Hugging Face's logo

distributed
/
optimized-gpt2-1b

Text Generation
Transformers
Safetensors
gpt_optimized
custom_code
Model card Files Community
kmfoda commited on
Commit
b1416f7
•
1 Parent(s): 85ea4eb

Epoch 460. Batch Size 4152. Peers 12.

Browse files
Files changed (2) hide show
  1. config.json +113 -113
  2. model.safetensors +1 -1
config.json CHANGED
@@ -4,104 +4,7 @@
4
  "all_reduce_scores": {
5
  "0": "NON_PARTICIPATING",
6
  "1": "NON_PARTICIPATING",
7
- "2": "SUCCESS",
8
- "3": "NON_PARTICIPATING",
9
- "4": "NON_PARTICIPATING",
10
- "5": "NON_PARTICIPATING",
11
- "6": "SUCCESS",
12
- "7": "SUCCESS",
13
- "8": "NON_PARTICIPATING",
14
- "9": "SUCCESS",
15
  "10": "NON_PARTICIPATING",
16
- "11": "NON_PARTICIPATING",
17
- "12": "NON_PARTICIPATING",
18
- "13": "NON_PARTICIPATING",
19
- "14": "NON_PARTICIPATING",
20
- "15": "NON_PARTICIPATING",
21
- "16": "NON_PARTICIPATING",
22
- "17": "NON_PARTICIPATING",
23
- "18": "NON_PARTICIPATING",
24
- "19": "NON_PARTICIPATING",
25
- "20": "NON_PARTICIPATING",
26
- "21": "NON_PARTICIPATING",
27
- "22": "SUCCESS",
28
- "23": "NON_PARTICIPATING",
29
- "24": "NON_PARTICIPATING",
30
- "25": "SUCCESS",
31
- "26": "NON_PARTICIPATING",
32
- "27": "NON_PARTICIPATING",
33
- "28": "SUCCESS",
34
- "29": "NON_PARTICIPATING",
35
- "30": "NON_PARTICIPATING",
36
- "31": "NON_PARTICIPATING",
37
- "32": "NON_PARTICIPATING",
38
- "33": "NON_PARTICIPATING",
39
- "34": "NON_PARTICIPATING",
40
- "35": "NON_PARTICIPATING",
41
- "36": "NON_PARTICIPATING",
42
- "37": "NON_PARTICIPATING",
43
- "38": "NON_PARTICIPATING",
44
- "39": "NON_PARTICIPATING",
45
- "40": "NON_PARTICIPATING",
46
- "41": "NON_PARTICIPATING",
47
- "42": "NON_PARTICIPATING",
48
- "43": "NON_PARTICIPATING",
49
- "44": "NON_PARTICIPATING",
50
- "45": "NON_PARTICIPATING",
51
- "46": "NON_PARTICIPATING",
52
- "47": "NON_PARTICIPATING",
53
- "48": "NON_PARTICIPATING",
54
- "49": "NON_PARTICIPATING",
55
- "50": "NON_PARTICIPATING",
56
- "51": "NON_PARTICIPATING",
57
- "52": "SUCCESS",
58
- "53": "NON_PARTICIPATING",
59
- "54": "NON_PARTICIPATING",
60
- "55": "NON_PARTICIPATING",
61
- "56": "NON_PARTICIPATING",
62
- "57": "NON_PARTICIPATING",
63
- "58": "SUCCESS",
64
- "59": "NON_PARTICIPATING",
65
- "60": "NON_PARTICIPATING",
66
- "61": "NON_PARTICIPATING",
67
- "62": "NON_PARTICIPATING",
68
- "63": "NON_PARTICIPATING",
69
- "64": "NON_PARTICIPATING",
70
- "65": "NON_PARTICIPATING",
71
- "66": "NON_PARTICIPATING",
72
- "67": "SUCCESS",
73
- "68": "NON_PARTICIPATING",
74
- "69": "NON_PARTICIPATING",
75
- "70": "NON_PARTICIPATING",
76
- "71": "NON_PARTICIPATING",
77
- "72": "NON_PARTICIPATING",
78
- "73": "NON_PARTICIPATING",
79
- "74": "NON_PARTICIPATING",
80
- "75": "SUCCESS",
81
- "76": "NON_PARTICIPATING",
82
- "77": "SUCCESS",
83
- "78": "NON_PARTICIPATING",
84
- "79": "NON_PARTICIPATING",
85
- "80": "NON_PARTICIPATING",
86
- "81": "NON_PARTICIPATING",
87
- "82": "NON_PARTICIPATING",
88
- "83": "NON_PARTICIPATING",
89
- "84": "NON_PARTICIPATING",
90
- "85": "NON_PARTICIPATING",
91
- "86": "NON_PARTICIPATING",
92
- "87": "NON_PARTICIPATING",
93
- "88": "NON_PARTICIPATING",
94
- "89": "NON_PARTICIPATING",
95
- "90": "NON_PARTICIPATING",
96
- "91": "NON_PARTICIPATING",
97
- "92": "NON_PARTICIPATING",
98
- "93": "NON_PARTICIPATING",
99
- "94": "NON_PARTICIPATING",
100
- "95": "SUCCESS",
101
- "96": "NON_PARTICIPATING",
102
- "97": "NON_PARTICIPATING",
103
- "98": "SUCCESS",
104
- "99": "NON_PARTICIPATING",
105
  "100": "NON_PARTICIPATING",
106
  "101": "NON_PARTICIPATING",
107
  "102": "NON_PARTICIPATING",
@@ -112,6 +15,7 @@
112
  "107": "NON_PARTICIPATING",
113
  "108": "NON_PARTICIPATING",
114
  "109": "NON_PARTICIPATING",
 
