Run 8. Outer Step 50. Inner Step 25.

config.json

@@ -264,15 +264,21 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6374200,
-    6374233
+    6374274,
+    6374294,
+    6374308,
+    6374324,
+    6374335,
+    6374354,
+    6374369,
+    6374384
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 4,
+  "inner_step": 24,
   "intermediate_size": 5632,
   "last_allreduce_block": 6374115,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:900b3e9a069465663a66ddda55107f0f24af166d6d696645b49e8bf28fc521bc
+oid sha256:263e230fe08461a0411be5578de3322ea3c9a87cf5827df0124fc3875f9901e0
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7f853cb86c3e9e8fa176d82eb483427a871b8fa37016babe03c937997f77508d
+oid sha256:953063e9fa6e26e6084e686c567281daa878b7500559788ba1f7bebc02ba78f9
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cac9265ef2edcc5c7132a74a3b638216f68f9ee72cd37c02506bbac39852e1b3
+oid sha256:c227d93c8834f5e47efc66559ed805c7697ee28c4feb0a38654a289086a45302
 size 4400216536