Run 11. Outer Step 80. Inner Step 0.

config.json

@@ -264,15 +264,16 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6584122,
-    6584151
+    6584182,
+    6584216,
+    6584247
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 31,
+  "inner_step": 0,
   "intermediate_size": 5632,
   "last_allreduce_block": 6583900,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2a7fca03e3eecaa690fa23a2da14a652255d0b64a0bb3932a0223cf90bcd775a
+oid sha256:7edcc57c0cb7ec52b7b23ab598c6344d033d65b2fb00c450c3a1181984c4b436
 size 88390091

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:36e0f04f17c4c633dea6601cc17dde902d4afdb66041e8ea14f95b20d78394bb
+oid sha256:12654d9f3235a1857aa9fef6461baba778a20d93f9c472e9b8c338a0b6c4efb8
 size 4924928047

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:58fb5a4b8be45546945ebdd0a4398ff5dc58a7e5d3b09d44b25e34e256045e60
+oid sha256:cf80a581592ebe97de5c32e4f6a15b0d927b32df77d141d06cd3674d9a50628c
 size 4400216536