Epoch 1285. Batch Size 78290. Peers 98.

config.json

@@ -3,7 +3,7 @@
   "activation_function": "gelu_new",
   "all_reduce_scores": {
     "0": "NON_PARTICIPATING",
-    "1": "NON_PARTICIPATING",
+    "1": "SUCCESS",
     "10": "NON_PARTICIPATING",
     "100": "NON_PARTICIPATING",
     "101": "NON_PARTICIPATING",
@@ -11,205 +11,205 @@
     "103": "NON_PARTICIPATING",
     "104": "NON_PARTICIPATING",
     "105": "NON_PARTICIPATING",
-    "106": "NON_PARTICIPATING",
+    "106": "SUCCESS",
     "107": "NON_PARTICIPATING",
     "108": "NON_PARTICIPATING",
-    "109": "NON_PARTICIPATING",
+    "109": "SUCCESS",
     "11": "NON_PARTICIPATING",
-    "110": "NON_PARTICIPATING",
+    "110": "SUCCESS",
     "111": "NON_PARTICIPATING",
-    "112": "NON_PARTICIPATING",
+    "112": "SUCCESS",
     "113": "NON_PARTICIPATING",
-    "114": "SUCCESS",
-    "115": "NON_PARTICIPATING",
-    "116": "NON_PARTICIPATING",
-    "117": "NON_PARTICIPATING",
-    "118": "NON_PARTICIPATING",
-    "119": "NON_PARTICIPATING",
+    "114": "NON_PARTICIPATING",
+    "115": "SUCCESS",
+    "116": "SUCCESS",
+    "117": "SUCCESS",
+    "118": "SUCCESS",
+    "119": "SUCCESS",
     "12": "NON_PARTICIPATING",
-    "120": "SUCCESS",
+    "120": "NON_PARTICIPATING",
     "121": "NON_PARTICIPATING",
     "122": "NON_PARTICIPATING",
     "123": "NON_PARTICIPATING",
     "124": "NON_PARTICIPATING",
     "125": "NON_PARTICIPATING",
-    "126": "NON_PARTICIPATING",
+    "126": "SUCCESS",
     "127": "NON_PARTICIPATING",
     "128": "NON_PARTICIPATING",
     "129": "NON_PARTICIPATING",
     "13": "SUCCESS",
-    "130": "SUCCESS",
+    "130": "NON_PARTICIPATING",
     "131": "NON_PARTICIPATING",
-    "132": "NON_PARTICIPATING",
+    "132": "SUCCESS",
     "133": "NON_PARTICIPATING",
-    "134": "SUCCESS",
-    "135": "NON_PARTICIPATING",
+    "134": "NON_PARTICIPATING",
+    "135": "SUCCESS",
     "136": "SUCCESS",
     "137": "SUCCESS",
     "138": "NON_PARTICIPATING",
-    "139": "NON_PARTICIPATING",
+    "139": "SUCCESS",
     "14": "SUCCESS",
     "140": "NON_PARTICIPATING",
     "141": "NON_PARTICIPATING",
-    "142": "SUCCESS",
+    "142": "NON_PARTICIPATING",
     "143": "NON_PARTICIPATING",
     "144": "NON_PARTICIPATING",
-    "145": "NON_PARTICIPATING",
-    "146": "SUCCESS",
+    "145": "SUCCESS",
+    "146": "NON_PARTICIPATING",
     "147": "SUCCESS",
-    "148": "NON_PARTICIPATING",
-    "149": "NON_PARTICIPATING",
+    "148": "SUCCESS",
+    "149": "SUCCESS",
     "15": "NON_PARTICIPATING",
-    "150": "NON_PARTICIPATING",
+    "150": "SUCCESS",
     "151": "NON_PARTICIPATING",
     "152": "NON_PARTICIPATING",
     "153": "NON_PARTICIPATING",
-    "154": "NON_PARTICIPATING",
