Run 8. Outer Step 130. Inner Step 64.

config.json

@@ -264,16 +264,16 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6446370,
-    6446379,
-    6446392
+    6446407,
+    6446422,
+    6446444
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 56,
+  "inner_step": 63,
   "intermediate_size": 5632,
   "last_allreduce_block": 6445839,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:478c43b370212c4e3b1fb52b9aa194098dd5edc874e51daa655bb6445882ab32
+oid sha256:f9706b8f7d9bc972803acca9211f669b2160e6c3b08f1688b37f6e5ef7c86866
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c6268dae1d40f8fb83e54d3e93b3ed6f358b5f5cf98d5319ca730e49dd5f2c26
+oid sha256:fd6307002d9f9d2faeb72336d213ddefcee518455e4dc18ca2b1995a86a7bf46
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:18d9efeb32b442ad99f539960d07b3d4091cc808ef6d6aa07c122149ae4c015f
+oid sha256:95fe698798ca37311714574537ab381ee286667d0cebb3ffc728b14b695c0bff
 size 4400216536