Run 8. Outer Step 133. Inner Step 18.

config.json

@@ -264,15 +264,16 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6447901,
-    6447910
+    6447940,
+    6447955,
+    6447979
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 10,
+  "inner_step": 17,
   "intermediate_size": 5632,
   "last_allreduce_block": 6447825,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4f3bfdec0deb22675a113b9dc07e85104c48e4998c64ad9684efcd2914078b4c
+oid sha256:b65fe1f551ede9eb49fa6a7adebc32d957bd488d132da435778253ffafe070c3
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6e4420ef6b9f82bb06e25c6f7dfa1f43e1d0f58405b5bbbeef0661ed5f55db26
+oid sha256:63c04728ad2a66a0fdea09fcaefb8002991acc59998b28d9a61d913c53cb9616
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:849cf33cb1617d08cea43ffc0d86114f941492f9ad91efd5115b2c4a08a92d3a
+oid sha256:bea4d3eec6a1f5165825925c672f3dd1695e4c25813e5ff7286dbb60427d83f2
 size 4400216536