Run 5. Outer Step 94. Inner Step 1.

config.json

@@ -5,11 +5,11 @@
     "1": "NON_PARTICIPATING",
     "10": "NON_PARTICIPATING",
     "100": "NOT_ALIVE",
-    "101": "NON_PARTICIPATING",
+    "101": "NOT_ALIVE",
     "102": "NON_PARTICIPATING",
     "103": "NON_PARTICIPATING",
-    "104": "NOT_ALIVE",
-    "105": "NOT_ALIVE",
+    "104": "NON_PARTICIPATING",
+    "105": "NON_PARTICIPATING",
     "106": "NON_PARTICIPATING",
     "107": "NOT_ALIVE",
     "108": "NON_PARTICIPATING",
@@ -20,7 +20,7 @@
     "112": "NON_PARTICIPATING",
     "113": "NOT_ALIVE",
     "114": "NON_PARTICIPATING",
-    "115": "NOT_ALIVE",
+    "115": "NON_PARTICIPATING",
     "116": "NON_PARTICIPATING",
     "117": "NON_PARTICIPATING",
     "118": "NON_PARTICIPATING",
@@ -35,60 +35,60 @@
     "126": "NOT_ALIVE",
     "127": "NON_PARTICIPATING",
     "128": "NOT_ALIVE",
-    "129": "NOT_ALIVE",
+    "129": "NON_PARTICIPATING",
     "13": "NON_PARTICIPATING",
-    "130": "NOT_ALIVE",
+    "130": "NON_PARTICIPATING",
     "131": "NOT_ALIVE",
     "132": "NON_PARTICIPATING",
-    "133": "SUCCESS",
+    "133": "NON_PARTICIPATING",
     "134": "NON_PARTICIPATING",
     "135": "NON_PARTICIPATING",
     "136": "NON_PARTICIPATING",
-    "137": "NON_PARTICIPATING",
-    "138": "NON_PARTICIPATING",
+    "137": "NOT_ALIVE",
+    "138": "SUCCESS",
     "139": "NON_PARTICIPATING",
-    "14": "NOT_ALIVE",
+    "14": "SUCCESS",
     "140": "NOT_ALIVE",
     "141": "NON_PARTICIPATING",
-    "142": "NON_PARTICIPATING",
+    "142": "NOT_ALIVE",
     "143": "NON_PARTICIPATING",
     "144": "NON_PARTICIPATING",
     "145": "NON_PARTICIPATING",
     "146": "NON_PARTICIPATING",
-    "147": "NON_PARTICIPATING",
-    "148": "SUCCESS",
-    "149": "NOT_ALIVE",
+    "147": "NOT_ALIVE",
+    "148": "NON_PARTICIPATING",
+    "149": "NON_PARTICIPATING",
     "15": "NON_PARTICIPATING",
-    "150": "NOT_ALIVE",
+    "150": "NON_PARTICIPATING",
     "151": "NOT_ALIVE",
-    "152": "NON_PARTICIPATING",
-    "153": "NOT_ALIVE",
+    "152": "NOT_ALIVE",
+    "153": "SUCCESS",
     "154": "NON_PARTICIPATING",
     "155": "NON_PARTICIPATING",
-    "156": "NON_PARTICIPATING",
+    "156": "SUCCESS",
     "157": "NON_PARTICIPATING",
     "158": "NON_PARTICIPATING",
     "159": "NON_PARTICIPATING",
     "16": "NON_PARTICIPATING",
     "160": "NOT_ALIVE",
     "161": "NON_PARTICIPATING",
-    "162": "NOT_ALIVE",
+    "162": "NON_PARTICIPATING",
     "163": "NON_PARTICIPATING",
     "164": "NON_PARTICIPATING",
     "165": "SUCCESS",
     "166": "NON_PARTICIPATING",
     "167": "NON_PARTICIPATING",
-    "168": "NON_PARTICIPATING",
+    "168": "SUCCESS",
     "169": "NOT_ALIVE",
     "17": "NON_PARTICIPATING",
-    "170": "NOT_ALIVE",
+    "170": "NON_PARTICIPATING",
     "171": "NOT_ALIVE",
-    "172": "NON_PARTICIPATING",
-    "173": "SUCCESS",
+    "172": "NOT_ALIVE",
+    "173": "NON_PARTICIPATING",
     "174": "NON_PARTICIPATING",
     "175": "NOT_ALIVE",
     "176": "NOT_ALIVE",
-    "177": "NOT_ALIVE",
+    "177": "NON_PARTICIPATING",
     "178": "NON_PARTICIPATING",
     "179": "NON_PARTICIPATING",
     "18": "NON_PARTICIPATING",
@@ -96,31 +96,31 @@
     "181": "NOT_ALIVE",
     "182": "NOT_ALIVE",
     "183": "SUCCESS",
-    "184": "NON_PARTICIPATING",
+    "184": "NOT_ALIVE",
     "185": "NON_PARTICIPATING",
-    "186": "NON_PARTICIPATING",
+    "186": "NOT_ALIVE",
     "187": "NON_PARTICIPATING",
-    "188": "NOT_ALIVE",
+    "188": "NON_PARTICIPATING",
