Run 8. Outer Step 133. Inner Step 12.

config.json

@@ -264,16 +264,16 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6447866,
-    6447885,
-    6447902
+    6447919,
+    6447945,
+    6447960
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 4,
+  "inner_step": 11,
   "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 sha256:95b8c8eb4290626b944ce22172b52f8210c4bf6def650e5ee884bdee5688497d
+oid sha256:a9044322614f052eff5d680f3c0be312ddd6bfc922933f735cc25d77d6888781
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:35137cf2a17a908f70870e7c63cb19e6267ca87b7a3c663e7cca3dd95eaf9863
+oid sha256:0711b424d88e4469c5f4a64fbc0e20804f3022db42168940294515ca501b3dff
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a23256392cef4712ab8f8799ffc1a1d9257c6255ac013c5c28eb870cd794f132
+oid sha256:b452060321ec3d9c46f9d140d72d7323606872164ca6b59fdc7beb61204de3e5
 size 4400216536