+    "154": "SUCCESS",
     "155": "NON_PARTICIPATING",
     "156": "NON_PARTICIPATING",
     "157": "SUCCESS",
-    "158": "SUCCESS",
+    "158": "NON_PARTICIPATING",
     "159": "NON_PARTICIPATING",
-    "16": "NON_PARTICIPATING",
+    "16": "SUCCESS",
     "160": "SUCCESS",
     "161": "NON_PARTICIPATING",
-    "162": "SUCCESS",
-    "163": "NON_PARTICIPATING",
-    "164": "NON_PARTICIPATING",
+    "162": "NON_PARTICIPATING",
+    "163": "SUCCESS",
+    "164": "SUCCESS",
     "165": "NON_PARTICIPATING",
-    "166": "NON_PARTICIPATING",
+    "166": "SUCCESS",
     "167": "NON_PARTICIPATING",
-    "168": "NON_PARTICIPATING",
-    "169": "NON_PARTICIPATING",
+    "168": "SUCCESS",
+    "169": "SUCCESS",
     "17": "NON_PARTICIPATING",
-    "170": "NON_PARTICIPATING",
+    "170": "SUCCESS",
     "171": "NON_PARTICIPATING",
-    "172": "NON_PARTICIPATING",
+    "172": "SUCCESS",
     "173": "SUCCESS",
     "174": "SUCCESS",
-    "175": "NON_PARTICIPATING",
-    "176": "SUCCESS",
+    "175": "SUCCESS",
+    "176": "NON_PARTICIPATING",
     "177": "NON_PARTICIPATING",
     "178": "SUCCESS",
-    "179": "NON_PARTICIPATING",
+    "179": "SUCCESS",
     "18": "SUCCESS",
-    "180": "SUCCESS",
-    "181": "NON_PARTICIPATING",
+    "180": "NON_PARTICIPATING",
+    "181": "SUCCESS",
     "182": "NON_PARTICIPATING",
-    "183": "SUCCESS",
-    "184": "SUCCESS",
+    "183": "NON_PARTICIPATING",
+    "184": "NON_PARTICIPATING",
     "185": "NON_PARTICIPATING",
-    "186": "NON_PARTICIPATING",
+    "186": "SUCCESS",
     "187": "SUCCESS",
     "188": "SUCCESS",
     "189": "NON_PARTICIPATING",
     "19": "NON_PARTICIPATING",
-    "190": "NON_PARTICIPATING",
+    "190": "SUCCESS",
     "191": "NON_PARTICIPATING",
     "192": "NON_PARTICIPATING",
     "193": "SUCCESS",
     "194": "NON_PARTICIPATING",
     "195": "NON_PARTICIPATING",
-    "196": "NON_PARTICIPATING",
-    "197": "SUCCESS",
+    "196": "SUCCESS",
+    "197": "NON_PARTICIPATING",
     "198": "SUCCESS",
     "199": "NON_PARTICIPATING",
-    "2": "SUCCESS",
+    "2": "NON_PARTICIPATING",
     "20": "NON_PARTICIPATING",
     "200": "NON_PARTICIPATING",
-    "201": "NON_PARTICIPATING",
+    "201": "SUCCESS",
     "202": "SUCCESS",
     "203": "NON_PARTICIPATING",
-    "204": "NON_PARTICIPATING",
-    "205": "NON_PARTICIPATING",
+    "204": "SUCCESS",
+    "205": "SUCCESS",
     "206": "SUCCESS",
     "207": "NON_PARTICIPATING",
     "208": "NON_PARTICIPATING",
-    "209": "NON_PARTICIPATING",
-    "21": "NON_PARTICIPATING",
+    "209": "SUCCESS",
+    "21": "SUCCESS",
     "210": "NON_PARTICIPATING",
-    "211": "NON_PARTICIPATING",
-    "212": "NON_PARTICIPATING",
+    "211": "SUCCESS",
+    "212": "SUCCESS",