     "189": "NOT_ALIVE",
     "19": "NOT_ALIVE",
     "190": "NON_PARTICIPATING",
     "191": "NOT_ALIVE",
-    "192": "NOT_ALIVE",
+    "192": "NON_PARTICIPATING",
     "193": "NON_PARTICIPATING",
     "194": "NON_PARTICIPATING",
     "195": "NON_PARTICIPATING",
-    "196": "NOT_ALIVE",
-    "197": "NOT_ALIVE",
-    "198": "NOT_ALIVE",
-    "199": "NON_PARTICIPATING",
+    "196": "NON_PARTICIPATING",
+    "197": "SUCCESS",
+    "198": "NON_PARTICIPATING",
+    "199": "NOT_ALIVE",
     "2": "NON_PARTICIPATING",
     "20": "NON_PARTICIPATING",
     "200": "NOT_ALIVE",
     "201": "NOT_ALIVE",
     "202": "NON_PARTICIPATING",
-    "203": "NOT_ALIVE",
+    "203": "NON_PARTICIPATING",
     "204": "NON_PARTICIPATING",
-    "205": "NON_PARTICIPATING",
+    "205": "NOT_ALIVE",
     "206": "NON_PARTICIPATING",
     "207": "NOT_ALIVE",
     "208": "NON_PARTICIPATING",
@@ -128,73 +128,73 @@
     "21": "NOT_ALIVE",
     "210": "NOT_ALIVE",
     "211": "NON_PARTICIPATING",
-    "212": "NOT_ALIVE",
+    "212": "NON_PARTICIPATING",
     "213": "NON_PARTICIPATING",
     "214": "NON_PARTICIPATING",
     "215": "SUCCESS",
     "216": "NON_PARTICIPATING",
     "217": "NON_PARTICIPATING",
-    "218": "NON_PARTICIPATING",
-    "219": "NOT_ALIVE",
+    "218": "NOT_ALIVE",
+    "219": "NON_PARTICIPATING",
     "22": "NON_PARTICIPATING",
     "220": "NON_PARTICIPATING",
     "221": "NOT_ALIVE",
     "222": "NON_PARTICIPATING",
     "223": "NON_PARTICIPATING",
     "224": "NOT_ALIVE",
-    "225": "SUCCESS",
+    "225": "NON_PARTICIPATING",
     "226": "NOT_ALIVE",
-    "227": "NOT_ALIVE",
-    "228": "NOT_ALIVE",
+    "227": "NON_PARTICIPATING",
+    "228": "NON_PARTICIPATING",
     "229": "NON_PARTICIPATING",
     "23": "NON_PARTICIPATING",
-    "230": "NOT_ALIVE",
+    "230": "NON_PARTICIPATING",
     "231": "NOT_ALIVE",
     "232": "NOT_ALIVE",
-    "233": "NON_PARTICIPATING",
-    "234": "NON_PARTICIPATING",
-    "235": "NON_PARTICIPATING",
+    "233": "NOT_ALIVE",
+    "234": "NOT_ALIVE",
+    "235": "NOT_ALIVE",
     "236": "NON_PARTICIPATING",
     "237": "NON_PARTICIPATING",
-    "238": "FAIL",
+    "238": "SUCCESS",
     "239": "NOT_ALIVE",
-    "24": "NON_PARTICIPATING",
+    "24": "NOT_ALIVE",
     "240": "NOT_ALIVE",
     "241": "NON_PARTICIPATING",
-    "242": "NON_PARTICIPATING",
+    "242": "NOT_ALIVE",
     "243": "NOT_ALIVE",
     "244": "NON_PARTICIPATING",
     "245": "NON_PARTICIPATING",
     "246": "NON_PARTICIPATING",
-    "247": "NOT_ALIVE",
+    "247": "NON_PARTICIPATING",
     "248": "NOT_ALIVE",
-    "249": "NON_PARTICIPATING",
+    "249": "NOT_ALIVE",
     "25": "SUCCESS",
     "250": "NON_PARTICIPATING",
-    "251": "NON_PARTICIPATING",
-    "252": "NON_PARTICIPATING",
+    "251": "NOT_ALIVE",
+    "252": "NOT_ALIVE",
     "253": "NON_PARTICIPATING",
     "254": "NON_PARTICIPATING",
     "255": "NOT_ALIVE",
-    "26": "NOT_ALIVE",
+    "26": "NON_PARTICIPATING",
     "27": "NON_PARTICIPATING",
     "28": "NOT_ALIVE",
     "29": "NON_PARTICIPATING",
     "3": "NON_PARTICIPATING",
-    "30": "NOT_ALIVE",
+    "30": "NON_PARTICIPATING",
     "31": "NON_PARTICIPATING",
-    "32": "SUCCESS",
+    "32": "NON_PARTICIPATING",
     "33": "NOT_ALIVE",
-    "34": "NON_PARTICIPATING",
-    "35": "SUCCESS",
+    "34": "NOT_ALIVE",
+    "35": "NON_PARTICIPATING",
     "36": "NON_PARTICIPATING",
     "37": "NOT_ALIVE",
-    "38": "NOT_ALIVE",
-    "39": "NOT_ALIVE",
+    "38": "SUCCESS",
+    "39": "SUCCESS",
