ec77cba6a0291
+oid sha256:4ea72011a3432d5ca295ee8b40b448f0246f4b22a69545c19d1891676dbd4f06
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4a69b037e75d3c06a3329ddf450aeb0d7e61d47b1c752a34ddb17aad81d137a0
+oid sha256:2193263b65880d05cf2be197277b550923d07efa76fe6e54f351562206708757
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:722a2e986d9dbd0b8bba21d26b9ab266513654331bf7d9fd90e14a184052d09d
+oid sha256:468210134ed335639c781e64723175eb3c9b22a5e3e9621cc694682e8ce5fcb7
 size 4400216536