Run 10. Outer Step 2. Inner Step 0.

config.json

@@ -264,16 +264,17 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6464223,
-    6464235,
-    6464248
+    6464265,
+    6464277,
+    6464300,
+    6464355
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 136,
+  "inner_step": 0,
   "intermediate_size": 5632,
   "last_allreduce_block": 6463065,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:40a8dee9ba7b8211ca6b0953571da4926602e05954a6560b7d3b67e0b2c4d726
+oid sha256:393212582b7f97b4bae6873b7518d4bb8f64b37f639e0ce7225a974d4c57e0b2
 size 88390091

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e7c5a0b41929f718a788bd28eab48e4307cd75a3fa677bedde653b1e05c25ec2
+oid sha256:9a0f677a1d258a9a7da7e8d8c8bff1e0e91b338d620cd1aec299bb2e94404502
 size 4924928047

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0906ade31739c25e22f83d4d68cdb54bdb32fc8fead6193c2509ce94d7fc34e8
+oid sha256:c488516121c2465d27d802486f4e1bf79f0969b457b427a1cec7a0d091ca3517
 size 4400216536