Epoch 851. Batch Size 35644. Peers 73.

config.json

@@ -3,110 +3,110 @@
   "activation_function": "gelu_new",
   "all_reduce_scores": {
     "0": "NON_PARTICIPATING",
-    "1": "SUCCESS",
-    "10": "SUCCESS",
+    "1": "NON_PARTICIPATING",
+    "10": "NON_PARTICIPATING",
     "100": "NON_PARTICIPATING",
     "101": "NON_PARTICIPATING",
     "102": "NON_PARTICIPATING",
-    "103": "NON_PARTICIPATING",
-    "104": "SUCCESS",
-    "105": "SUCCESS",
-    "106": "NON_PARTICIPATING",
+    "103": "SUCCESS",
+    "104": "NON_PARTICIPATING",
+    "105": "NON_PARTICIPATING",
+    "106": "SUCCESS",
     "107": "NON_PARTICIPATING",
     "108": "NON_PARTICIPATING",
     "109": "NON_PARTICIPATING",
-    "11": "NON_PARTICIPATING",
-    "110": "SUCCESS",
+    "11": "SUCCESS",
+    "110": "NON_PARTICIPATING",
     "111": "NON_PARTICIPATING",
     "112": "NON_PARTICIPATING",
     "113": "NON_PARTICIPATING",
-    "114": "SUCCESS",
-    "115": "NON_PARTICIPATING",
-    "116": "NON_PARTICIPATING",
-    "117": "NON_PARTICIPATING",
-    "118": "SUCCESS",
+    "114": "NON_PARTICIPATING",
+    "115": "SUCCESS",
+    "116": "SUCCESS",
+    "117": "SUCCESS",
+    "118": "NON_PARTICIPATING",
     "119": "SUCCESS",
-    "12": "NON_PARTICIPATING",
-    "120": "NON_PARTICIPATING",
+    "12": "SUCCESS",
+    "120": "SUCCESS",
     "121": "NON_PARTICIPATING",
     "122": "SUCCESS",
     "123": "NON_PARTICIPATING",
-    "124": "NON_PARTICIPATING",
+    "124": "SUCCESS",
     "125": "NON_PARTICIPATING",
     "126": "NON_PARTICIPATING",
-    "127": "SUCCESS",
+    "127": "NON_PARTICIPATING",
     "128": "SUCCESS",
     "129": "SUCCESS",
     "13": "NON_PARTICIPATING",
-    "130": "NON_PARTICIPATING",
+    "130": "SUCCESS",
     "131": "NON_PARTICIPATING",
     "132": "NON_PARTICIPATING",
-    "133": "SUCCESS",
+    "133": "NON_PARTICIPATING",
     "134": "NON_PARTICIPATING",
     "135": "NON_PARTICIPATING",
-    "136": "NON_PARTICIPATING",
-    "137": "SUCCESS",
+    "136": "SUCCESS",
+    "137": "NON_PARTICIPATING",
     "138": "NON_PARTICIPATING",
     "139": "NON_PARTICIPATING",
-    "14": "NON_PARTICIPATING",
+    "14": "SUCCESS",
     "140": "NON_PARTICIPATING",
     "141": "SUCCESS",
-    "142": "SUCCESS",
+    "142": "NON_PARTICIPATING",
     "143": "NON_PARTICIPATING",
     "144": "NON_PARTICIPATING",
-    "145": "SUCCESS",
+    "145": "NON_PARTICIPATING",
     "146": "NON_PARTICIPATING",
     "147": "NON_PARTICIPATING",
     "148": "NON_PARTICIPATING",
-    "149": "SUCCESS",
+    "149": "NON_PARTICIPATING",
     "15": "NON_PARTICIPATING",
-    "150": "SUCCESS",
+    "150": "NON_PARTICIPATING",
     "151": "NON_PARTICIPATING",
     "152": "SUCCESS",
     "153": "NON_PARTICIPATING",
-    "154": "NON_PARTICIPATING",
