Run 8. Outer Step 130. Inner Step 100.

config.json

@@ -264,16 +264,16 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6446618,
-    6446644,
-    6446654
+    6446678,
+    6446695,
+    6446705
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 92,
+  "inner_step": 99,
   "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 sha256:3bc3fe65e44ba133d840b70f286c9deab7e5ce9a2c9b1437d3639a066f9ab13f
+oid sha256:92eb0fdbde37ab142af9566b622da6f06c20f5a7e100b9730cb7af9000570fcc
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dd1a56400cfd2ae9e1828b05e2fcfb9366585cc063f8791a0122fc4982849b97
+oid sha256:472209db8a7ce564af8d79a902ac29616278aee1c7986d90810917b1c5cc932f
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4ca9135f6aa40eda32789ba869f8fcd7cf9070a8a7f4208338bdd2702818bf2b
+oid sha256:23a1300b7931fad9bd0781e8bbd83475255c94d3a0b1de786371abc3618c9bc9
 size 4400216536