Epoch 608. Batch Size 16034. Peers 100.

config.json

@@ -3,20 +3,20 @@
   "activation_function": "gelu_new",
   "all_reduce_scores": {
     "0": "NON_PARTICIPATING",
-    "1": "NON_PARTICIPATING",
-    "10": "NON_PARTICIPATING",
+    "1": "SUCCESS",
+    "10": "SUCCESS",
     "100": "NON_PARTICIPATING",
-    "101": "SUCCESS",
+    "101": "NON_PARTICIPATING",
     "102": "SUCCESS",
-    "103": "NON_PARTICIPATING",
-    "104": "NON_PARTICIPATING",
+    "103": "SUCCESS",
+    "104": "SUCCESS",
     "105": "NON_PARTICIPATING",
-    "106": "NON_PARTICIPATING",
-    "107": "NON_PARTICIPATING",
+    "106": "SUCCESS",
+    "107": "SUCCESS",
     "108": "SUCCESS",
     "109": "SUCCESS",
     "11": "SUCCESS",
-    "110": "SUCCESS",
+    "110": "NON_PARTICIPATING",
     "111": "SUCCESS",
     "112": "NON_PARTICIPATING",
     "113": "NON_PARTICIPATING",
@@ -25,237 +25,237 @@
     "116": "NON_PARTICIPATING",
     "117": "NON_PARTICIPATING",
     "118": "SUCCESS",
-    "119": "NON_PARTICIPATING",
+    "119": "SUCCESS",
     "12": "NON_PARTICIPATING",
-    "120": "SUCCESS",
-    "121": "SUCCESS",
-    "122": "SUCCESS",
+    "120": "NON_PARTICIPATING",
+    "121": "NON_PARTICIPATING",
+    "122": "NON_PARTICIPATING",
     "123": "NON_PARTICIPATING",
     "124": "NON_PARTICIPATING",
-    "125": "SUCCESS",
+    "125": "NON_PARTICIPATING",
     "126": "NON_PARTICIPATING",
-    "127": "NON_PARTICIPATING",
+    "127": "SUCCESS",
     "128": "SUCCESS",
-    "129": "SUCCESS",
+    "129": "NON_PARTICIPATING",
     "13": "NON_PARTICIPATING",
     "130": "NON_PARTICIPATING",
     "131": "NON_PARTICIPATING",
-    "132": "SUCCESS",
+    "132": "NON_PARTICIPATING",
     "133": "NON_PARTICIPATING",
     "134": "NON_PARTICIPATING",
     "135": "NON_PARTICIPATING",
-    "136": "SUCCESS",
-    "137": "NON_PARTICIPATING",
+    "136": "NON_PARTICIPATING",
+    "137": "SUCCESS",
     "138": "NON_PARTICIPATING",
     "139": "NON_PARTICIPATING",
-    "14": "SUCCESS",
-    "140": "SUCCESS",
+    "14": "NON_PARTICIPATING",
+    "140": "NON_PARTICIPATING",
     "141": "NON_PARTICIPATING",
     "142": "NON_PARTICIPATING",
     "143": "NON_PARTICIPATING",
     "144": "NON_PARTICIPATING",
-    "145": "SUCCESS",
+    "145": "NON_PARTICIPATING",
     "146": "NON_PARTICIPATING",
-    "147": "NON_PARTICIPATING",
+    "147": "SUCCESS",
     "148": "NON_PARTICIPATING",
     "149": "NON_PARTICIPATING",
     "15": "NON_PARTICIPATING",
-    "150": "SUCCESS",
-    "151": "SUCCESS",
+    "150": "NON_PARTICIPATING",
+    "151": "NON_PARTICIPATING",
     "152": "SUCCESS",
     "153": "NON_PARTICIPATING",
-    "154": "SUCCESS",
+    "154": "NON_PARTICIPATING",
     "155": "NON_PARTICIPATING",
     "156": "NON_PARTICIPATING",
     "157": "NON_PARTICIPATING",
-    "158": "NON_PARTICIPATING",
+    "158": "SUCCESS",
     "159": "SUCCESS",
     "16": "NON_PARTICIPATING",
