Run 4. Outer Step 20. Inner Step 26.

config.json

@@ -268,28 +268,28 @@
     "AutoModelForCausalLM": "distributed/optimized-gpt2-500m--modeling_gpt_optimized.GPTOptim"
   },
   "block_list": [
-    5525338,
-    5525340,
-    5525341,
-    5525341,
-    5525342,
-    5525345,
-    5525346,
-    5525348,
-    5525349,
-    5525351,
-    5525353,
-    5525355,
-    5525357,
-    5525359,
-    5525360
+    5525362,
+    5525365,
+    5525366,
+    5525368,
+    5525370,
+    5525371,
+    5525373,
+    5525374,
+    5525376,
+    5525377,
+    5525378,
+    5525380,
+    5525382,
+    5525383,
+    5525385
   ],
   "block_size": 1024,
   "bos_token_id": 50256,
   "embd_pdrop": 0.1,
   "eos_token_id": 50256,
   "initializer_range": 0.02,
-  "inner_step": 24,
+  "inner_step": 26,
   "inner_steps": 0,
   "last_allreduce_block": 5523943,
   "layer_norm_epsilon": 1e-05,

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:caa82b65fd1699b1f5c32e0a8e78dc033213e757126ee39be6c1d43a7eefaa00
+oid sha256:e219ec3ac03f16d8de031842fe7794f9e1efce2dfb6e02c31a54969acff87cd5
 size 8081782026

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8b0a5eef0b616a2193fe70f2de00d32fb1a41fb081ca8259f37b5d3f11bdd638
+oid sha256:82106756061d7b601ad52113a139e306aaa253048cf38f0e6488bb65f6252be8
 size 4040701744