Run 8. Outer Step 66. Inner Step 64.

config.json

@@ -264,15 +264,15 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6391066,
-    6391077
+    6391117,
+    6391128
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 56,
+  "inner_step": 64,
   "intermediate_size": 5632,
   "last_allreduce_block": 6390689,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c11f871e531db90b6f3645966670f0efe24fe711b400d26f50c39e3b5cbf5bc1
+oid sha256:08bfba5e9cf990f912e72824d2f34212769091a7df38c228310be63558086dd7
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:689ace9e7eb41ca605e79810f1bedc83860429320e9521462cc531308a88dfba
+oid sha256:98420dad5202f150a5e8cccba85a1f61cb466b2bd59f6e109ff6ff25f5385156
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b7982e60fcc317ca65216f60860871eeda6bfe94022b15d91b05423383c29989
+oid sha256:0baef740ca0678ea2785f214979830a4a0f82e8e215c009fd64f0c3ea8e61718
 size 4400216536