Run 8. Outer Step 75. Inner Step 16.

config.json

@@ -264,15 +264,17 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6396600,
-    6396619
+    6396652,
+    6396664,
+    6396680,
+    6396687
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 4,
+  "inner_step": 16,
   "intermediate_size": 5632,
   "last_allreduce_block": 6396547,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4e3cd59193bf924c77ee6602668ecaa8b64ba22cc94355b3f7835b7bedcabc35
+oid sha256:2167affba2105304abac23cf2bafb8d81911610eb98f48c63bb8bd9451361096
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6df19314ee5c74049ce831aedb3e42058e77b7012c128c7b83b3bbfb8edca2d7
+oid sha256:789c528708dcbaa2dedbe1f4986c729211de2a566edae67afd3f32a08bfdc606
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f280dc2a5c1643706fdb3b5d2c726df11a82e1d905eec68251fb2f371cc317f1
+oid sha256:8a013b0f563960d4fa4a1e93a78d151003d94217f838dec6f75f2f176408d51d
 size 4400216536