+    "154": "SUCCESS",
     "155": "NON_PARTICIPATING",
     "156": "NON_PARTICIPATING",
-    "157": "SUCCESS",
-    "158": "NON_PARTICIPATING",
+    "157": "NON_PARTICIPATING",
+    "158": "SUCCESS",
     "159": "NON_PARTICIPATING",
     "16": "NON_PARTICIPATING",
-    "160": "NON_PARTICIPATING",
+    "160": "SUCCESS",
     "161": "NON_PARTICIPATING",
     "162": "NON_PARTICIPATING",
-    "163": "NON_PARTICIPATING",
+    "163": "SUCCESS",
     "164": "NON_PARTICIPATING",
     "165": "NON_PARTICIPATING",
     "166": "SUCCESS",
     "167": "NON_PARTICIPATING",
     "168": "NON_PARTICIPATING",
     "169": "NON_PARTICIPATING",
-    "17": "NON_PARTICIPATING",
-    "170": "NON_PARTICIPATING",
-    "171": "SUCCESS",
+    "17": "SUCCESS",
+    "170": "SUCCESS",
+    "171": "NON_PARTICIPATING",
     "172": "NON_PARTICIPATING",
-    "173": "SUCCESS",
-    "174": "SUCCESS",
-    "175": "SUCCESS",
-    "176": "SUCCESS",
+    "173": "NON_PARTICIPATING",
+    "174": "NON_PARTICIPATING",
+    "175": "NON_PARTICIPATING",
+    "176": "NON_PARTICIPATING",
     "177": "NON_PARTICIPATING",
-    "178": "SUCCESS",
-    "179": "NON_PARTICIPATING",
+    "178": "NON_PARTICIPATING",
+    "179": "SUCCESS",
     "18": "NON_PARTICIPATING",
-    "180": "NON_PARTICIPATING",
-    "181": "NON_PARTICIPATING",
+    "180": "SUCCESS",
+    "181": "SUCCESS",
     "182": "NON_PARTICIPATING",
     "183": "NON_PARTICIPATING",
-    "184": "SUCCESS",
+    "184": "NON_PARTICIPATING",
     "185": "NON_PARTICIPATING",
     "186": "NON_PARTICIPATING",
     "187": "SUCCESS",
-    "188": "SUCCESS",
-    "189": "SUCCESS",
-    "19": "NON_PARTICIPATING",
-    "190": "NON_PARTICIPATING",
-    "191": "NON_PARTICIPATING",
-    "192": "NON_PARTICIPATING",
+    "188": "NON_PARTICIPATING",
+    "189": "NON_PARTICIPATING",
+    "19": "SUCCESS",
+    "190": "SUCCESS",
+    "191": "SUCCESS",
+    "192": "SUCCESS",
     "193": "SUCCESS",
     "194": "NON_PARTICIPATING",
     "195": "NON_PARTICIPATING",
@@ -116,7 +116,7 @@
     "199": "NON_PARTICIPATING",
     "2": "NON_PARTICIPATING",
     "20": "NON_PARTICIPATING",
-    "200": "SUCCESS",
+    "200": "NON_PARTICIPATING",
     "201": "NON_PARTICIPATING",
     "202": "NON_PARTICIPATING",
     "203": "NON_PARTICIPATING",
@@ -124,54 +124,54 @@
     "205": "NON_PARTICIPATING",
     "206": "NON_PARTICIPATING",
     "207": "NON_PARTICIPATING",
-    "208": "SUCCESS",
-    "209": "SUCCESS",
+    "208": "NON_PARTICIPATING",
+    "209": "NON_PARTICIPATING",
     "21": "NON_PARTICIPATING",
-    "210": "SUCCESS",
-    "211": "SUCCESS",
+    "210": "NON_PARTICIPATING",
+    "211": "NON_PARTICIPATING",
     "212": "NON_PARTICIPATING",
     "213": "NON_PARTICIPATING",
-    "214": "SUCCESS",
+    "214": "NON_PARTICIPATING",
