Epoch 1586. Batch Size 59034. Peers 101.

config.json

@@ -3,39 +3,39 @@
   "activation_function": "gelu_new",
   "all_reduce_scores": {
     "0": "NON_PARTICIPATING",
-    "1": "NON_PARTICIPATING",
+    "1": "SUCCESS",
     "10": "NON_PARTICIPATING",
     "100": "NON_PARTICIPATING",
     "101": "SUCCESS",
     "102": "NON_PARTICIPATING",
     "103": "NON_PARTICIPATING",
-    "104": "NON_PARTICIPATING",
+    "104": "SUCCESS",
     "105": "NON_PARTICIPATING",
     "106": "SUCCESS",
-    "107": "SUCCESS",
-    "108": "NON_PARTICIPATING",
+    "107": "NON_PARTICIPATING",
+    "108": "SUCCESS",
     "109": "SUCCESS",
-    "11": "NON_PARTICIPATING",
+    "11": "SUCCESS",
     "110": "NON_PARTICIPATING",
-    "111": "SUCCESS",
-    "112": "SUCCESS",
-    "113": "NON_PARTICIPATING",
-    "114": "SUCCESS",
-    "115": "SUCCESS",
-    "116": "SUCCESS",
+    "111": "NON_PARTICIPATING",
+    "112": "NON_PARTICIPATING",
+    "113": "SUCCESS",
+    "114": "NON_PARTICIPATING",
+    "115": "NON_PARTICIPATING",
+    "116": "NON_PARTICIPATING",
     "117": "NON_PARTICIPATING",
-    "118": "NON_PARTICIPATING",
+    "118": "SUCCESS",
     "119": "SUCCESS",
     "12": "NON_PARTICIPATING",
     "120": "NON_PARTICIPATING",
     "121": "NON_PARTICIPATING",
-    "122": "NON_PARTICIPATING",
-    "123": "NON_PARTICIPATING",
+    "122": "SUCCESS",
+    "123": "SUCCESS",
     "124": "SUCCESS",
-    "125": "SUCCESS",
-    "126": "NON_PARTICIPATING",
-    "127": "NON_PARTICIPATING",
-    "128": "SUCCESS",
+    "125": "NON_PARTICIPATING",
+    "126": "SUCCESS",
+    "127": "SUCCESS",
+    "128": "NON_PARTICIPATING",
     "129": "NON_PARTICIPATING",
     "13": "SUCCESS",
     "130": "NON_PARTICIPATING",
@@ -43,101 +43,101 @@
     "132": "NON_PARTICIPATING",
     "133": "SUCCESS",
     "134": "NON_PARTICIPATING",
-    "135": "NON_PARTICIPATING",
-    "136": "SUCCESS",
-    "137": "SUCCESS",
-    "138": "NON_PARTICIPATING",
+    "135": "SUCCESS",
+    "136": "NON_PARTICIPATING",
+    "137": "NON_PARTICIPATING",
+    "138": "SUCCESS",
     "139": "NON_PARTICIPATING",
-    "14": "SUCCESS",
-    "140": "NON_PARTICIPATING",
-    "141": "SUCCESS",
+    "14": "NON_PARTICIPATING",
+    "140": "SUCCESS",
+    "141": "NON_PARTICIPATING",
     "142": "SUCCESS",
     "143": "NON_PARTICIPATING",
-    "144": "NON_PARTICIPATING",
+    "144": "SUCCESS",
     "145": "SUCCESS",
-    "146": "NON_PARTICIPATING",
+    "146": "SUCCESS",
     "147": "NON_PARTICIPATING",
-    "148": "SUCCESS",
+    "148": "NON_PARTICIPATING",
     "149": "NON_PARTICIPATING",
-    "15": "SUCCESS",
+    "15": "NON_PARTICIPATING",
     "150": "NON_PARTICIPATING",
     "151": "NON_PARTICIPATING",
     "152": "NON_PARTICIPATING",
     "153": "NON_PARTICIPATING",
     "154": "SUCCESS",
     "155": "SUCCESS",
-    "156": "NON_PARTICIPATING",
-    "157": "NON_PARTICIPATING",
