Run 8. Outer Step 54. Inner Step 0.

config.json

@@ -264,16 +264,17 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6376668,
-    6376681,
-    6376696
+    6376717,
+    6376731,
+    6376745,
+    6376788
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 33,
+  "inner_step": 0,
   "intermediate_size": 5632,
   "last_allreduce_block": 6376174,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:59b3f41db3ff2fed09b2628fe0f083f532e81fb177108515779214eb47e11bef
+oid sha256:d1cf995d2a7ec4e863b84e98a5bb9c0107b1db4a0df312ee87f48b72010354cf
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dea0e9603687eec0ec4057b5e37bcd77a0e74ec640573b6438e252f3351f7882
+oid sha256:3465ee602f7f3a69600a875f2df71694575c719de220fe1cc3c5e6988f9a0d93
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:555cf1a546b5e1b907936b84337bed72a172df1be6d8ee3d7084a8ee5b70093b
+oid sha256:41e4115a4022fe18994109cef5b5b5c331e2f009a3d02a28b6e24cf5629f5a43
 size 4400216536