Run 6. Outer Step 97. Inner Step 129.

config.json

@@ -264,16 +264,15 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6198820,
-    6198827,
-    6198836
+    6198862,
+    6198872
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 121,
+  "inner_step": 128,
   "intermediate_size": 5632,
   "last_allreduce_block": 6198252,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9fb9ea4c3ee24abf3d6c96ec862d21861ee80eaeb64aac19fae8d44c4cab9f32
+oid sha256:a6f251bdd3fb2059e27673471788556c670a92eef0676dfe7dd02a664f00122f
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8f20b5c33f393089bb297c0943c6daca4b0e1f3ae6d8d2d3d78e53de4e7165d3
+oid sha256:d4ea6b8578764be0c3c84cc6eed737c327a50f41cbb326c7c31c2edfe855e026
 size 8800564076

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:533488967d1556ccd42d740e831fa19570fbb27b2325f1f08efce3728e26440c
+oid sha256:37c598822f5b2aa72ed4009253808af5984054433dce5404d8e008b17deb0e14
 size 4400216536