Run 11. Outer Step 214. Inner Step 0.

config.json

@@ -264,8 +264,7 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6664324,
-    6664361
+    6664446
   ],
   "bos_token_id": 1,
   "dtype": "float32",
@@ -274,7 +273,7 @@
   "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:c729c536719271c3eed79596a8ca073a6616a657002209a725fecd1c9cce869f
+oid sha256:6052da1ee8d0a9d49a0ba25e91ed76f7080ec1d5ed3ddd3a25e11015939c9a15
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0225ac86f79fb1525c6b0841141325ab3341bc10f5d8b86087103234593ad496
+oid sha256:086d12f68c69f44a270f20dc277558e3620c56ec163a27b37e5628bd3236e33a
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0ae9b6c8722b718c94c80072d756240cae1fef77be2c7321334b594b56971a79
+oid sha256:0e777709a86e5ff4ffdd073d1c5a3cdc511741cd60cc2b67bdab51655a07c520
 size 4400216536