Run 8. Outer Step 96. Inner Step 0.

config.json

@@ -264,15 +264,16 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6418595,
-    6418609
+    6418641,
+    6418656,
+    6418673
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 471,
+  "inner_step": 0,
   "intermediate_size": 5632,
   "last_allreduce_block": 6411158,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a26fbd3fc7aec53b7b0f2cb356e1464f2d7de00f96f61e88a011a89f896775eb
+oid sha256:a2b37816870fc9a439e166d588db0862f2f6ee0bd246a3f7b71cc9f445d434b9
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4c804d7c96658b5c32bfcbd03c858c6230d2d43e095ce577343cad15697c2977
+oid sha256:caffa2d1cbf6585bc93d60794444dcaf47eec8105052131dc5b30bbb226b6ab0
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:76bcb49706fbef7e2adf139aa2bb644cbdc2c633d3035c87aa6591c30b6a6b59
+oid sha256:71aae3db7c9b282d31e7344c4c113aac4fd72a5868e33d9479a679c1cf39bd49
 size 4400216536