Run 5. Outer Step 81. Inner Step 1.

config.json

@@ -3,98 +3,98 @@
   "all_reduce_scores": {
     "0": "NOT_ALIVE",
     "1": "NON_PARTICIPATING",
-    "10": "NON_PARTICIPATING",
+    "10": "NOT_ALIVE",
     "100": "NOT_ALIVE",
-    "101": "NON_PARTICIPATING",
+    "101": "SUCCESS",
     "102": "NOT_ALIVE",
     "103": "NON_PARTICIPATING",
     "104": "NOT_ALIVE",
     "105": "NON_PARTICIPATING",
     "106": "NOT_ALIVE",
-    "107": "NON_PARTICIPATING",
+    "107": "NOT_ALIVE",
     "108": "NON_PARTICIPATING",
-    "109": "NON_PARTICIPATING",
+    "109": "SUCCESS",
     "11": "NON_PARTICIPATING",
-    "110": "NOT_ALIVE",
+    "110": "NON_PARTICIPATING",
     "111": "NON_PARTICIPATING",
     "112": "NOT_ALIVE",
-    "113": "NON_PARTICIPATING",
+    "113": "NOT_ALIVE",
     "114": "NON_PARTICIPATING",
     "115": "NOT_ALIVE",
-    "116": "NOT_ALIVE",
-    "117": "SUCCESS",
-    "118": "NON_PARTICIPATING",
+    "116": "NON_PARTICIPATING",
+    "117": "NON_PARTICIPATING",
+    "118": "NOT_ALIVE",
     "119": "NOT_ALIVE",
     "12": "NON_PARTICIPATING",
     "120": "NOT_ALIVE",
-    "121": "NON_PARTICIPATING",
+    "121": "NOT_ALIVE",
     "122": "NON_PARTICIPATING",
     "123": "NON_PARTICIPATING",
     "124": "NON_PARTICIPATING",
-    "125": "NOT_ALIVE",
+    "125": "NON_PARTICIPATING",
     "126": "NON_PARTICIPATING",
-    "127": "NON_PARTICIPATING",
-    "128": "NOT_ALIVE",
+    "127": "NOT_ALIVE",
+    "128": "NON_PARTICIPATING",
     "129": "NON_PARTICIPATING",
-    "13": "NON_PARTICIPATING",
+    "13": "SUCCESS",
     "130": "NON_PARTICIPATING",
     "131": "NOT_ALIVE",
     "132": "NON_PARTICIPATING",
     "133": "NON_PARTICIPATING",
     "134": "NON_PARTICIPATING",
     "135": "NON_PARTICIPATING",
-    "136": "NOT_ALIVE",
+    "136": "NON_PARTICIPATING",
     "137": "NOT_ALIVE",
-    "138": "SUCCESS",
-    "139": "NOT_ALIVE",
+    "138": "NON_PARTICIPATING",
+    "139": "NON_PARTICIPATING",
     "14": "NOT_ALIVE",
     "140": "NOT_ALIVE",
     "141": "NON_PARTICIPATING",
     "142": "NON_PARTICIPATING",
-    "143": "NON_PARTICIPATING",
-    "144": "NON_PARTICIPATING",
-    "145": "NON_PARTICIPATING",
+    "143": "NOT_ALIVE",
+    "144": "SUCCESS",
+    "145": "SUCCESS",
     "146": "NON_PARTICIPATING",
-    "147": "NON_PARTICIPATING",
+    "147": "NOT_ALIVE",
     "148": "NON_PARTICIPATING",
     "149": "NON_PARTICIPATING",
-    "15": "NON_PARTICIPATING",
+    "15": "SUCCESS",
     "150": "NON_PARTICIPATING",
     "151": "NOT_ALIVE",
     "152": "NON_PARTICIPATING",
-    "153": "SUCCESS",
-    "154": "NOT_ALIVE",
+    "153": "NON_PARTICIPATING",
+    "154": "NON_PARTICIPATING",
     "155": "NON_PARTICIPATING",
     "156": "NON_PARTICIPATING",
