Run 8. Outer Step 76. Inner Step 125.

config.json

@@ -264,19 +264,17 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6397641,
-    6397648,
-    6397655,
-    6397663,
-    6397672,
-    6397680
+    6397701,
+    6397708,
+    6397716,
+    6397725
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 113,
+  "inner_step": 125,
   "intermediate_size": 5632,
   "last_allreduce_block": 6397200,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:675e23f844cb848a940cca542935a41d2ef1ff91f9e478159f582e56e3b18fcb
+oid sha256:2a48c77c51c89036872a0211e6fc4db68e6b77a70268e3fcda69c0d53d54b4b2
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:df24ba930ba1b16784e41b91b557504f17e137acb504a05df0ee098e3c74cab9
+oid sha256:588968118e045e701cb2b1ac27c4e138ea8482ae1cea205049eb482fed590d8c
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bf84cd84b4ff00c33e45d7621378739101b6bbbc18bbe57f74051835b92015cc
+oid sha256:262fff82d4646a078e0109b2723ce2aa1f37effbbe9417bf10c2796734656ded
 size 4400216536