Run 8. Outer Step 75. Inner Step 0.

config.json

@@ -264,15 +264,17 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6396393,
-    6396404
+    6396442,
+    6396454,
+    6396467,
+    6396529
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 69,
+  "inner_step": 0,
   "intermediate_size": 5632,
   "last_allreduce_block": 6395861,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:67acebe8333aae21e8741ac957ced6f14e8eeed259794af51090fac1b06457c8
+oid sha256:5a920b829555ee1268b8b17d5bf41bc47e5555b215d8fe54ac6788d736f058e8
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:63cf0557c8950743a2e9ebaac4256b22f31e2bdd789ffa629c06bc67029fec9d
+oid sha256:a0cce58c6d98aabfcc596822902fe9458d2a95b153ac3abcc0a5e6226918a201
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eaab7911546b8ddc8c1dafd986e02fd33bfc060e84cbf8b791da1acd0f703ca1
+oid sha256:2dd1725e78472e16ea845b5163658f5a8c30eaa94df7a0bbbf28697e7948fbaa
 size 4400216536