-    "157": "NOT_ALIVE",
+    "157": "NON_PARTICIPATING",
     "158": "NOT_ALIVE",
     "159": "NON_PARTICIPATING",
-    "16": "NON_PARTICIPATING",
+    "16": "NOT_ALIVE",
     "160": "NON_PARTICIPATING",
     "161": "NON_PARTICIPATING",
-    "162": "NON_PARTICIPATING",
-    "163": "NOT_ALIVE",
+    "162": "SUCCESS",
+    "163": "NON_PARTICIPATING",
     "164": "NOT_ALIVE",
-    "165": "SUCCESS",
-    "166": "NON_PARTICIPATING",
+    "165": "NON_PARTICIPATING",
+    "166": "NOT_ALIVE",
     "167": "NON_PARTICIPATING",
-    "168": "NOT_ALIVE",
-    "169": "NOT_ALIVE",
+    "168": "NON_PARTICIPATING",
+    "169": "NON_PARTICIPATING",
     "17": "NON_PARTICIPATING",
     "170": "NON_PARTICIPATING",
     "171": "NOT_ALIVE",
-    "172": "NON_PARTICIPATING",
+    "172": "NOT_ALIVE",
     "173": "NON_PARTICIPATING",
-    "174": "NON_PARTICIPATING",
+    "174": "NOT_ALIVE",
     "175": "NOT_ALIVE",
     "176": "NOT_ALIVE",
     "177": "NON_PARTICIPATING",
-    "178": "NON_PARTICIPATING",
+    "178": "NOT_ALIVE",
     "179": "NON_PARTICIPATING",
     "18": "NON_PARTICIPATING",
-    "180": "NON_PARTICIPATING",
+    "180": "NOT_ALIVE",
     "181": "SUCCESS",
-    "182": "NON_PARTICIPATING",
+    "182": "NOT_ALIVE",
     "183": "NON_PARTICIPATING",
     "184": "NOT_ALIVE",
     "185": "NON_PARTICIPATING",
@@ -106,123 +106,123 @@
     "190": "NON_PARTICIPATING",
     "191": "NOT_ALIVE",
     "192": "NON_PARTICIPATING",
-    "193": "NOT_ALIVE",
+    "193": "NON_PARTICIPATING",
     "194": "NOT_ALIVE",
     "195": "NOT_ALIVE",
-    "196": "NOT_ALIVE",
+    "196": "NON_PARTICIPATING",
     "197": "NON_PARTICIPATING",
     "198": "NOT_ALIVE",
-    "199": "NOT_ALIVE",
-    "2": "NON_PARTICIPATING",
+    "199": "NON_PARTICIPATING",
+    "2": "NOT_ALIVE",
     "20": "NOT_ALIVE",
-    "200": "NON_PARTICIPATING",
+    "200": "SUCCESS",
     "201": "NON_PARTICIPATING",
-    "202": "NON_PARTICIPATING",
+    "202": "NOT_ALIVE",
     "203": "NON_PARTICIPATING",
     "204": "NON_PARTICIPATING",
     "205": "NOT_ALIVE",
     "206": "NON_PARTICIPATING",
     "207": "NOT_ALIVE",
-    "208": "NON_PARTICIPATING",
+    "208": "NOT_ALIVE",
     "209": "NON_PARTICIPATING",
-    "21": "SUCCESS",
-    "210": "NON_PARTICIPATING",
+    "21": "NON_PARTICIPATING",
+    "210": "NOT_ALIVE",
     "211": "NOT_ALIVE",
     "212": "NOT_ALIVE",
-    "213": "SUCCESS",
+    "213": "NON_PARTICIPATING",
     "214": "NON_PARTICIPATING",
     "215": "NOT_ALIVE",
     "216": "NOT_ALIVE",
     "217": "NON_PARTICIPATING",
-    "218": "NOT_ALIVE",
+    "218": "NON_PARTICIPATING",
     "219": "NON_PARTICIPATING",
     "22": "NON_PARTICIPATING",
     "220": "NON_PARTICIPATING",
