Run 8. Outer Step 40. Inner Step 256.

config.json

@@ -264,18 +264,18 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6366992,
-    6367012,
-    6367031,
-    6367048,
-    6367068
+    6367116,
+    6367136,
+    6367159,
+    6367179,
+    6367200
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 241,
+  "inner_step": 256,
   "intermediate_size": 5632,
   "last_allreduce_block": 6361535,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0b1276d277a527ef5a3893949fb700e34ab589e78c21d99d1e4a7183ca53aee9
+oid sha256:b6adde2c4269edda4ebd47335bb70ecbf06c82b80eaf92d56ee4960e144e4944
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:929765242d4c2efd960f40c8dc17deeba17814088bd3f951f51f586d5a965680
+oid sha256:63d54c6d5c7f677f555faa3a15c707db47740dc3b945374da6dca96479a1e6fb
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5ff78d757da60db422e05801d9f3462fc39176ad316d45a67856a5e41c90dbd0
+oid sha256:c5215c43b326ac0f9e4bc7a6be61aeb4e53821472b81dac670f8d249b268a39a
 size 4400216536