Outer Step 3. Inner Step 10. Batch Size 60

config.json

@@ -268,18 +268,18 @@
     "AutoModelForCausalLM": "distributed/optimized-gpt2-500m--modeling_gpt_optimized.GPTOptim"
   },
   "block_list": [
-    5076055,
-    5076059,
-    5076062,
-    5076066,
-    5076070
+    5076086,
+    5076089,
+    5076093,
+    5076097,
+    5076101
   ],
   "block_size": 1024,
   "bos_token_id": 50256,
   "embd_pdrop": 0.1,
   "eos_token_id": 50256,
   "initializer_range": 0.02,
-  "inner_step": 5,
+  "inner_step": 10,
   "inner_steps": 0,
   "last_allreduce_block": 4016775,
   "layer_norm_epsilon": 1e-05,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:86cfa05231a9babae02d0b08b2a3f949183fa4b72807bf016c5f831ff31d740e
+oid sha256:d72750a580aa8c48fbcce3b0e6dcb99f1072981e8759db47ab379e03aeaf793d
 size 4040701744