     "213": "NON_PARTICIPATING",
     "214": "NON_PARTICIPATING",
     "215": "SUCCESS",
-    "216": "NON_PARTICIPATING",
+    "216": "SUCCESS",
     "217": "SUCCESS",
     "218": "NON_PARTICIPATING",
     "219": "NON_PARTICIPATING",
     "22": "NON_PARTICIPATING",
-    "220": "SUCCESS",
-    "221": "NON_PARTICIPATING",
+    "220": "NON_PARTICIPATING",
+    "221": "SUCCESS",
     "222": "NON_PARTICIPATING",
     "223": "NON_PARTICIPATING",
     "224": "NON_PARTICIPATING",
-    "225": "NON_PARTICIPATING",
+    "225": "SUCCESS",
     "226": "NON_PARTICIPATING",
-    "227": "NON_PARTICIPATING",
+    "227": "SUCCESS",
     "228": "NON_PARTICIPATING",
     "229": "NON_PARTICIPATING",
     "23": "NON_PARTICIPATING",
     "230": "SUCCESS",
     "231": "NON_PARTICIPATING",
-    "232": "SUCCESS",
+    "232": "NON_PARTICIPATING",
     "233": "NON_PARTICIPATING",
-    "234": "SUCCESS",
+    "234": "NON_PARTICIPATING",
     "235": "NON_PARTICIPATING",
-    "236": "NON_PARTICIPATING",
+    "236": "SUCCESS",
     "237": "NON_PARTICIPATING",
     "238": "NON_PARTICIPATING",
     "239": "NON_PARTICIPATING",
-    "24": "NON_PARTICIPATING",
+    "24": "SUCCESS",
     "240": "NON_PARTICIPATING",
     "241": "SUCCESS",
     "242": "NON_PARTICIPATING",
-    "243": "SUCCESS",
+    "243": "NON_PARTICIPATING",
     "244": "NON_PARTICIPATING",
     "245": "NON_PARTICIPATING",
     "246": "NON_PARTICIPATING",
-    "247": "SUCCESS",
+    "247": "NON_PARTICIPATING",
     "248": "NON_PARTICIPATING",
     "249": "NON_PARTICIPATING",
-    "25": "NON_PARTICIPATING",
+    "25": "SUCCESS",
     "250": "NON_PARTICIPATING",
     "251": "NON_PARTICIPATING",
     "252": "SUCCESS",
-    "253": "SUCCESS",
-    "254": "NON_PARTICIPATING",
+    "253": "NON_PARTICIPATING",
+    "254": "SUCCESS",
     "255": "NON_PARTICIPATING",
-    "26": "SUCCESS",
+    "26": "NON_PARTICIPATING",
     "27": "NON_PARTICIPATING",
     "28": "NON_PARTICIPATING",
     "29": "NON_PARTICIPATING",
     "3": "NON_PARTICIPATING",
-    "30": "SUCCESS",
+    "30": "NON_PARTICIPATING",
     "31": "NON_PARTICIPATING",
     "32": "NON_PARTICIPATING",
     "33": "NON_PARTICIPATING",
-    "34": "SUCCESS",
-    "35": "NON_PARTICIPATING",
+    "34": "NON_PARTICIPATING",
+    "35": "SUCCESS",
     "36": "SUCCESS",
     "37": "NON_PARTICIPATING",
     "38": "NON_PARTICIPATING",
     "39": "NON_PARTICIPATING",
     "4": "NON_PARTICIPATING",
     "40": "SUCCESS",
-    "41": "NON_PARTICIPATING",
+    "41": "SUCCESS",
     "42": "SUCCESS",
-    "43": "SUCCESS",
-    "44": "NON_PARTICIPATING",
-    "45": "NON_PARTICIPATING",
+    "43": "NON_PARTICIPATING",
+    "44": "SUCCESS",
+    "45": "SUCCESS",
     "46": "SUCCESS",
     "47": "SUCCESS",