115
  "110": "NON_PARTICIPATING",
116
  "111": "NON_PARTICIPATING",
117
  "112": "NON_PARTICIPATING",
@@ -122,6 +26,7 @@
122
  "117": "NON_PARTICIPATING",
123
  "118": "NON_PARTICIPATING",
124
  "119": "NON_PARTICIPATING",
 
125
  "120": "NON_PARTICIPATING",
126
  "121": "NON_PARTICIPATING",
127
  "122": "NON_PARTICIPATING",
@@ -132,19 +37,21 @@
132
  "127": "NON_PARTICIPATING",
133
  "128": "NON_PARTICIPATING",
134
  "129": "NON_PARTICIPATING",
 
135
  "130": "NON_PARTICIPATING",
136
  "131": "NON_PARTICIPATING",
137
  "132": "NON_PARTICIPATING",
138
  "133": "NON_PARTICIPATING",
139
- "134": "SUCCESS",
140
  "135": "NON_PARTICIPATING",
141
  "136": "NON_PARTICIPATING",
142
  "137": "NON_PARTICIPATING",
143
  "138": "NON_PARTICIPATING",
144
  "139": "NON_PARTICIPATING",
145
- "140": "NON_PARTICIPATING",
146
- "141": "SUCCESS",
147
- "142": "NON_PARTICIPATING",
 
148
  "143": "NON_PARTICIPATING",
149
  "144": "NON_PARTICIPATING",
150
  "145": "NON_PARTICIPATING",
@@ -152,8 +59,9 @@
152
  "147": "NON_PARTICIPATING",
153
  "148": "NON_PARTICIPATING",
154
  "149": "NON_PARTICIPATING",
 
155
  "150": "NON_PARTICIPATING",
156
- "151": "SUCCESS",
157
  "152": "NON_PARTICIPATING",
158
  "153": "NON_PARTICIPATING",
159
  "154": "NON_PARTICIPATING",
@@ -161,7 +69,8 @@
161
  "156": "NON_PARTICIPATING",
162
  "157": "NON_PARTICIPATING",
163
  "158": "NON_PARTICIPATING",
164
- "159": "NON_PARTICIPATING",
 