     "4": "NON_PARTICIPATING",
     "40": "NON_PARTICIPATING",
-    "41": "NON_PARTICIPATING",
-    "42": "NOT_ALIVE",
+    "41": "SUCCESS",
+    "42": "NON_PARTICIPATING",
     "43": "NON_PARTICIPATING",
     "44": "NON_PARTICIPATING",
     "45": "NOT_ALIVE",
@@ -202,46 +202,46 @@
     "47": "NON_PARTICIPATING",
     "48": "NON_PARTICIPATING",
     "49": "NON_PARTICIPATING",
-    "5": "NOT_ALIVE",
+    "5": "NON_PARTICIPATING",
     "50": "NON_PARTICIPATING",
-    "51": "NOT_ALIVE",
-    "52": "NOT_ALIVE",
+    "51": "NON_PARTICIPATING",
+    "52": "NON_PARTICIPATING",
     "53": "NON_PARTICIPATING",
-    "54": "NOT_ALIVE",
-    "55": "NON_PARTICIPATING",
-    "56": "SUCCESS",
+    "54": "NON_PARTICIPATING",
+    "55": "NOT_ALIVE",
+    "56": "NON_PARTICIPATING",
     "57": "NON_PARTICIPATING",
     "58": "NON_PARTICIPATING",
-    "59": "SUCCESS",
-    "6": "NOT_ALIVE",
+    "59": "NON_PARTICIPATING",
+    "6": "NON_PARTICIPATING",
     "60": "NON_PARTICIPATING",
     "61": "NON_PARTICIPATING",
-    "62": "NON_PARTICIPATING",
+    "62": "SUCCESS",
     "63": "SUCCESS",
-    "64": "NOT_ALIVE",
+    "64": "NON_PARTICIPATING",
     "65": "NON_PARTICIPATING",
-    "66": "NON_PARTICIPATING",
+    "66": "NOT_ALIVE",
     "67": "NOT_ALIVE",
-    "68": "NON_PARTICIPATING",
+    "68": "NOT_ALIVE",
     "69": "NON_PARTICIPATING",
-    "7": "SUCCESS",
-    "70": "NOT_ALIVE",
+    "7": "NON_PARTICIPATING",
+    "70": "NON_PARTICIPATING",
     "71": "NON_PARTICIPATING",
-    "72": "NOT_ALIVE",
+    "72": "NON_PARTICIPATING",
     "73": "NON_PARTICIPATING",
     "74": "NON_PARTICIPATING",
     "75": "NON_PARTICIPATING",
     "76": "NON_PARTICIPATING",
-    "77": "NON_PARTICIPATING",
+    "77": "NOT_ALIVE",
     "78": "NON_PARTICIPATING",
     "79": "NON_PARTICIPATING",
     "8": "NON_PARTICIPATING",
     "80": "NON_PARTICIPATING",
-    "81": "NOT_ALIVE",
+    "81": "NON_PARTICIPATING",
     "82": "NON_PARTICIPATING",
-    "83": "NOT_ALIVE",
+    "83": "NON_PARTICIPATING",
     "84": "NON_PARTICIPATING",
-    "85": "NOT_ALIVE",
+    "85": "NON_PARTICIPATING",
     "86": "NON_PARTICIPATING",
     "87": "NON_PARTICIPATING",
     "88": "NON_PARTICIPATING",
@@ -249,13 +249,13 @@
     "9": "NOT_ALIVE",
     "90": "NON_PARTICIPATING",
     "91": "NON_PARTICIPATING",
-    "92": "NON_PARTICIPATING",
-    "93": "NON_PARTICIPATING",
+    "92": "NOT_ALIVE",
+    "93": "NOT_ALIVE",
     "94": "NON_PARTICIPATING",
     "95": "NOT_ALIVE",
-    "96": "NOT_ALIVE",
+    "96": "NON_PARTICIPATING",
     "97": "NON_PARTICIPATING",
-    "98": "NOT_ALIVE",
+    "98": "NON_PARTICIPATING",
     "99": "NOT_ALIVE"
   },
   "architectures": [
@@ -264,17 +264,17 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    5945042,
-    5945053
+    5945169,
+    5945186
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 46,
+  "inner_step": 1,
   "intermediate_size": 5632,
-  "last_allreduce_block": 5943784,
+  "last_allreduce_block": 5945066,
   "max_position_embeddings": 2048,
   "mlp_bias": false,
   "model_type": "llama",

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2f9941e08cfde70e712bb3135b4e5eb0accc79e3f0a024efb99678954e28d1f5
+oid sha256:e86d54797cdfb8fc88c86a2fd435ce0dcdab4abe8de9b6016162c327c90c85fe
 size 8800564076

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ffe7f97cf2cd65c0390f7e3d57b29b6c918f84a442f3ef07d860d2ff4be5cecf
+oid sha256:5a8f102c041a568e5bc1386ebec64db57f20f0244dedb4aebbd8a14455abb91a
 size 4400216536