+    "156": "SUCCESS",
+    "157": "SUCCESS",
     "158": "NON_PARTICIPATING",
-    "159": "NON_PARTICIPATING",
+    "159": "SUCCESS",
     "16": "NON_PARTICIPATING",
     "160": "NON_PARTICIPATING",
-    "161": "NON_PARTICIPATING",
+    "161": "SUCCESS",
     "162": "NON_PARTICIPATING",
     "163": "NON_PARTICIPATING",
-    "164": "SUCCESS",
+    "164": "NON_PARTICIPATING",
     "165": "NON_PARTICIPATING",
     "166": "NON_PARTICIPATING",
-    "167": "SUCCESS",
-    "168": "SUCCESS",
-    "169": "NON_PARTICIPATING",
+    "167": "NON_PARTICIPATING",
+    "168": "NON_PARTICIPATING",
+    "169": "SUCCESS",
     "17": "NON_PARTICIPATING",
-    "170": "NON_PARTICIPATING",
-    "171": "NON_PARTICIPATING",
-    "172": "NON_PARTICIPATING",
+    "170": "SUCCESS",
+    "171": "SUCCESS",
+    "172": "SUCCESS",
     "173": "NON_PARTICIPATING",
-    "174": "SUCCESS",
-    "175": "SUCCESS",
+    "174": "NON_PARTICIPATING",
+    "175": "NON_PARTICIPATING",
     "176": "NON_PARTICIPATING",
-    "177": "NON_PARTICIPATING",
+    "177": "SUCCESS",
     "178": "SUCCESS",
-    "179": "SUCCESS",
-    "18": "NON_PARTICIPATING",
+    "179": "NON_PARTICIPATING",
+    "18": "SUCCESS",
     "180": "SUCCESS",
-    "181": "SUCCESS",
-    "182": "SUCCESS",
+    "181": "NON_PARTICIPATING",
+    "182": "NON_PARTICIPATING",
     "183": "NON_PARTICIPATING",
     "184": "NON_PARTICIPATING",
-    "185": "SUCCESS",
+    "185": "NON_PARTICIPATING",
     "186": "NON_PARTICIPATING",
     "187": "NON_PARTICIPATING",
     "188": "NON_PARTICIPATING",
     "189": "SUCCESS",
-    "19": "NON_PARTICIPATING",
-    "190": "SUCCESS",
+    "19": "SUCCESS",
+    "190": "NON_PARTICIPATING",
     "191": "NON_PARTICIPATING",
     "192": "SUCCESS",
-    "193": "SUCCESS",
+    "193": "NON_PARTICIPATING",
     "194": "SUCCESS",
-    "195": "NON_PARTICIPATING",
-    "196": "NON_PARTICIPATING",
-    "197": "SUCCESS",
+    "195": "SUCCESS",
+    "196": "SUCCESS",
+    "197": "NON_PARTICIPATING",
     "198": "SUCCESS",
     "199": "NON_PARTICIPATING",
-    "2": "NON_PARTICIPATING",
+    "2": "SUCCESS",
     "20": "NON_PARTICIPATING",
-    "200": "SUCCESS",
-    "201": "NON_PARTICIPATING",
-    "202": "SUCCESS",
+    "200": "NON_PARTICIPATING",
+    "201": "SUCCESS",
+    "202": "NON_PARTICIPATING",
     "203": "NON_PARTICIPATING",
     "204": "NON_PARTICIPATING",
-    "205": "SUCCESS",
-    "206": "SUCCESS",
+    "205": "NON_PARTICIPATING",
+    "206": "NON_PARTICIPATING",
     "207": "NON_PARTICIPATING",
     "208": "NON_PARTICIPATING",
-    "209": "NON_PARTICIPATING",
+    "209": "SUCCESS",
     "21": "NON_PARTICIPATING",
     "210": "NON_PARTICIPATING",
-    "211": "NON_PARTICIPATING",
-    "212": "SUCCESS",
-    "213": "SUCCESS",
+    "211": "SUCCESS",
+    "212": "NON_PARTICIPATING",
+    "213": "NON_PARTICIPATING",
     "214": "NON_PARTICIPATING",