165
  "160": "NON_PARTICIPATING",
166
  "161": "NON_PARTICIPATING",
167
  "162": "NON_PARTICIPATING",
@@ -172,6 +81,7 @@
172
  "167": "NON_PARTICIPATING",
173
  "168": "NON_PARTICIPATING",
174
  "169": "NON_PARTICIPATING",
 
175
  "170": "NON_PARTICIPATING",
176
  "171": "NON_PARTICIPATING",
177
  "172": "NON_PARTICIPATING",
@@ -180,8 +90,9 @@
180
  "175": "NON_PARTICIPATING",
181
  "176": "NON_PARTICIPATING",
182
  "177": "NON_PARTICIPATING",
183
- "178": "SUCCESS",
184
  "179": "NON_PARTICIPATING",
 
185
  "180": "NON_PARTICIPATING",
186
  "181": "NON_PARTICIPATING",
187
  "182": "NON_PARTICIPATING",
@@ -192,36 +103,41 @@
192
  "187": "NON_PARTICIPATING",
193
  "188": "NON_PARTICIPATING",
194
  "189": "NON_PARTICIPATING",
 
195
  "190": "NON_PARTICIPATING",
196
  "191": "NON_PARTICIPATING",
197
  "192": "NON_PARTICIPATING",
198
  "193": "NON_PARTICIPATING",
199
  "194": "NON_PARTICIPATING",
200
- "195": "SUCCESS",
201
  "196": "NON_PARTICIPATING",
202
  "197": "NON_PARTICIPATING",
203
  "198": "NON_PARTICIPATING",
204
  "199": "NON_PARTICIPATING",
 
 
205
  "200": "NON_PARTICIPATING",
206
- "201": "NON_PARTICIPATING",
207
- "202": "SUCCESS",
208
  "203": "NON_PARTICIPATING",
209
- "204": "SUCCESS",
210
  "205": "NON_PARTICIPATING",
211
  "206": "NON_PARTICIPATING",
212
  "207": "NON_PARTICIPATING",
213
  "208": "NON_PARTICIPATING",
214
- "209": "SUCCESS",
 
215
  "210": "NON_PARTICIPATING",
216
  "211": "NON_PARTICIPATING",
217
  "212": "NON_PARTICIPATING",
218
- "213": "SUCCESS",
219
  "214": "NON_PARTICIPATING",
220
  "215": "NON_PARTICIPATING",
221
- "216": "SUCCESS",
222
- "217": "SUCCESS",
223
  "218": "NON_PARTICIPATING",
224
  "219": "NON_PARTICIPATING",
 
225
  "220": "NON_PARTICIPATING",
226
  "221": "NON_PARTICIPATING",
227
  "222": "NON_PARTICIPATING",
@@ -232,6 +148,7 @@
232
  "227": "NON_PARTICIPATING",
233
  "228": "NON_PARTICIPATING",
234
  "229": "NON_PARTICIPATING",
 
235
  "230": "NON_PARTICIPATING",
236
  "231": "NON_PARTICIPATING",
237
  "232": "NON_PARTICIPATING",
@@ -242,6 +159,7 @@
242
  "237": "NON_PARTICIPATING",
243
  "238": "NON_PARTICIPATING",
244
  "239": "NON_PARTICIPATING",
 
245
  "240": "NON_PARTICIPATING",
246
  "241": "NON_PARTICIPATING",
247
  "242": "NON_PARTICIPATING",
@@ -252,12 +170,94 @@
252
  "247": "NON_PARTICIPATING",
253
  "248": "NON_PARTICIPATING",
254
  "249": "NON_PARTICIPATING",
 
