Run 11. Outer Step 211. Inner Step 0.

config.json

@@ -264,8 +264,9 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6663112,
-    6663150
+    6663218,
+    6663249,
+    6663278
   ],
   "bos_token_id": 1,
   "dtype": "float32",
@@ -274,7 +275,7 @@
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 36,
+  "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:b31053507d759f0be5b7d3454755f55a3b471f220e6fc8492333324c0978c11a
+oid sha256:3b9345f6975efe7aea8f924f93364dc812a75e0c4e460c930c05d76bc8d6479c
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:62af352551ba31cbaff5bd316f92b2011de8f451d38bb19716af283732bc9a74
+oid sha256:f173e89e66651531f71c7c7de336404da146603fa7205f7b11a9be88f18a3db6
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c40941a224a51a630034b5ce98f750c115cfd3a661b785493b26fe6fc725f04b
+oid sha256:be36741848b3659ff19d5aec0f6710c955335addc748006810e97dd5ce15a450
 size 4400216536