Run 11. Outer Step 410. Inner Step 658.

config.json

@@ -264,8 +264,6 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6755407,
-    6755430,
     6755465,
     6755497,
     6755527,
@@ -274,7 +272,9 @@
     6755622,
     6755661,
     6755694,
-    6755717
+    6755717,
+    6755753,
+    6755775
   ],
   "bos_token_id": 1,
   "dtype": "float32",
@@ -283,7 +283,7 @@
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 648,
+  "inner_step": 657,
   "intermediate_size": 5632,
   "last_allreduce_block": 6750947,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9f86f0fcce75a2fbb53a7ec1bca6f3a8d1eb6e406cd5f5ed1f779ec2b1f5660f
+oid sha256:0eb929dc4ebb45a9d4f517231d1363440068a61a25843747f4cb0286a2ae1547
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f1c4ac83902ee7eda0b85eb8ee7a77093a79c0129ad1105ccc2ce0a351bca152
+oid sha256:b4b7414dd95bb52f064542ea6da7ff1bca68bcd384bff1ceb0b989b53c7a97ab
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e838d06ebfdba10024ef8a473989345d8e006fd5c7bce6ba33bda916b6182b5e
+oid sha256:69e4d66248615556308bd73b7f07373f1af394aa5a35a52d1119ed1864165d56
 size 4400216536