255
  "250": "NON_PARTICIPATING",
256
  "251": "NON_PARTICIPATING",
257
  "252": "NON_PARTICIPATING",
258
  "253": "NON_PARTICIPATING",
259
  "254": "NON_PARTICIPATING",
260
- "255": "NON_PARTICIPATING"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
261
  },
262
  "architectures": [
263
  "GPTOptim"
 
4
  "all_reduce_scores": {
5
  "0": "NON_PARTICIPATING",
6
  "1": "NON_PARTICIPATING",
 
 
 
 
 
 
 
 
7
  "10": "NON_PARTICIPATING",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8
  "100": "NON_PARTICIPATING",
9
  "101": "NON_PARTICIPATING",
10
  "102": "NON_PARTICIPATING",
 
15
  "107": "NON_PARTICIPATING",
16
  "108": "NON_PARTICIPATING",
17
  "109": "NON_PARTICIPATING",
18
+ "11": "NON_PARTICIPATING",
19
  "110": "NON_PARTICIPATING",
20
  "111": "NON_PARTICIPATING",
21
  "112": "NON_PARTICIPATING",
 
26
  "117": "NON_PARTICIPATING",
27
  "118": "NON_PARTICIPATING",
28
  "119": "NON_PARTICIPATING",
29
+ "12": "NON_PARTICIPATING",
30
  "120": "NON_PARTICIPATING",
31
  "121": "NON_PARTICIPATING",
32
  "122": "NON_PARTICIPATING",
 
37
  "127": "NON_PARTICIPATING",
38
  "128": "NON_PARTICIPATING",
39
  "129": "NON_PARTICIPATING",
40
+ "13": "NON_PARTICIPATING",
41
  "130": "NON_PARTICIPATING",
42
  "131": "NON_PARTICIPATING",
43
  "132": "NON_PARTICIPATING",
44
  "133": "NON_PARTICIPATING",
45
+ "134": "NON_PARTICIPATING",
46
  "135": "NON_PARTICIPATING",
47
  "136": "NON_PARTICIPATING",
48
  "137": "NON_PARTICIPATING",
49
  "138": "NON_PARTICIPATING",
50
  "139": "NON_PARTICIPATING",
51
+ "14": "NON_PARTICIPATING",
52
+ "140": "SUCCESS",
53
+ "141": "NON_PARTICIPATING",
54
+ "142": "SUCCESS",
55
  "143": "NON_PARTICIPATING",
56
  "144": "NON_PARTICIPATING",
57
  "145": "NON_PARTICIPATING",
 
59
  "147": "NON_PARTICIPATING",
60
  "148": "NON_PARTICIPATING",
61
  "149": "NON_PARTICIPATING",
62
+ "15": "NON_PARTICIPATING",
63
  "150": "NON_PARTICIPATING",
64
+ "151": "NON_PARTICIPATING",
65
  "152": "NON_PARTICIPATING",
66
  "153": "NON_PARTICIPATING",
67
  "154": "NON_PARTICIPATING",
 
69
  "156": "NON_PARTICIPATING",
70
  "157": "NON_PARTICIPATING",
71
  "158": "NON_PARTICIPATING",
72
+ "159": "SUCCESS",
73
+ "16": "NON_PARTICIPATING",
74
  "160": "NON_PARTICIPATING",
75
  "161": "NON_PARTICIPATING",
76
  "162": "NON_PARTICIPATING",
 
81
  "167": "NON_PARTICIPATING",
82
  "168": "NON_PARTICIPATING",
83
  "169": "NON_PARTICIPATING",
84
+ "17": "NON_PARTICIPATING",
85
  "170": "NON_PARTICIPATING",
86
  "171": "NON_PARTICIPATING",
87
  "172": "NON_PARTICIPATING",
 
90
  "175": "NON_PARTICIPATING",
91
  "176": "NON_PARTICIPATING",
92
  "177": "NON_PARTICIPATING",
93
+ "178": "NON_PARTICIPATING",
94
  "179": "NON_PARTICIPATING",
95
+ "18": "NON_PARTICIPATING",
96
  "180": "NON_PARTICIPATING",
97
  "181": "NON_PARTICIPATING",
98
  "182": "NON_PARTICIPATING",
 
