Run 8. Outer Step 40. Inner Step 133.

config.json

@@ -264,15 +264,16 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6362592,
-    6362607
+    6362641,
+    6362656,
+    6362682
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 125,
+  "inner_step": 133,
   "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 sha256:2dcd2bb7c044cfed2e98bc2907cef2747792c1677dbfb5cb5b5c19aa519706c6
+oid sha256:4970397e49da63ff2a0437b3fa854b03d38a3b475aa71316fb79613bba815ad7
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:484621424aa410d7fe08083be9b33b1edf5054b6da96b57e0551f15a94805791
+oid sha256:af611b1c45a390ed202be6e18867dfed725971abc64ad494b7f76de88163d706
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ef49591c1466d91cb0fceaaae499eeca19d612a34c060919fb6d0d5ef2b05ad9
+oid sha256:73f55cc4aed032c21af8f98171869d38983460f70f2755217bc6238574c571cb
 size 4400216536