Epoch 1680. Batch Size 69380. Peers 101.

config.json

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

model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4668dc910c1b6e17589a5ee4050e074ca5a555aca0d9efa8dc4447062515206f
+oid sha256:4f5aa7d02724aed77be9f93588ed2ea8f0214b70b3226a254e8b6fc4c679930d
 size 4975530824

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3e585dc7d833319ed7935eac76b6732711829fe2d97af8c74b8d1ba32782df29
+oid sha256:030e671c0ba325782148eb1d27fe6b0da82dd8f2f9cf2e0dd6a858a8de265a90
 size 3075942944

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:368028938a6a8f3563f23ab7b9c66a544cda91dd6cb1ed32682bc74b0e2e770d
+oid sha256:3d48ddab2a4515dd558074405f1c1029d9777a6993c3a6648232fd0284e527ae
 size 4675755870