Run 11. Outer Step 212. Inner Step 0.

config.json

@@ -264,8 +264,7 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6663561,
-    6663597
+    6663668
   ],
   "bos_token_id": 1,
   "dtype": "float32",
@@ -274,7 +273,7 @@
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 40,
+  "inner_step": 0,
   "intermediate_size": 5632,
   "last_allreduce_block": 6661467,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1b1566548a438d046345558a7bebf41edae69efb2fbf1becd1b1938a866e7766
+oid sha256:ab381f6238c047f4fae4c01e25097ea9b1162ff9311df6d4e07eae9f9f82224c
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3139aa907e6462809b00549c051ae637600c418ca5ef66f2d34fb5d6a96f2ae3
+oid sha256:0d0dc9e05ddba505828710146868b068ede811106e5c93de245d78dffa3aaab9
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:46258309a3106d46b6059034f093d26bee0fdaa423b90a598e406f275a81012a
+oid sha256:6cdadfc6f53b0a9a38f07b704573a53712eedbc4598a33e7d2e60602d49ed100
 size 4400216536