Run 11. Outer Step 410. Inner Step 667.

config.json

@@ -264,8 +264,6 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6755465,
-    6755497,
     6755527,
     6755552,
     6755587,
@@ -274,7 +272,9 @@
     6755694,
     6755717,
     6755753,
-    6755775
+    6755775,
+    6755809,
+    6755832
   ],
   "bos_token_id": 1,
   "dtype": "float32",
@@ -283,7 +283,7 @@
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 657,
+  "inner_step": 666,
   "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 sha256:0eb929dc4ebb45a9d4f517231d1363440068a61a25843747f4cb0286a2ae1547
+oid sha256:57886776c1152ace5a8fbfbe375d6f060e42ef0bf984504d902bd4c27d8ea4b4
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b4b7414dd95bb52f064542ea6da7ff1bca68bcd384bff1ceb0b989b53c7a97ab
+oid sha256:39d93e4b90cf9043de450bd31e938787689f319d8664676f6fa90b3706d1c119
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:69e4d66248615556308bd73b7f07373f1af394aa5a35a52d1119ed1864165d56
+oid sha256:fe882d9b16d5247ca33f1e27dfa9c07d0c99b9866dd759b1e593958a8f83b608
 size 4400216536