     "160": "NON_PARTICIPATING",
-    "161": "SUCCESS",
+    "161": "NON_PARTICIPATING",
     "162": "SUCCESS",
     "163": "SUCCESS",
     "164": "NON_PARTICIPATING",
     "165": "NON_PARTICIPATING",
     "166": "SUCCESS",
-    "167": "NON_PARTICIPATING",
+    "167": "SUCCESS",
     "168": "NON_PARTICIPATING",
-    "169": "SUCCESS",
-    "17": "NON_PARTICIPATING",
+    "169": "NON_PARTICIPATING",
+    "17": "SUCCESS",
     "170": "NON_PARTICIPATING",
-    "171": "SUCCESS",
+    "171": "NON_PARTICIPATING",
     "172": "NON_PARTICIPATING",
     "173": "NON_PARTICIPATING",
     "174": "SUCCESS",
-    "175": "SUCCESS",
+    "175": "NON_PARTICIPATING",
     "176": "NON_PARTICIPATING",
-    "177": "SUCCESS",
-    "178": "NON_PARTICIPATING",
+    "177": "NON_PARTICIPATING",
+    "178": "SUCCESS",
     "179": "NON_PARTICIPATING",
     "18": "SUCCESS",
-    "180": "SUCCESS",
-    "181": "SUCCESS",
-    "182": "NON_PARTICIPATING",
-    "183": "NON_PARTICIPATING",
-    "184": "NON_PARTICIPATING",
+    "180": "NON_PARTICIPATING",
+    "181": "NON_PARTICIPATING",
+    "182": "SUCCESS",
+    "183": "SUCCESS",
+    "184": "SUCCESS",
     "185": "NON_PARTICIPATING",
-    "186": "NON_PARTICIPATING",
+    "186": "SUCCESS",
     "187": "NON_PARTICIPATING",
-    "188": "NON_PARTICIPATING",
-    "189": "NON_PARTICIPATING",
-    "19": "SUCCESS",
-    "190": "SUCCESS",
-    "191": "SUCCESS",
+    "188": "SUCCESS",
+    "189": "SUCCESS",
+    "19": "NON_PARTICIPATING",
+    "190": "NON_PARTICIPATING",
+    "191": "NON_PARTICIPATING",
     "192": "NON_PARTICIPATING",
-    "193": "NON_PARTICIPATING",
+    "193": "SUCCESS",
     "194": "SUCCESS",
-    "195": "NON_PARTICIPATING",
+    "195": "SUCCESS",
     "196": "NON_PARTICIPATING",
-    "197": "NON_PARTICIPATING",
-    "198": "NON_PARTICIPATING",
-    "199": "SUCCESS",
-    "2": "NON_PARTICIPATING",
+    "197": "SUCCESS",
+    "198": "SUCCESS",
+    "199": "NON_PARTICIPATING",
+    "2": "SUCCESS",
     "20": "NON_PARTICIPATING",
     "200": "NON_PARTICIPATING",
-    "201": "NON_PARTICIPATING",
+    "201": "SUCCESS",
     "202": "SUCCESS",
-    "203": "SUCCESS",
-    "204": "NON_PARTICIPATING",
+    "203": "NON_PARTICIPATING",
+    "204": "SUCCESS",
     "205": "NON_PARTICIPATING",
     "206": "SUCCESS",
     "207": "NON_PARTICIPATING",
-    "208": "NON_PARTICIPATING",
+    "208": "SUCCESS",
     "209": "NON_PARTICIPATING",
     "21": "SUCCESS",
     "210": "NON_PARTICIPATING",
     "211": "NON_PARTICIPATING",
     "212": "NON_PARTICIPATING",
-    "213": "SUCCESS",
+    "213": "NON_PARTICIPATING",
     "214": "SUCCESS",
     "215": "NON_PARTICIPATING",
-    "216": "NON_PARTICIPATING",
-    "217": "SUCCESS",
-    "218": "NON_PARTICIPATING",
-    "219": "SUCCESS",
-    "22": "NON_PARTICIPATING",
-    "220": "SUCCESS",
-    "221": "SUCCESS",
+    "216": "SUCCESS",
+    "217": "NON_PARTICIPATING",
+    "218": "SUCCESS",
