Epoch 1665. Batch Size 49732. Peers 99.

config.json

@@ -3,30 +3,30 @@
   "activation_function": "gelu_new",
   "all_reduce_scores": {
     "0": "NON_PARTICIPATING",
-    "1": "NON_PARTICIPATING",
+    "1": "SUCCESS",
     "10": "SUCCESS",
     "100": "NON_PARTICIPATING",
     "101": "SUCCESS",
-    "102": "SUCCESS",
+    "102": "NON_PARTICIPATING",
     "103": "NON_PARTICIPATING",
-    "104": "SUCCESS",
-    "105": "SUCCESS",
-    "106": "SUCCESS",
-    "107": "NON_PARTICIPATING",
-    "108": "NON_PARTICIPATING",
+    "104": "NON_PARTICIPATING",
+    "105": "NON_PARTICIPATING",
+    "106": "NON_PARTICIPATING",
+    "107": "SUCCESS",
+    "108": "SUCCESS",
     "109": "NON_PARTICIPATING",
-    "11": "NON_PARTICIPATING",
-    "110": "SUCCESS",
+    "11": "SUCCESS",
+    "110": "NON_PARTICIPATING",
     "111": "NON_PARTICIPATING",
-    "112": "NON_PARTICIPATING",
+    "112": "SUCCESS",
     "113": "SUCCESS",
-    "114": "SUCCESS",
+    "114": "NON_PARTICIPATING",
     "115": "SUCCESS",
-    "116": "SUCCESS",
+    "116": "NON_PARTICIPATING",
     "117": "NON_PARTICIPATING",
-    "118": "NON_PARTICIPATING",
+    "118": "SUCCESS",
     "119": "NON_PARTICIPATING",
-    "12": "NON_PARTICIPATING",
+    "12": "SUCCESS",
     "120": "NON_PARTICIPATING",
     "121": "SUCCESS",
     "122": "SUCCESS",
@@ -34,111 +34,111 @@
     "124": "NON_PARTICIPATING",
     "125": "NON_PARTICIPATING",
     "126": "NON_PARTICIPATING",
-    "127": "NON_PARTICIPATING",
-    "128": "NON_PARTICIPATING",
+    "127": "SUCCESS",
+    "128": "SUCCESS",
     "129": "SUCCESS",
-    "13": "SUCCESS",
+    "13": "NON_PARTICIPATING",
     "130": "NON_PARTICIPATING",
     "131": "NON_PARTICIPATING",
-    "132": "SUCCESS",
+    "132": "NON_PARTICIPATING",
     "133": "NON_PARTICIPATING",
     "134": "NON_PARTICIPATING",
     "135": "NON_PARTICIPATING",
-    "136": "NON_PARTICIPATING",
+    "136": "SUCCESS",
     "137": "SUCCESS",
-    "138": "SUCCESS",
-    "139": "NON_PARTICIPATING",
-    "14": "NON_PARTICIPATING",
+    "138": "NON_PARTICIPATING",
+    "139": "SUCCESS",
+    "14": "SUCCESS",
     "140": "SUCCESS",
     "141": "NON_PARTICIPATING",
     "142": "NON_PARTICIPATING",
     "143": "NON_PARTICIPATING",
     "144": "SUCCESS",
     "145": "NON_PARTICIPATING",
-    "146": "SUCCESS",
-    "147": "NON_PARTICIPATING",
+    "146": "NON_PARTICIPATING",
+    "147": "SUCCESS",
     "148": "NON_PARTICIPATING",
-    "149": "SUCCESS",
+    "149": "NON_PARTICIPATING",
     "15": "NON_PARTICIPATING",
     "150": "NON_PARTICIPATING",
     "151": "NON_PARTICIPATING",
-    "152": "SUCCESS",
-    "153": "SUCCESS",
+    "152": "NON_PARTICIPATING",
+    "153": "NON_PARTICIPATING",
     "154": "NON_PARTICIPATING",
     "155": "NON_PARTICIPATING",