     "215": "SUCCESS",
     "216": "NON_PARTICIPATING",
-    "217": "SUCCESS",
-    "218": "SUCCESS",
+    "217": "NON_PARTICIPATING",
+    "218": "NON_PARTICIPATING",
     "219": "NON_PARTICIPATING",
-    "22": "SUCCESS",
+    "22": "NON_PARTICIPATING",
     "220": "NON_PARTICIPATING",
     "221": "SUCCESS",
     "222": "NON_PARTICIPATING",
     "223": "NON_PARTICIPATING",
     "224": "NON_PARTICIPATING",
-    "225": "SUCCESS",
+    "225": "NON_PARTICIPATING",
     "226": "NON_PARTICIPATING",
-    "227": "NON_PARTICIPATING",
-    "228": "NON_PARTICIPATING",
-    "229": "SUCCESS",
+    "227": "SUCCESS",
+    "228": "SUCCESS",
+    "229": "NON_PARTICIPATING",
     "23": "NON_PARTICIPATING",
     "230": "NON_PARTICIPATING",
     "231": "NON_PARTICIPATING",
-    "232": "NON_PARTICIPATING",
+    "232": "SUCCESS",
     "233": "NON_PARTICIPATING",
     "234": "NON_PARTICIPATING",
     "235": "NON_PARTICIPATING",
-    "236": "NON_PARTICIPATING",
+    "236": "SUCCESS",
     "237": "SUCCESS",
-    "238": "SUCCESS",
+    "238": "NON_PARTICIPATING",
     "239": "NON_PARTICIPATING",
-    "24": "NON_PARTICIPATING",
+    "24": "SUCCESS",
     "240": "NON_PARTICIPATING",
     "241": "SUCCESS",
-    "242": "SUCCESS",
-    "243": "SUCCESS",
+    "242": "NON_PARTICIPATING",
+    "243": "NON_PARTICIPATING",
     "244": "SUCCESS",
-    "245": "NON_PARTICIPATING",
+    "245": "SUCCESS",
     "246": "NON_PARTICIPATING",
-    "247": "SUCCESS",
+    "247": "NON_PARTICIPATING",
     "248": "NON_PARTICIPATING",
     "249": "NON_PARTICIPATING",
     "25": "SUCCESS",
-    "250": "SUCCESS",
+    "250": "NON_PARTICIPATING",
     "251": "NON_PARTICIPATING",
     "252": "NON_PARTICIPATING",
     "253": "NON_PARTICIPATING",
@@ -182,31 +182,31 @@
     "28": "SUCCESS",
     "29": "NON_PARTICIPATING",
     "3": "NON_PARTICIPATING",
-    "30": "SUCCESS",
-    "31": "NON_PARTICIPATING",
-    "32": "NON_PARTICIPATING",
+    "30": "NON_PARTICIPATING",
+    "31": "SUCCESS",
+    "32": "SUCCESS",
     "33": "NON_PARTICIPATING",
-    "34": "NON_PARTICIPATING",
-    "35": "SUCCESS",
+    "34": "SUCCESS",
+    "35": "NON_PARTICIPATING",
     "36": "NON_PARTICIPATING",
     "37": "NON_PARTICIPATING",
     "38": "NON_PARTICIPATING",
-    "39": "NON_PARTICIPATING",
-    "4": "NON_PARTICIPATING",
+    "39": "SUCCESS",
+    "4": "SUCCESS",
     "40": "NON_PARTICIPATING",
-    "41": "SUCCESS",
-    "42": "SUCCESS",
+    "41": "NON_PARTICIPATING",
+    "42": "NON_PARTICIPATING",
     "43": "NON_PARTICIPATING",
-    "44": "SUCCESS",
+    "44": "NON_PARTICIPATING",
     "45": "NON_PARTICIPATING",
-    "46": "NON_PARTICIPATING",
+    "46": "SUCCESS",
     "47": "SUCCESS",
     "48": "NON_PARTICIPATING",
-    "49": "SUCCESS",
