Outer Step 8. Inner Step 12. Batch Size 8

config.json

@@ -268,23 +268,23 @@
     "AutoModelForCausalLM": "distributed/optimized-gpt2-500m--modeling_gpt_optimized.GPTOptim"
   },
   "block_list": [
-    5290877,
-    5290881,
-    5290884,
-    5290887,
-    5290891,
-    5290894,
-    5290898,
-    5290902,
-    5290905,
-    5290909
+    5291103,
+    5291107,
+    5291110,
+    5291114,
+    5291118,
+    5291120,
+    5291125,
+    5291128,
+    5291132,
+    5291135
   ],
   "block_size": 1024,
   "bos_token_id": 50256,
   "embd_pdrop": 0.1,
   "eos_token_id": 50256,
   "initializer_range": 0.02,
-  "inner_step": 6,
+  "inner_step": 12,
   "inner_steps": 0,
   "last_allreduce_block": 5283755,
   "layer_norm_epsilon": 1e-05,

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0db58b13d413e62194c31076081d6fcabed039883c8dc33b396575d82b5cebe4
+oid sha256:c155068eef073dd51527454a0d0924339e1ac14d2110017a94b8159285fb07be
 size 8081782026

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:87fb520d88a83da31066d63675beda0a7f8edd77406ac6636d4f0ccc3df1a2d4
+oid sha256:aaa0a15a53da77b74bb19f781dcac1bced4b2432ae35192fc92812de76c7c5e8
 size 4040701744