Run 11. Outer Step 200. Inner Step 32.

config.json

@@ -264,8 +264,11 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6659127,
-    6659147
+    6659167,
+    6659195,
+    6659214,
+    6659234,
+    6659255
   ],
   "bos_token_id": 1,
   "dtype": "float32",
@@ -274,7 +277,7 @@
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 9,
+  "inner_step": 31,
   "intermediate_size": 5632,
   "last_allreduce_block": 6659044,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e9425e8b27b5fd2edf1a75680538dd357836941a0b1666d2123816255fdd0e49
+oid sha256:cc433d0ebe9776897654df7ac1ec7d0e67da88f2978d87cb4ef1f72f2898971b
 size 88390091

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7a407685f16ceb193df6b840c812fcbd0725dc613569e22a8ae9018bd52e7680
+oid sha256:3e55abf066ef40be3fb198a39e3f6bcdc74fb8554547137a8895fde890204e66
 size 4924928111

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6c27d2f098ebbd2b064541b45b73cb28a9b4bca92282bac6b1d07bd2786d3e65
+oid sha256:1b8d279bbb4ebd07cfff48e8360679680e8af7862fb0078cc59a3c6334801c7b
 size 4400216536