+    "219": "NON_PARTICIPATING",
+    "22": "SUCCESS",
+    "220": "NON_PARTICIPATING",
+    "221": "NON_PARTICIPATING",
     "222": "NON_PARTICIPATING",
     "223": "NON_PARTICIPATING",
-    "224": "NON_PARTICIPATING",
+    "224": "SUCCESS",
     "225": "SUCCESS",
-    "226": "SUCCESS",
-    "227": "SUCCESS",
+    "226": "NON_PARTICIPATING",
+    "227": "NON_PARTICIPATING",
     "228": "SUCCESS",
-    "229": "NON_PARTICIPATING",
+    "229": "SUCCESS",
     "23": "NON_PARTICIPATING",
-    "230": "SUCCESS",
-    "231": "SUCCESS",
-    "232": "SUCCESS",
-    "233": "NON_PARTICIPATING",
+    "230": "NON_PARTICIPATING",
+    "231": "NON_PARTICIPATING",
+    "232": "NON_PARTICIPATING",
+    "233": "SUCCESS",
     "234": "NON_PARTICIPATING",
-    "235": "NON_PARTICIPATING",
-    "236": "SUCCESS",
+    "235": "SUCCESS",
+    "236": "NON_PARTICIPATING",
     "237": "SUCCESS",
     "238": "SUCCESS",
-    "239": "NON_PARTICIPATING",
-    "24": "NON_PARTICIPATING",
-    "240": "SUCCESS",
-    "241": "NON_PARTICIPATING",
+    "239": "SUCCESS",
+    "24": "SUCCESS",
+    "240": "NON_PARTICIPATING",
+    "241": "SUCCESS",
     "242": "NON_PARTICIPATING",
-    "243": "NON_PARTICIPATING",
-    "244": "NON_PARTICIPATING",
+    "243": "SUCCESS",
+    "244": "SUCCESS",
     "245": "NON_PARTICIPATING",
     "246": "SUCCESS",
-    "247": "NON_PARTICIPATING",
+    "247": "SUCCESS",
     "248": "NON_PARTICIPATING",
     "249": "NON_PARTICIPATING",
     "25": "NON_PARTICIPATING",
     "250": "NON_PARTICIPATING",
-    "251": "SUCCESS",
+    "251": "NON_PARTICIPATING",
     "252": "NON_PARTICIPATING",
     "253": "NON_PARTICIPATING",
     "254": "NON_PARTICIPATING",
     "255": "NON_PARTICIPATING",
-    "26": "SUCCESS",
-    "27": "SUCCESS",
-    "28": "NON_PARTICIPATING",
-    "29": "SUCCESS",
-    "3": "NON_PARTICIPATING",
-    "30": "SUCCESS",
-    "31": "NON_PARTICIPATING",
-    "32": "NON_PARTICIPATING",
+    "26": "NON_PARTICIPATING",
+    "27": "NON_PARTICIPATING",
+    "28": "SUCCESS",
+    "29": "NON_PARTICIPATING",
+    "3": "SUCCESS",
+    "30": "NON_PARTICIPATING",
+    "31": "SUCCESS",
+    "32": "SUCCESS",
     "33": "NON_PARTICIPATING",
     "34": "SUCCESS",
-    "35": "SUCCESS",
-    "36": "NON_PARTICIPATING",
+    "35": "NON_PARTICIPATING",
+    "36": "SUCCESS",
     "37": "SUCCESS",
     "38": "NON_PARTICIPATING",
     "39": "NON_PARTICIPATING",
-    "4": "SUCCESS",
-    "40": "SUCCESS",
+    "4": "NON_PARTICIPATING",
+    "40": "NON_PARTICIPATING",
     "41": "SUCCESS",
-    "42": "NON_PARTICIPATING",
-    "43": "NON_PARTICIPATING",
+    "42": "SUCCESS",
+    "43": "SUCCESS",
     "44": "NON_PARTICIPATING",
-    "45": "SUCCESS",
-    "46": "SUCCESS",
+    "45": "NON_PARTICIPATING",
+    "46": "NON_PARTICIPATING",
     "47": "NON_PARTICIPATING",
     "48": "NON_PARTICIPATING",
     "49": "NON_PARTICIPATING",
     "5": "NON_PARTICIPATING",