     "156": "NON_PARTICIPATING",
-    "157": "NON_PARTICIPATING",
-    "158": "SUCCESS",
+    "157": "SUCCESS",
+    "158": "NON_PARTICIPATING",
     "159": "SUCCESS",
     "16": "SUCCESS",
     "160": "NON_PARTICIPATING",
-    "161": "SUCCESS",
+    "161": "NON_PARTICIPATING",
     "162": "NON_PARTICIPATING",
-    "163": "SUCCESS",
+    "163": "NON_PARTICIPATING",
     "164": "SUCCESS",
     "165": "NON_PARTICIPATING",
-    "166": "NON_PARTICIPATING",
-    "167": "SUCCESS",
+    "166": "SUCCESS",
+    "167": "NON_PARTICIPATING",
     "168": "NON_PARTICIPATING",
     "169": "NON_PARTICIPATING",
-    "17": "SUCCESS",
+    "17": "NON_PARTICIPATING",
     "170": "NON_PARTICIPATING",
     "171": "NON_PARTICIPATING",
     "172": "SUCCESS",
-    "173": "SUCCESS",
-    "174": "NON_PARTICIPATING",
-    "175": "NON_PARTICIPATING",
+    "173": "NON_PARTICIPATING",
+    "174": "SUCCESS",
+    "175": "SUCCESS",
     "176": "NON_PARTICIPATING",
     "177": "SUCCESS",
     "178": "NON_PARTICIPATING",
     "179": "NON_PARTICIPATING",
-    "18": "SUCCESS",
+    "18": "NON_PARTICIPATING",
     "180": "SUCCESS",
     "181": "SUCCESS",
     "182": "SUCCESS",
-    "183": "SUCCESS",
+    "183": "NON_PARTICIPATING",
     "184": "NON_PARTICIPATING",
-    "185": "SUCCESS",
-    "186": "SUCCESS",
-    "187": "NON_PARTICIPATING",
+    "185": "NON_PARTICIPATING",
+    "186": "NON_PARTICIPATING",
+    "187": "SUCCESS",
     "188": "NON_PARTICIPATING",
     "189": "SUCCESS",
-    "19": "NON_PARTICIPATING",
+    "19": "SUCCESS",
     "190": "NON_PARTICIPATING",
     "191": "SUCCESS",
     "192": "NON_PARTICIPATING",
-    "193": "SUCCESS",
-    "194": "SUCCESS",
-    "195": "NON_PARTICIPATING",
-    "196": "NON_PARTICIPATING",
+    "193": "NON_PARTICIPATING",
+    "194": "NON_PARTICIPATING",
+    "195": "SUCCESS",
+    "196": "SUCCESS",
     "197": "SUCCESS",
     "198": "SUCCESS",
-    "199": "NON_PARTICIPATING",
+    "199": "SUCCESS",
     "2": "NON_PARTICIPATING",
-    "20": "SUCCESS",
-    "200": "NON_PARTICIPATING",
+    "20": "NON_PARTICIPATING",
+    "200": "SUCCESS",
     "201": "NON_PARTICIPATING",
     "202": "NON_PARTICIPATING",
     "203": "SUCCESS",
     "204": "NON_PARTICIPATING",
-    "205": "SUCCESS",
-    "206": "SUCCESS",
+    "205": "NON_PARTICIPATING",
+    "206": "NON_PARTICIPATING",
     "207": "NON_PARTICIPATING",
-    "208": "NON_PARTICIPATING",
+    "208": "SUCCESS",
     "209": "NON_PARTICIPATING",
     "21": "NON_PARTICIPATING",
     "210": "NON_PARTICIPATING",
     "211": "NON_PARTICIPATING",
     "212": "SUCCESS",
     "213": "SUCCESS",
-    "214": "NON_PARTICIPATING",
-    "215": "NON_PARTICIPATING",
+    "214": "SUCCESS",
+    "215": "SUCCESS",
     "216": "NON_PARTICIPATING",
