Run 8. Outer Step 63. Inner Step 85.

config.json

@@ -264,23 +264,21 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6385235,
-    6385242,
-    6385249,
-    6385256,
-    6385265,
-    6385274,
-    6385281,
-    6385290,
-    6385299,
-    6385306
+    6385327,
+    6385335,
+    6385344,
+    6385353,
+    6385361,
+    6385370,
+    6385378,
+    6385387
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 62,
+  "inner_step": 85,
   "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:d2b3ad4529124e39796f8a8a490c353ee96889f61c7df9d0df298ff20f524e90
+oid sha256:4da6f43b0fe9318328706002e35761081e472d9d16784d390045fdcc6c166eb3
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0db7b56daa765f0cf079b15f44d4098e517669f08053538b135791eb33b1c4e6
+oid sha256:11dba823aa057a00a35864ddceb20ac498fa65ac3b87e324fa5139734c20a439
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:85674714d9680b6e0309376995390b3d2d1ed2551083a4394084b88107a3c74b
+oid sha256:76f0e7426ea3297bf0fbe3c211858b2690107b78ea4de4d20d442f271794d73e
 size 4400216536