Run 8. Outer Step 96. Inner Step 0.

config.json

@@ -264,15 +264,14 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6418624,
-    6418636
+    6418675
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 495,
+  "inner_step": 0,
   "intermediate_size": 5632,
   "last_allreduce_block": 6412880,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:41d6f5317974803a2bd020bd11987cde06c6c96da3e587db14761e24fe28175f
+oid sha256:0cf06c405b159b7bc0d19d8ecda9d099a7f21878f598903cedaf7ef39a5e350d
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fa068824c0757275f593a5de50e45f97fecc94b4be7e959e3d9b1874587dcc15
+oid sha256:10e4d4ec2c63413c07afe87549516b279212d177159862895987322f9dde2545
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fe9db871ca62ed1565a83fe1753332b75d9b5af338ee132245076d7d9c967548
+oid sha256:2e3cceac30b0d126e342913229e6c3a0cd19ebdabd09c19cb1b530591c35a2ee
 size 4400216536