Run 11. Outer Step 350. Inner Step 0.

config.json

@@ -264,8 +264,7 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6722150,
-    6722182
+    6722212
   ],
   "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": 6719484,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e4ace8c7c52a53f5ba8468ca15da37fc1e2cbde6dfb30b0c01fdaa599d4716ee
+oid sha256:efe5a647ca7ebd6ca90642f6177d6ad94c13dc98fbdd7a256c4da4c065e28ded
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1bb5c1ed598f424dbbf86775623f545462b4b570555dc1365a1f5b46482cde20
+oid sha256:47e49711118a0c29be81172f8205f8d10ef35bc84c6c604d5c2a308a10d5479c
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6d6d732fef13a919dee3eec5d1c6dc9ae25b8f5a798397db6e256485199916fe
+oid sha256:dc542cf08a2440dcc54904906dd99ff2f7e5775fd407dd1f318f2368e5dc0c6b
 size 4400216536