-    "50": "SUCCESS",
+    "50": "NON_PARTICIPATING",
     "51": "NON_PARTICIPATING",
     "52": "SUCCESS",
     "53": "NON_PARTICIPATING",
     "54": "NON_PARTICIPATING",
     "55": "NON_PARTICIPATING",
     "56": "NON_PARTICIPATING",
-    "57": "SUCCESS",
+    "57": "NON_PARTICIPATING",
     "58": "SUCCESS",
-    "59": "NON_PARTICIPATING",
-    "6": "SUCCESS",
+    "59": "SUCCESS",
+    "6": "NON_PARTICIPATING",
     "60": "SUCCESS",
-    "61": "SUCCESS",
-    "62": "NON_PARTICIPATING",
-    "63": "SUCCESS",
+    "61": "NON_PARTICIPATING",
+    "62": "SUCCESS",
+    "63": "NON_PARTICIPATING",
     "64": "NON_PARTICIPATING",
     "65": "SUCCESS",
     "66": "NON_PARTICIPATING",
     "67": "NON_PARTICIPATING",
-    "68": "SUCCESS",
-    "69": "NON_PARTICIPATING",
+    "68": "NON_PARTICIPATING",
+    "69": "SUCCESS",
     "7": "SUCCESS",
-    "70": "NON_PARTICIPATING",
-    "71": "SUCCESS",
+    "70": "SUCCESS",
+    "71": "NON_PARTICIPATING",
     "72": "NON_PARTICIPATING",
-    "73": "SUCCESS",
+    "73": "NON_PARTICIPATING",
     "74": "NON_PARTICIPATING",
-    "75": "NON_PARTICIPATING",
+    "75": "SUCCESS",
     "76": "SUCCESS",
-    "77": "NON_PARTICIPATING",
+    "77": "SUCCESS",
     "78": "NON_PARTICIPATING",
-    "79": "NON_PARTICIPATING",
-    "8": "NON_PARTICIPATING",
+    "79": "SUCCESS",
+    "8": "SUCCESS",
     "80": "SUCCESS",
     "81": "NON_PARTICIPATING",
-    "82": "SUCCESS",
-    "83": "SUCCESS",
-    "84": "NON_PARTICIPATING",
+    "82": "NON_PARTICIPATING",
+    "83": "NON_PARTICIPATING",
+    "84": "SUCCESS",
     "85": "SUCCESS",
-    "86": "NON_PARTICIPATING",
-    "87": "NON_PARTICIPATING",
+    "86": "SUCCESS",
+    "87": "SUCCESS",
     "88": "NON_PARTICIPATING",
-    "89": "NON_PARTICIPATING",
-    "9": "SUCCESS",
+    "89": "SUCCESS",
+    "9": "NON_PARTICIPATING",
     "90": "NON_PARTICIPATING",
     "91": "NON_PARTICIPATING",
-    "92": "NON_PARTICIPATING",
+    "92": "SUCCESS",
     "93": "NON_PARTICIPATING",
-    "94": "NON_PARTICIPATING",
+    "94": "SUCCESS",
     "95": "NON_PARTICIPATING",
-    "96": "SUCCESS",
-    "97": "NON_PARTICIPATING",
+    "96": "NON_PARTICIPATING",
+    "97": "SUCCESS",
     "98": "NON_PARTICIPATING",
     "99": "SUCCESS"
   },

model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e4e3b80af5b40f3bf86b7acbcdc8f871b3420ccbd8e0f69ecf74a4303c753fa1
+oid sha256:f215dba7148fcfc0ae8d6adbc8edf41fa05d85c666c8e5408e77a36390cec5ba
 size 4975530824

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c2369136d0f1cb815b4034517c820ccf4394391dec11673faaf6d6ed0f381819
+oid sha256:f1d3d78bc3626d1f452784e6ba723184e2b6c9d24f79e810d0033b2323c19c42
 size 3075942944

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:41b7c6a43f09e58c3a0e28b181fed962898722a0aea5af664d43a9a5e1af44f4
+oid sha256:cf9a5b5246614e569a207bd11f90921075c02eb5e57a57e574e1aa8bf08bfe36
 size 4675755870