Run 2. Outer Step 11. Inner Step 0.

config.json

@@ -1,20 +1,20 @@
 {
-  "_name_or_path": "crypt0trading/c66-h8",
+  "_name_or_path": "distributed/optimized-gpt2-1b",
   "activation_function": "gelu_new",
   "all_reduce_scores": {
     "0": "NON_PARTICIPATING",
-    "1": "SUCCESS",
+    "1": "NON_PARTICIPATING",
     "10": "NON_PARTICIPATING",
     "100": "NON_PARTICIPATING",
-    "101": "NON_PARTICIPATING",
-    "102": "NON_PARTICIPATING",
+    "101": "SUCCESS",
+    "102": "SUCCESS",
     "103": "NON_PARTICIPATING",
     "104": "NON_PARTICIPATING",
     "105": "NON_PARTICIPATING",
     "106": "NON_PARTICIPATING",
     "107": "NON_PARTICIPATING",
     "108": "NON_PARTICIPATING",
-    "109": "SUCCESS",
+    "109": "NON_PARTICIPATING",
     "11": "NON_PARTICIPATING",
     "110": "NON_PARTICIPATING",
     "111": "NON_PARTICIPATING",
@@ -23,7 +23,7 @@
     "114": "NON_PARTICIPATING",
     "115": "NON_PARTICIPATING",
     "116": "NON_PARTICIPATING",
-    "117": "NON_PARTICIPATING",
+    "117": "SUCCESS",
     "118": "NON_PARTICIPATING",
     "119": "NON_PARTICIPATING",
     "12": "NON_PARTICIPATING",
@@ -75,7 +75,7 @@
     "161": "NON_PARTICIPATING",
     "162": "NON_PARTICIPATING",
     "163": "NON_PARTICIPATING",
-    "164": "NON_PARTICIPATING",
+    "164": "SUCCESS",
     "165": "NON_PARTICIPATING",
     "166": "NON_PARTICIPATING",
     "167": "NON_PARTICIPATING",
@@ -89,7 +89,7 @@
     "174": "NON_PARTICIPATING",
     "175": "NON_PARTICIPATING",
     "176": "NON_PARTICIPATING",
-    "177": "NON_PARTICIPATING",
+    "177": "SUCCESS",
     "178": "NON_PARTICIPATING",
     "179": "NON_PARTICIPATING",
     "18": "NON_PARTICIPATING",
@@ -106,8 +106,8 @@
     "19": "NON_PARTICIPATING",
     "190": "NON_PARTICIPATING",
     "191": "NON_PARTICIPATING",
-    "192": "NON_PARTICIPATING",
-    "193": "NON_PARTICIPATING",
+    "192": "SUCCESS",
+    "193": "SUCCESS",
     "194": "NON_PARTICIPATING",
     "195": "NON_PARTICIPATING",
     "196": "NON_PARTICIPATING",
@@ -139,12 +139,12 @@
     "219": "NON_PARTICIPATING",
     "22": "SUCCESS",
     "220": "NON_PARTICIPATING",
-    "221": "NON_PARTICIPATING",
+    "221": "SUCCESS",
     "222": "NON_PARTICIPATING",
-    "223": "NON_PARTICIPATING",
+    "223": "FAIL",
     "224": "NON_PARTICIPATING",
     "225": "NON_PARTICIPATING",
-    "226": "NON_PARTICIPATING",
+    "226": "SUCCESS",
     "227": "NON_PARTICIPATING",
     "228": "NON_PARTICIPATING",
     "229": "NON_PARTICIPATING",
@@ -154,7 +154,7 @@
     "232": "NON_PARTICIPATING",
     "233": "NON_PARTICIPATING",
     "234": "NON_PARTICIPATING",
-    "235": "NON_PARTICIPATING",
+    "235": "SUCCESS",
     "236": "NON_PARTICIPATING",
     "237": "NON_PARTICIPATING",
     "238": "NON_PARTICIPATING",
@@ -169,9 +169,9 @@
     "246": "NON_PARTICIPATING",
     "247": "NON_PARTICIPATING",
     "248": "NON_PARTICIPATING",
-    "249": "NON_PARTICIPATING",
+    "249": "SUCCESS",
     "25": "SUCCESS",
-    "250": "NON_PARTICIPATING",
+    "250": "SUCCESS",
     "251": "NON_PARTICIPATING",
     "252": "NON_PARTICIPATING",
     "253": "NON_PARTICIPATING",
@@ -191,12 +191,12 @@
     "36": "NON_PARTICIPATING",
     "37": "NON_PARTICIPATING",
     "38": "NON_PARTICIPATING",
-    "39": "SUCCESS",
+    "39": "NON_PARTICIPATING",
     "4": "NON_PARTICIPATING",
     "40": "NON_PARTICIPATING",
     "41": "NON_PARTICIPATING",
-    "42": "NON_PARTICIPATING",
-    "43": "NON_PARTICIPATING",
+    "42": "SUCCESS",
+    "43": "SUCCESS",
     "44": "NON_PARTICIPATING",
     "45": "NON_PARTICIPATING",
     "46": "NON_PARTICIPATING",
@@ -217,12 +217,12 @@
     "6": "NON_PARTICIPATING",
     "60": "NON_PARTICIPATING",
     "61": "NON_PARTICIPATING",
-    "62": "SUCCESS",
+    "62": "NON_PARTICIPATING",
     "63": "NON_PARTICIPATING",
     "64": "NON_PARTICIPATING",
     "65": "NON_PARTICIPATING",
     "66": "NON_PARTICIPATING",
-    "67": "SUCCESS",
+    "67": "NON_PARTICIPATING",
     "68": "NON_PARTICIPATING",
     "69": "NON_PARTICIPATING",
     "7": "NON_PARTICIPATING",
@@ -237,7 +237,7 @@
     "78": "NON_PARTICIPATING",
     "79": "NON_PARTICIPATING",
     "8": "NON_PARTICIPATING",
-    "80": "NON_PARTICIPATING",
+    "80": "SUCCESS",
     "81": "NON_PARTICIPATING",
     "82": "NON_PARTICIPATING",
     "83": "NON_PARTICIPATING",
@@ -267,21 +267,15 @@
     "AutoConfig": "distributed/optimized-gpt2-500m--configuration_gpt_optimized.GPTOptimConfig",
     "AutoModelForCausalLM": "distributed/optimized-gpt2-500m--modeling_gpt_optimized.GPTOptim"
   },
-  "block_list": [
-    5363652,
-    5363657,
-    5363661,
-    5363665,
-    5363669
-  ],
+  "block_list": [],
   "block_size": 1024,
   "bos_token_id": 50256,
   "embd_pdrop": 0.1,
   "eos_token_id": 50256,
   "initializer_range": 0.02,
-  "inner_step": 26,
+  "inner_step": 0,
   "inner_steps": 0,
-  "last_allreduce_block": 5339896,
+  "last_allreduce_block": 5363508,
   "layer_norm_epsilon": 1e-05,
   "model_type": "gpt_optimized",
   "n_embd": 1280,

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2275be2a61fb29cfe39f8ef164b66708d490ae891a5db06fa5b21a07368f4f71
+oid sha256:e1e7676cf1cd9ec2c4a00c238d39c1f344ee5a7dd27ad65e72ab3c13419f95fc
 size 8081782026

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bd989a8c8b2c1a3caa8adbc358659ff57c66096a808a9e6582e8ce3bfbaa24b2
+oid sha256:7ba59bf6f504d702d1349e92b0cf12064e903119ec0fd54b44f5b0348e647f9e
 size 4040701744