-    "217": "NON_PARTICIPATING",
+    "217": "SUCCESS",
     "218": "NON_PARTICIPATING",
-    "219": "NON_PARTICIPATING",
+    "219": "SUCCESS",
     "22": "SUCCESS",
-    "220": "NON_PARTICIPATING",
+    "220": "SUCCESS",
     "221": "NON_PARTICIPATING",
     "222": "NON_PARTICIPATING",
     "223": "NON_PARTICIPATING",
@@ -149,113 +149,113 @@
     "228": "NON_PARTICIPATING",
     "229": "NON_PARTICIPATING",
     "23": "NON_PARTICIPATING",
-    "230": "SUCCESS",
+    "230": "NON_PARTICIPATING",
     "231": "SUCCESS",
-    "232": "SUCCESS",
+    "232": "NON_PARTICIPATING",
     "233": "NON_PARTICIPATING",
-    "234": "NON_PARTICIPATING",
+    "234": "SUCCESS",
     "235": "SUCCESS",
     "236": "SUCCESS",
-    "237": "NON_PARTICIPATING",
-    "238": "NON_PARTICIPATING",
-    "239": "SUCCESS",
+    "237": "SUCCESS",
+    "238": "SUCCESS",
+    "239": "NON_PARTICIPATING",
     "24": "NON_PARTICIPATING",
-    "240": "NON_PARTICIPATING",
+    "240": "SUCCESS",
     "241": "NON_PARTICIPATING",
     "242": "NON_PARTICIPATING",
-    "243": "SUCCESS",
+    "243": "NON_PARTICIPATING",
     "244": "NON_PARTICIPATING",
-    "245": "SUCCESS",
-    "246": "SUCCESS",
-    "247": "NON_PARTICIPATING",
+    "245": "NON_PARTICIPATING",
+    "246": "NON_PARTICIPATING",
+    "247": "SUCCESS",
     "248": "NON_PARTICIPATING",
-    "249": "SUCCESS",
+    "249": "NON_PARTICIPATING",
     "25": "SUCCESS",
-    "250": "SUCCESS",
+    "250": "NON_PARTICIPATING",
     "251": "NON_PARTICIPATING",
     "252": "NON_PARTICIPATING",
     "253": "NON_PARTICIPATING",
     "254": "SUCCESS",
-    "255": "NON_PARTICIPATING",
+    "255": "SUCCESS",
     "26": "NON_PARTICIPATING",
     "27": "NON_PARTICIPATING",
     "28": "SUCCESS",
     "29": "SUCCESS",
-    "3": "SUCCESS",
+    "3": "NON_PARTICIPATING",
     "30": "NON_PARTICIPATING",
     "31": "SUCCESS",
-    "32": "NON_PARTICIPATING",
+    "32": "SUCCESS",
     "33": "NON_PARTICIPATING",
     "34": "NON_PARTICIPATING",
     "35": "NON_PARTICIPATING",
     "36": "NON_PARTICIPATING",
-    "37": "SUCCESS",
+    "37": "NON_PARTICIPATING",
     "38": "NON_PARTICIPATING",
-    "39": "NON_PARTICIPATING",
+    "39": "SUCCESS",
     "4": "NON_PARTICIPATING",
     "40": "NON_PARTICIPATING",
     "41": "NON_PARTICIPATING",
-    "42": "SUCCESS",
+    "42": "NON_PARTICIPATING",
     "43": "NON_PARTICIPATING",
     "44": "NON_PARTICIPATING",
     "45": "NON_PARTICIPATING",
-    "46": "SUCCESS",
-    "47": "SUCCESS",
+    "46": "NON_PARTICIPATING",
+    "47": "NON_PARTICIPATING",
     "48": "NON_PARTICIPATING",
-    "49": "SUCCESS",
-    "5": "NON_PARTICIPATING",
+    "49": "NON_PARTICIPATING",
+    "5": "SUCCESS",
     "50": "NON_PARTICIPATING",
-    "51": "SUCCESS",
