Run 8. Outer Step 40. Inner Step 200.

config.json

@@ -264,30 +264,15 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6365538,
-    6365557,
-    6365573,
-    6365593,
-    6365614,
-    6365634,
-    6365656,
-    6365676,
-    6365696,
-    6365717,
-    6365739,
-    6365759,
-    6365779,
-    6365801,
-    6365824,
-    6365845,
-    6365870
+    6366247,
+    6366266
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 195,
+  "inner_step": 200,
   "intermediate_size": 5632,
   "last_allreduce_block": 6361535,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c60723145730314d77498a424f0ebef6b2310f772753117bed4ff90f0fbe1d4f
+oid sha256:25562e9ccd7b575224f68d6e32f63ae6613ca7969d72716e113b81e2e67533a8
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b7e1372483948c55a0c351f72f935853e863fc605a9bd4f3c02dbd42f85269fa
+oid sha256:4306ddbd60691f266ebd0ce73b65f1253a4890e03696392238c0acd844915bac
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7c57be6c0ecaabdf0b5ada8372cd478ee99986fea3039bc025959a7a013604b5
+oid sha256:c95f0cce0db66ab358376c30c792df4e00fcc4ac64db69582c16ac23bbfb6e0c
 size 4400216536