Run 8. Outer Step 62. Inner Step 0.

config.json

@@ -264,15 +264,16 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6383479,
-    6383494
+    6383523,
+    6383537,
+    6383553
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 80,
+  "inner_step": 0,
   "intermediate_size": 5632,
   "last_allreduce_block": 6382929,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1577c420c5ea1fcda40a34cb424042849b200cff06b4ae61eaaa2325f76bbb0e
+oid sha256:6a552bdd7f82102da8fb474297001dd9ac1cc2cb2675919d0961f855e846c413
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:79bcebc912ba6281d14934f0a9e4d0880f0e0a2c19da1fc0928ff6224b7f1c02
+oid sha256:4d004bda7545e772056ec69f454a6bd421d743bf05c68036337653ff4236d458
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fb6b5a8cb5f2f8659b8b1953508d201067f50fecf46b82111add896102784d63
+oid sha256:61a14e7f6810965822c4e0ea673b6a351ebcf79f96600722cd401afa280430a6
 size 4400216536