-    "52": "SUCCESS",
+    "51": "NON_PARTICIPATING",
+    "52": "NON_PARTICIPATING",
     "53": "NON_PARTICIPATING",
     "54": "NON_PARTICIPATING",
     "55": "NON_PARTICIPATING",
     "56": "SUCCESS",
     "57": "SUCCESS",
     "58": "SUCCESS",
-    "59": "NON_PARTICIPATING",
+    "59": "SUCCESS",
     "6": "NON_PARTICIPATING",
     "60": "SUCCESS",
-    "61": "SUCCESS",
+    "61": "NON_PARTICIPATING",
     "62": "NON_PARTICIPATING",
-    "63": "NON_PARTICIPATING",
+    "63": "SUCCESS",
     "64": "SUCCESS",
     "65": "SUCCESS",
     "66": "NON_PARTICIPATING",
     "67": "NON_PARTICIPATING",
     "68": "SUCCESS",
-    "69": "NON_PARTICIPATING",
+    "69": "SUCCESS",
     "7": "SUCCESS",
     "70": "SUCCESS",
     "71": "NON_PARTICIPATING",
-    "72": "NON_PARTICIPATING",
+    "72": "SUCCESS",
     "73": "NON_PARTICIPATING",
     "74": "SUCCESS",
-    "75": "NON_PARTICIPATING",
-    "76": "NON_PARTICIPATING",
+    "75": "SUCCESS",
+    "76": "SUCCESS",
     "77": "NON_PARTICIPATING",
     "78": "NON_PARTICIPATING",
-    "79": "NON_PARTICIPATING",
-    "8": "SUCCESS",
-    "80": "SUCCESS",
+    "79": "SUCCESS",
+    "8": "NON_PARTICIPATING",
+    "80": "NON_PARTICIPATING",
     "81": "NON_PARTICIPATING",
     "82": "NON_PARTICIPATING",
-    "83": "NON_PARTICIPATING",
-    "84": "NON_PARTICIPATING",
+    "83": "SUCCESS",
+    "84": "SUCCESS",
     "85": "NON_PARTICIPATING",
-    "86": "NON_PARTICIPATING",
+    "86": "SUCCESS",
     "87": "NON_PARTICIPATING",
-    "88": "NON_PARTICIPATING",
+    "88": "SUCCESS",
     "89": "NON_PARTICIPATING",
-    "9": "NON_PARTICIPATING",
-    "90": "SUCCESS",
-    "91": "NON_PARTICIPATING",
-    "92": "NON_PARTICIPATING",
-    "93": "NON_PARTICIPATING",
+    "9": "SUCCESS",
+    "90": "NON_PARTICIPATING",
+    "91": "SUCCESS",
+    "92": "SUCCESS",
+    "93": "SUCCESS",
     "94": "NON_PARTICIPATING",
-    "95": "NON_PARTICIPATING",
+    "95": "SUCCESS",
     "96": "NON_PARTICIPATING",
-    "97": "SUCCESS",
+    "97": "NON_PARTICIPATING",
     "98": "NON_PARTICIPATING",
     "99": "NON_PARTICIPATING"
   },

model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:82fdbba708eae70899311abd353b6325b3a43909fc8bc1b60ec7cec31db6f406
+oid sha256:305c63f6584279f67c0c19e50b0b89906c7e034612ef9e5a63f993e1fa098b96
 size 4975530824

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1eb35c4213aa3fc98b7a4d111d516756dc5572c9cb815c560071a36aa403c79c
+oid sha256:a8e89e19ea83a84d822f1a7694cf91115c43a892dcff3cc7b9ab6e7a565e32d6
 size 3075942944

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c225580d013a64de1b4effe9cfdeecc8e3c3f360332a994ae13012f9516e12d6
+oid sha256:b747913cd8e3f28fada0dde00830804cfa3a17ad3b08594fa948a5ba061ada61
 size 4675755870