Run 8. Outer Step 40. Inner Step 265.

config.json

@@ -264,17 +264,17 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6364840,
-    6364856,
-    6364873,
-    6364889
+    6364929,
+    6364944,
+    6364960,
+    6364975
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 253,
+  "inner_step": 264,
   "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:be5646d8b8ce624319330cf82c515035b3613e39249e1963eb3d0ea91dcdef79
+oid sha256:f126e809898c93d645ccf3addc464c5e5f4c94d1ed2b4db2225cd7884b443c96
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:19dcb50acd82d60724943d539b8ddd47eb7ddff39fcf5c6f0f9302b83870611b
+oid sha256:45ab4157c2e88616ed91da903dc1d40b6271167815493e4720083f4480d8afb3
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:021f9a960d67ae77b4e61b1df2c2d9d088dfa7c7886fcc58a5d5ea43b3266a7f
+oid sha256:8f8c83366017edce22fe1f22a3ec361f8edc911709b69e43933ebc3d5b98b373
 size 4400216536