     "215": "NON_PARTICIPATING",
-    "216": "SUCCESS",
+    "216": "NON_PARTICIPATING",
     "217": "NON_PARTICIPATING",
     "218": "SUCCESS",
-    "219": "SUCCESS",
-    "22": "NON_PARTICIPATING",
+    "219": "NON_PARTICIPATING",
+    "22": "SUCCESS",
     "220": "NON_PARTICIPATING",
     "221": "SUCCESS",
     "222": "NON_PARTICIPATING",
@@ -145,118 +145,118 @@
     "224": "NON_PARTICIPATING",
     "225": "NON_PARTICIPATING",
     "226": "NON_PARTICIPATING",
-    "227": "NON_PARTICIPATING",
-    "228": "SUCCESS",
-    "229": "NON_PARTICIPATING",
+    "227": "SUCCESS",
+    "228": "NON_PARTICIPATING",
+    "229": "SUCCESS",
     "23": "NON_PARTICIPATING",
     "230": "NON_PARTICIPATING",
     "231": "NON_PARTICIPATING",
-    "232": "SUCCESS",
-    "233": "NON_PARTICIPATING",
+    "232": "NON_PARTICIPATING",
+    "233": "SUCCESS",
     "234": "NON_PARTICIPATING",
-    "235": "NON_PARTICIPATING",
-    "236": "SUCCESS",
-    "237": "NON_PARTICIPATING",
-    "238": "SUCCESS",
-    "239": "SUCCESS",
+    "235": "SUCCESS",
+    "236": "NON_PARTICIPATING",
+    "237": "SUCCESS",
+    "238": "NON_PARTICIPATING",
+    "239": "NON_PARTICIPATING",
     "24": "SUCCESS",
     "240": "NON_PARTICIPATING",
-    "241": "SUCCESS",
-    "242": "SUCCESS",
+    "241": "NON_PARTICIPATING",
+    "242": "NON_PARTICIPATING",
     "243": "SUCCESS",
     "244": "NON_PARTICIPATING",
-    "245": "SUCCESS",
+    "245": "NON_PARTICIPATING",
     "246": "SUCCESS",
     "247": "NON_PARTICIPATING",
     "248": "SUCCESS",
     "249": "NON_PARTICIPATING",
     "25": "SUCCESS",
     "250": "SUCCESS",
-    "251": "SUCCESS",
-    "252": "NON_PARTICIPATING",
+    "251": "NON_PARTICIPATING",
+    "252": "SUCCESS",
     "253": "SUCCESS",
-    "254": "SUCCESS",
+    "254": "NON_PARTICIPATING",
     "255": "NON_PARTICIPATING",
     "26": "NON_PARTICIPATING",
-    "27": "NON_PARTICIPATING",
+    "27": "SUCCESS",
     "28": "NON_PARTICIPATING",
-    "29": "SUCCESS",
-    "3": "NON_PARTICIPATING",
-    "30": "NON_PARTICIPATING",
+    "29": "NON_PARTICIPATING",
+    "3": "SUCCESS",
+    "30": "SUCCESS",
     "31": "NON_PARTICIPATING",
     "32": "NON_PARTICIPATING",
     "33": "NON_PARTICIPATING",
     "34": "SUCCESS",
     "35": "NON_PARTICIPATING",
     "36": "NON_PARTICIPATING",
-    "37": "NON_PARTICIPATING",
+    "37": "SUCCESS",
     "38": "NON_PARTICIPATING",
     "39": "SUCCESS",
     "4": "NON_PARTICIPATING",
     "40": "NON_PARTICIPATING",
-    "41": "NON_PARTICIPATING",
-    "42": "NON_PARTICIPATING",
+    "41": "SUCCESS",
+    "42": "SUCCESS",
     "43": "NON_PARTICIPATING",
-    "44": "SUCCESS",
-    "45": "NON_PARTICIPATING",
+    "44": "NON_PARTICIPATING",
+    "45": "SUCCESS",
     "46": "SUCCESS",
     "47": "NON_PARTICIPATING",
-    "48": "SUCCESS",
-    "49": "NON_PARTICIPATING",
-    "5": "NON_PARTICIPATING",
-    "50": "NON_PARTICIPATING",
