Run 3. Outer Step 0. Inner Step 1.

config.json

@@ -268,18 +268,18 @@
     "AutoModelForCausalLM": "distributed/optimized-gpt2-500m--modeling_gpt_optimized.GPTOptim"
   },
   "block_list": [
-    5365626,
-    5365627,
-    5365628,
-    5365629,
-    5365630
+    5365644,
+    5365645,
+    5365646,
+    5365647,
+    5365648
   ],
   "block_size": 1024,
   "bos_token_id": 50256,
   "embd_pdrop": 0.1,
   "eos_token_id": 50256,
   "initializer_range": 0.02,
-  "inner_step": 0,
+  "inner_step": 1,
   "inner_steps": 0,
   "last_allreduce_block": 5363566,
   "layer_norm_epsilon": 1e-05,

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:50852309c11bba1c68b1fc599de303950b01245d3f8e70c6ec061ebf1229b0c3
-size 2944
+oid sha256:634601fa549c021cb58e0721278daca27be247af08c3bcf056cd969b1206dae5
+size 8081782026