Run 2. Outer Step 7. Inner Step 0.

config.json

@@ -1,9 +1,9 @@
 {
-  "_name_or_path": "crypt0trading/c66-h5",
+  "_name_or_path": "distributed/optimized-gpt2-1b",
   "activation_function": "gelu_new",
   "all_reduce_scores": {
     "0": "NON_PARTICIPATING",
-    "1": "NON_PARTICIPATING",
+    "1": "SUCCESS",
     "10": "NON_PARTICIPATING",
     "100": "NON_PARTICIPATING",
     "101": "NON_PARTICIPATING",
@@ -14,7 +14,7 @@
     "106": "NON_PARTICIPATING",
     "107": "NON_PARTICIPATING",
     "108": "NON_PARTICIPATING",
-    "109": "NON_PARTICIPATING",
+    "109": "SUCCESS",
     "11": "NON_PARTICIPATING",
     "110": "NON_PARTICIPATING",
     "111": "NON_PARTICIPATING",
@@ -23,7 +23,7 @@
     "114": "NON_PARTICIPATING",
     "115": "NON_PARTICIPATING",
     "116": "NON_PARTICIPATING",
-    "117": "SUCCESS",
+    "117": "NON_PARTICIPATING",
     "118": "NON_PARTICIPATING",
     "119": "NON_PARTICIPATING",
     "12": "NON_PARTICIPATING",
@@ -106,7 +106,7 @@
     "19": "NON_PARTICIPATING",
     "190": "NON_PARTICIPATING",
     "191": "NON_PARTICIPATING",
-    "192": "SUCCESS",
+    "192": "NON_PARTICIPATING",
     "193": "NON_PARTICIPATING",
     "194": "NON_PARTICIPATING",
     "195": "NON_PARTICIPATING",
@@ -116,7 +116,7 @@
     "199": "NON_PARTICIPATING",
     "2": "NON_PARTICIPATING",
     "20": "NON_PARTICIPATING",
-    "200": "SUCCESS",
+    "200": "NON_PARTICIPATING",
     "201": "NON_PARTICIPATING",
     "202": "NON_PARTICIPATING",
     "203": "NON_PARTICIPATING",
@@ -137,7 +137,7 @@
     "217": "NON_PARTICIPATING",
     "218": "NON_PARTICIPATING",
     "219": "NON_PARTICIPATING",
-    "22": "NON_PARTICIPATING",
+    "22": "SUCCESS",
     "220": "NON_PARTICIPATING",
     "221": "NON_PARTICIPATING",
     "222": "NON_PARTICIPATING",
@@ -146,7 +146,7 @@
     "225": "NON_PARTICIPATING",
     "226": "NON_PARTICIPATING",
     "227": "NON_PARTICIPATING",
-    "228": "SUCCESS",
+    "228": "NON_PARTICIPATING",
     "229": "NON_PARTICIPATING",
     "23": "SUCCESS",
     "230": "NON_PARTICIPATING",
@@ -169,7 +169,7 @@
     "246": "NON_PARTICIPATING",
     "247": "NON_PARTICIPATING",
     "248": "NON_PARTICIPATING",
-    "249": "SUCCESS",
+    "249": "NON_PARTICIPATING",
     "25": "SUCCESS",
     "250": "NON_PARTICIPATING",
     "251": "NON_PARTICIPATING",
@@ -188,7 +188,7 @@
     "33": "NON_PARTICIPATING",
     "34": "NON_PARTICIPATING",
     "35": "NON_PARTICIPATING",
-    "36": "SUCCESS",
+    "36": "NON_PARTICIPATING",
     "37": "NON_PARTICIPATING",
     "38": "NON_PARTICIPATING",
     "39": "SUCCESS",
@@ -198,7 +198,7 @@
     "42": "NON_PARTICIPATING",
     "43": "NON_PARTICIPATING",
     "44": "NON_PARTICIPATING",
-    "45": "SUCCESS",
+    "45": "NON_PARTICIPATING",
     "46": "NON_PARTICIPATING",
     "47": "NON_PARTICIPATING",
     "48": "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": [
-    5338057,
-    5338062,
-    5338066,
-    5338070,
-    5338075
-  ],
+  "block_list": [],
   "block_size": 1024,
   "bos_token_id": 50256,
   "embd_pdrop": 0.1,
   "eos_token_id": 50256,
   "initializer_range": 0.02,
-  "inner_step": 558,
+  "inner_step": 0,
   "inner_steps": 0,
-  "last_allreduce_block": 5332469,
+  "last_allreduce_block": 5339896,
   "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:f07f922908cbe3583e954b54018b0d3adbe4157195cbd3725550f6efefbfab4d
+oid sha256:ed94bc36dce83a7e9f7b2a0fa58b6d3de86f514816a5452bc24b18d96b3944e1
 size 8081782026

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5f3dd47ac89203fbd6eb6eb4e07f6c3b8df76fb7cb3e2911934218fcecfb946d
+oid sha256:bcdfb45c1c7b670e719f10e78091cec2c601ac898763796339f16712e03c08c1
 size 4040701744