     "221": "NOT_ALIVE",
-    "222": "NON_PARTICIPATING",
-    "223": "NOT_ALIVE",
+    "222": "SUCCESS",
+    "223": "NON_PARTICIPATING",
     "224": "NOT_ALIVE",
     "225": "NON_PARTICIPATING",
     "226": "NOT_ALIVE",
     "227": "NON_PARTICIPATING",
     "228": "NON_PARTICIPATING",
     "229": "NOT_ALIVE",
-    "23": "NON_PARTICIPATING",
+    "23": "SUCCESS",
     "230": "NOT_ALIVE",
     "231": "NON_PARTICIPATING",
     "232": "NOT_ALIVE",
     "233": "NON_PARTICIPATING",
-    "234": "NON_PARTICIPATING",
+    "234": "SUCCESS",
     "235": "NOT_ALIVE",
-    "236": "NOT_ALIVE",
+    "236": "NON_PARTICIPATING",
     "237": "NON_PARTICIPATING",
-    "238": "SUCCESS",
+    "238": "NON_PARTICIPATING",
     "239": "NON_PARTICIPATING",
-    "24": "NOT_ALIVE",
+    "24": "NON_PARTICIPATING",
     "240": "NON_PARTICIPATING",
     "241": "NON_PARTICIPATING",
-    "242": "NOT_ALIVE",
+    "242": "NON_PARTICIPATING",
     "243": "NON_PARTICIPATING",
-    "244": "NON_PARTICIPATING",
-    "245": "NON_PARTICIPATING",
+    "244": "NOT_ALIVE",
+    "245": "NOT_ALIVE",
     "246": "NON_PARTICIPATING",
-    "247": "NON_PARTICIPATING",
+    "247": "SUCCESS",
     "248": "NON_PARTICIPATING",
-    "249": "NON_PARTICIPATING",
+    "249": "NOT_ALIVE",
     "25": "SUCCESS",
     "250": "NON_PARTICIPATING",
     "251": "NON_PARTICIPATING",
-    "252": "NON_PARTICIPATING",
-    "253": "NON_PARTICIPATING",
-    "254": "NOT_ALIVE",
+    "252": "SUCCESS",
+    "253": "NOT_ALIVE",
+    "254": "NON_PARTICIPATING",
     "255": "NON_PARTICIPATING",
-    "26": "NOT_ALIVE",
+    "26": "NON_PARTICIPATING",
     "27": "NON_PARTICIPATING",
     "28": "NON_PARTICIPATING",
-    "29": "NON_PARTICIPATING",
-    "3": "NON_PARTICIPATING",
-    "30": "SUCCESS",
+    "29": "NOT_ALIVE",
+    "3": "NOT_ALIVE",
+    "30": "NOT_ALIVE",
     "31": "NON_PARTICIPATING",
     "32": "NON_PARTICIPATING",
     "33": "NOT_ALIVE",
-    "34": "NOT_ALIVE",
-    "35": "SUCCESS",
+    "34": "NON_PARTICIPATING",
+    "35": "NON_PARTICIPATING",
     "36": "NON_PARTICIPATING",
-    "37": "NON_PARTICIPATING",
-    "38": "SUCCESS",
-    "39": "SUCCESS",
+    "37": "NOT_ALIVE",
+    "38": "NON_PARTICIPATING",
+    "39": "NON_PARTICIPATING",
     "4": "NOT_ALIVE",
-    "40": "SUCCESS",
-    "41": "NOT_ALIVE",
+    "40": "NON_PARTICIPATING",
+    "41": "NON_PARTICIPATING",
     "42": "NON_PARTICIPATING",
-    "43": "NOT_ALIVE",
+    "43": "NON_PARTICIPATING",
     "44": "NON_PARTICIPATING",
-    "45": "NOT_ALIVE",
-    "46": "NON_PARTICIPATING",
+    "45": "NON_PARTICIPATING",
+    "46": "SUCCESS",
     "47": "NON_PARTICIPATING",
     "48": "NON_PARTICIPATING",
     "49": "NON_PARTICIPATING",
-    "5": "NOT_ALIVE",