-    "5": "NON_PARTICIPATING",
+    "49": "NON_PARTICIPATING",
+    "5": "SUCCESS",
     "50": "NON_PARTICIPATING",
     "51": "NON_PARTICIPATING",
-    "52": "SUCCESS",
+    "52": "NON_PARTICIPATING",
     "53": "SUCCESS",
     "54": "SUCCESS",
     "55": "NON_PARTICIPATING",
@@ -214,13 +214,13 @@
     "57": "NON_PARTICIPATING",
     "58": "NON_PARTICIPATING",
     "59": "NON_PARTICIPATING",
-    "6": "SUCCESS",
-    "60": "NON_PARTICIPATING",
-    "61": "NON_PARTICIPATING",
-    "62": "SUCCESS",
-    "63": "SUCCESS",
+    "6": "NON_PARTICIPATING",
+    "60": "SUCCESS",
+    "61": "SUCCESS",
+    "62": "NON_PARTICIPATING",
+    "63": "NON_PARTICIPATING",
     "64": "NON_PARTICIPATING",
-    "65": "NON_PARTICIPATING",
+    "65": "SUCCESS",
     "66": "NON_PARTICIPATING",
     "67": "NON_PARTICIPATING",
     "68": "NON_PARTICIPATING",
@@ -228,34 +228,34 @@
     "7": "SUCCESS",
     "70": "SUCCESS",
     "71": "NON_PARTICIPATING",
-    "72": "NON_PARTICIPATING",
+    "72": "SUCCESS",
     "73": "NON_PARTICIPATING",
-    "74": "NON_PARTICIPATING",
+    "74": "SUCCESS",
     "75": "SUCCESS",
-    "76": "SUCCESS",
+    "76": "NON_PARTICIPATING",
     "77": "SUCCESS",
     "78": "SUCCESS",
-    "79": "NON_PARTICIPATING",
-    "8": "SUCCESS",
+    "79": "SUCCESS",
+    "8": "NON_PARTICIPATING",
     "80": "NON_PARTICIPATING",
     "81": "NON_PARTICIPATING",
-    "82": "NON_PARTICIPATING",
-    "83": "SUCCESS",
+    "82": "SUCCESS",
+    "83": "NON_PARTICIPATING",
     "84": "NON_PARTICIPATING",
     "85": "NON_PARTICIPATING",
-    "86": "SUCCESS",
+    "86": "NON_PARTICIPATING",
     "87": "NON_PARTICIPATING",
     "88": "NON_PARTICIPATING",
-    "89": "NON_PARTICIPATING",
+    "89": "SUCCESS",
     "9": "NON_PARTICIPATING",
-    "90": "NON_PARTICIPATING",
+    "90": "SUCCESS",
     "91": "NON_PARTICIPATING",
-    "92": "SUCCESS",
-    "93": "SUCCESS",
-    "94": "SUCCESS",
+    "92": "NON_PARTICIPATING",
+    "93": "NON_PARTICIPATING",
+    "94": "NON_PARTICIPATING",
     "95": "NON_PARTICIPATING",
-    "96": "SUCCESS",
-    "97": "SUCCESS",
+    "96": "NON_PARTICIPATING",
+    "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 sha256:324f722c2501277887e60e392070d63ce88d393aa6f436ecd134ad45a256ab75
+oid sha256:ced5fcf654f3017f3a0b26282296b609cea6448ca3de4b9b1badeed814d90039
 size 4975530824

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ed91212c83a6a01a17f29654ae31b03fefbdf5aaa3b962db0ae6e35fe2be50bb
+oid sha256:9554330faa42f129455ef74817ee64b130374c880cf25ec0fe1fe8f9f183935c
 size 3075942944

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ee589c81463d3695e3c8ecaf3c73da73e96ac9f2b7bd9413a61927347d77f417
+oid sha256:2c5455060bb9231eeec48597a05eb44003b1b47d330866080d0b18d8da39ae16
 size 4675755870