+    "48": "NON_PARTICIPATING",
+    "49": "SUCCESS",
+    "5": "SUCCESS",
+    "50": "SUCCESS",
     "51": "SUCCESS",
-    "52": "NON_PARTICIPATING",
-    "53": "NON_PARTICIPATING",
-    "54": "SUCCESS",
+    "52": "SUCCESS",
+    "53": "SUCCESS",
+    "54": "NON_PARTICIPATING",
     "55": "NON_PARTICIPATING",
     "56": "SUCCESS",
     "57": "NON_PARTICIPATING",
     "58": "NON_PARTICIPATING",
-    "59": "SUCCESS",
-    "6": "NON_PARTICIPATING",
+    "59": "NON_PARTICIPATING",
+    "6": "SUCCESS",
     "60": "SUCCESS",
-    "61": "NON_PARTICIPATING",
+    "61": "SUCCESS",
     "62": "SUCCESS",
     "63": "SUCCESS",
-    "64": "NON_PARTICIPATING",
+    "64": "SUCCESS",
     "65": "NON_PARTICIPATING",
-    "66": "SUCCESS",
-    "67": "NON_PARTICIPATING",
+    "66": "NON_PARTICIPATING",
+    "67": "SUCCESS",
     "68": "SUCCESS",
-    "69": "NON_PARTICIPATING",
+    "69": "SUCCESS",
     "7": "SUCCESS",
     "70": "NON_PARTICIPATING",
-    "71": "SUCCESS",
-    "72": "NON_PARTICIPATING",
+    "71": "NON_PARTICIPATING",
+    "72": "SUCCESS",
     "73": "NON_PARTICIPATING",
-    "74": "SUCCESS",
-    "75": "SUCCESS",
-    "76": "SUCCESS",
-    "77": "SUCCESS",
+    "74": "NON_PARTICIPATING",
+    "75": "NON_PARTICIPATING",
+    "76": "NON_PARTICIPATING",
+    "77": "NON_PARTICIPATING",
     "78": "NON_PARTICIPATING",
-    "79": "SUCCESS",
+    "79": "NON_PARTICIPATING",
     "8": "SUCCESS",
     "80": "NON_PARTICIPATING",
     "81": "NON_PARTICIPATING",
-    "82": "NON_PARTICIPATING",
+    "82": "SUCCESS",
     "83": "SUCCESS",
     "84": "SUCCESS",
     "85": "NON_PARTICIPATING",
-    "86": "NON_PARTICIPATING",
+    "86": "SUCCESS",
     "87": "NON_PARTICIPATING",
     "88": "NON_PARTICIPATING",
-    "89": "NON_PARTICIPATING",
-    "9": "SUCCESS",
+    "89": "SUCCESS",
+    "9": "NON_PARTICIPATING",
     "90": "NON_PARTICIPATING",
-    "91": "SUCCESS",
+    "91": "NON_PARTICIPATING",
     "92": "NON_PARTICIPATING",
     "93": "NON_PARTICIPATING",
     "94": "NON_PARTICIPATING",
     "95": "NON_PARTICIPATING",
     "96": "NON_PARTICIPATING",
     "97": "NON_PARTICIPATING",
-    "98": "SUCCESS",
+    "98": "NON_PARTICIPATING",
     "99": "SUCCESS"
   },
   "architectures": [

model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4959093bd096caa23fcc3b5035a0c7c0d1b86a8c43b59fd378f5a67933dac9df
+oid sha256:e1dda5d704f63b0def3d56a3921351ac9733a7bda33da9c2418e174be93b897f
 size 4975530824

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:efba35e4b3bc41b847f14c0b0b8c57f2efe2bd58c84d91badd35a4dcae0e3d13
+oid sha256:8facc461976b1d0fe1ee0edf09b9f48a91493309b549c368b0c488fa22dfa95e
 size 3075942944

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a2b06fbbc1fe6c6a6e86cfa961531a4b18a7a1abb0a78cf04fdf10137d9c208a
+oid sha256:93e7b96643f3081fa9f0f985e8537c02288793deb69e31cff0291b63b1752bf3
 size 4675755870