-    "50": "NON_PARTICIPATING",
-    "51": "NON_PARTICIPATING",
-    "52": "NOT_ALIVE",
+    "5": "NON_PARTICIPATING",
+    "50": "SUCCESS",
+    "51": "SUCCESS",
+    "52": "NON_PARTICIPATING",
     "53": "NOT_ALIVE",
-    "54": "NON_PARTICIPATING",
+    "54": "NOT_ALIVE",
     "55": "NOT_ALIVE",
     "56": "NON_PARTICIPATING",
     "57": "NON_PARTICIPATING",
-    "58": "NON_PARTICIPATING",
-    "59": "SUCCESS",
+    "58": "NOT_ALIVE",
+    "59": "NON_PARTICIPATING",
     "6": "NON_PARTICIPATING",
-    "60": "NON_PARTICIPATING",
+    "60": "SUCCESS",
     "61": "NOT_ALIVE",
     "62": "NOT_ALIVE",
-    "63": "NOT_ALIVE",
+    "63": "NON_PARTICIPATING",
     "64": "NON_PARTICIPATING",
     "65": "NON_PARTICIPATING",
     "66": "NON_PARTICIPATING",
-    "67": "NOT_ALIVE",
-    "68": "NOT_ALIVE",
+    "67": "SUCCESS",
+    "68": "NON_PARTICIPATING",
     "69": "NON_PARTICIPATING",
     "7": "NON_PARTICIPATING",
     "70": "NOT_ALIVE",
@@ -231,7 +231,7 @@
     "73": "NON_PARTICIPATING",
     "74": "NON_PARTICIPATING",
     "75": "NON_PARTICIPATING",
-    "76": "NON_PARTICIPATING",
+    "76": "SUCCESS",
     "77": "NOT_ALIVE",
     "78": "NON_PARTICIPATING",
     "79": "NON_PARTICIPATING",
@@ -239,14 +239,14 @@
     "80": "NON_PARTICIPATING",
     "81": "NOT_ALIVE",
     "82": "NON_PARTICIPATING",
-    "83": "NON_PARTICIPATING",
+    "83": "NOT_ALIVE",
     "84": "NON_PARTICIPATING",
     "85": "NOT_ALIVE",
     "86": "NON_PARTICIPATING",
-    "87": "NOT_ALIVE",
+    "87": "NON_PARTICIPATING",
     "88": "NON_PARTICIPATING",
     "89": "NOT_ALIVE",
-    "9": "NON_PARTICIPATING",
+    "9": "NOT_ALIVE",
     "90": "NON_PARTICIPATING",
     "91": "NON_PARTICIPATING",
     "92": "NON_PARTICIPATING",
@@ -256,7 +256,7 @@
     "96": "NON_PARTICIPATING",
     "97": "NON_PARTICIPATING",
     "98": "NON_PARTICIPATING",
-    "99": "NON_PARTICIPATING"
+    "99": "NOT_ALIVE"
   },
   "architectures": [
     "LlamaForCausalLM"
@@ -264,17 +264,17 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    5918455,
-    5918476
+    5928893,
+    5928911
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 23,
+  "inner_step": 1,
   "intermediate_size": 5632,
-  "last_allreduce_block": 5917724,
+  "last_allreduce_block": 5927546,
   "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 sha256:4cfaac9d061404badef6fd6d07ce36e6877165140981be954b7bbb86a908ad44
+oid sha256:6b378ed4d30821437b3a571584c02ec8f5ba7de3fde0359bc8e7e1a049785e0c
 size 8800564076

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ff7edf097913b1ee6280a349d0d55d30a835e8d6a8624813fa7dce4b7b5eee73
+oid sha256:d25083688730c3dbe7904940ee63f2545ced313a10c4e8119bbfb22d6d853c15
 size 4400216536