     "48": "SUCCESS",
-    "49": "NON_PARTICIPATING",
+    "49": "SUCCESS",
     "5": "NON_PARTICIPATING",
-    "50": "NON_PARTICIPATING",
-    "51": "NON_PARTICIPATING",
-    "52": "NON_PARTICIPATING",
-    "53": "SUCCESS",
-    "54": "SUCCESS",
-    "55": "SUCCESS",
+    "50": "SUCCESS",
+    "51": "SUCCESS",
+    "52": "SUCCESS",
+    "53": "NON_PARTICIPATING",
+    "54": "NON_PARTICIPATING",
+    "55": "NON_PARTICIPATING",
     "56": "NON_PARTICIPATING",
     "57": "NON_PARTICIPATING",
     "58": "NON_PARTICIPATING",
@@ -219,43 +219,43 @@
     "61": "NON_PARTICIPATING",
     "62": "NON_PARTICIPATING",
     "63": "NON_PARTICIPATING",
-    "64": "NON_PARTICIPATING",
+    "64": "SUCCESS",
     "65": "SUCCESS",
     "66": "NON_PARTICIPATING",
     "67": "NON_PARTICIPATING",
     "68": "NON_PARTICIPATING",
-    "69": "SUCCESS",
+    "69": "NON_PARTICIPATING",
     "7": "NON_PARTICIPATING",
     "70": "NON_PARTICIPATING",
-    "71": "SUCCESS",
+    "71": "NON_PARTICIPATING",
     "72": "NON_PARTICIPATING",
     "73": "NON_PARTICIPATING",
-    "74": "NON_PARTICIPATING",
+    "74": "SUCCESS",
     "75": "NON_PARTICIPATING",
     "76": "NON_PARTICIPATING",
     "77": "NON_PARTICIPATING",
-    "78": "NON_PARTICIPATING",
+    "78": "SUCCESS",
     "79": "NON_PARTICIPATING",
     "8": "NON_PARTICIPATING",
-    "80": "NON_PARTICIPATING",
+    "80": "SUCCESS",
     "81": "SUCCESS",
     "82": "NON_PARTICIPATING",
     "83": "SUCCESS",
     "84": "NON_PARTICIPATING",
     "85": "NON_PARTICIPATING",
-    "86": "NON_PARTICIPATING",
+    "86": "SUCCESS",
     "87": "SUCCESS",
-    "88": "SUCCESS",
-    "89": "SUCCESS",
+    "88": "NON_PARTICIPATING",
+    "89": "NON_PARTICIPATING",
     "9": "NON_PARTICIPATING",
     "90": "NON_PARTICIPATING",
-    "91": "SUCCESS",
-    "92": "SUCCESS",
+    "91": "NON_PARTICIPATING",
+    "92": "NON_PARTICIPATING",
     "93": "NON_PARTICIPATING",
-    "94": "NON_PARTICIPATING",
+    "94": "SUCCESS",
     "95": "SUCCESS",
-    "96": "NON_PARTICIPATING",
-    "97": "SUCCESS",
+    "96": "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 sha256:9212542a2d7826d1ee8b9b432eef450fd68eb18d9c457c0d1c8c913f92418ac2
+oid sha256:031924a44ccc02783caed3e2a76a71b98bf810cf6e93911a8b27b1bb05f264ee
 size 4975530824

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8441adca0bc6d9917879fd3ac61f91d3f2cebdc08afaf81a94638474c3df1685
+oid sha256:9d436947cbe8026b548f2cfdce323f43804cb525e109ae75b02b7b8ee2f98045
 size 3075942944

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c54ae8cbb1df4596d3ac6b60ce30b463965af7a28057fd2a274e1949fb2c7526
+oid sha256:a3914409a9509ee2d1338c864f7f427c9c5eee502b617baca7488e485936e812
 size 4675755870