103
  "187": "NON_PARTICIPATING",
104
  "188": "NON_PARTICIPATING",
105
  "189": "NON_PARTICIPATING",
106
+ "19": "NON_PARTICIPATING",
107
  "190": "NON_PARTICIPATING",
108
  "191": "NON_PARTICIPATING",
109
  "192": "NON_PARTICIPATING",
110
  "193": "NON_PARTICIPATING",
111
  "194": "NON_PARTICIPATING",
112
+ "195": "NON_PARTICIPATING",
113
  "196": "NON_PARTICIPATING",
114
  "197": "NON_PARTICIPATING",
115
  "198": "NON_PARTICIPATING",
116
  "199": "NON_PARTICIPATING",
117
+ "2": "SUCCESS",
118
+ "20": "NON_PARTICIPATING",
119
  "200": "NON_PARTICIPATING",
120
+ "201": "SUCCESS",
121
+ "202": "NON_PARTICIPATING",
122
  "203": "NON_PARTICIPATING",
123
+ "204": "NON_PARTICIPATING",
124
  "205": "NON_PARTICIPATING",
125
  "206": "NON_PARTICIPATING",
126
  "207": "NON_PARTICIPATING",
127
  "208": "NON_PARTICIPATING",
128
+ "209": "NON_PARTICIPATING",
129
+ "21": "NON_PARTICIPATING",
130
  "210": "NON_PARTICIPATING",
131
  "211": "NON_PARTICIPATING",
132
  "212": "NON_PARTICIPATING",
133
+ "213": "NON_PARTICIPATING",
134
  "214": "NON_PARTICIPATING",
135
  "215": "NON_PARTICIPATING",
136
+ "216": "NON_PARTICIPATING",
137
+ "217": "NON_PARTICIPATING",
138
  "218": "NON_PARTICIPATING",
139
  "219": "NON_PARTICIPATING",
140
+ "22": "NON_PARTICIPATING",
141
  "220": "NON_PARTICIPATING",
142
  "221": "NON_PARTICIPATING",
143
  "222": "NON_PARTICIPATING",
 
148
  "227": "NON_PARTICIPATING",
149
  "228": "NON_PARTICIPATING",
150
  "229": "NON_PARTICIPATING",
151
+ "23": "NON_PARTICIPATING",
152
  "230": "NON_PARTICIPATING",
153
  "231": "NON_PARTICIPATING",
154
  "232": "NON_PARTICIPATING",
 
159
  "237": "NON_PARTICIPATING",
160
  "238": "NON_PARTICIPATING",
161
  "239": "NON_PARTICIPATING",
162
+ "24": "NON_PARTICIPATING",
163
  "240": "NON_PARTICIPATING",
164
  "241": "NON_PARTICIPATING",
165
  "242": "NON_PARTICIPATING",
 
170
  "247": "NON_PARTICIPATING",
171
  "248": "NON_PARTICIPATING",
172
  "249": "NON_PARTICIPATING",
173
+ "25": "SUCCESS",
174
  "250": "NON_PARTICIPATING",
175
  "251": "NON_PARTICIPATING",
176
  "252": "NON_PARTICIPATING",
177
  "253": "NON_PARTICIPATING",
178
  "254": "NON_PARTICIPATING",
179
+ "255": "NON_PARTICIPATING",
180
+ "26": "NON_PARTICIPATING",
181
+ "27": "NON_PARTICIPATING",
182
+ "28": "NON_PARTICIPATING",
183
+ "29": "NON_PARTICIPATING",
184
+ "3": "NON_PARTICIPATING",
185
+ "30": "NON_PARTICIPATING",
186
+ "31": "NON_PARTICIPATING",
187
+ "32": "NON_PARTICIPATING",
188
+ "33": "NON_PARTICIPATING",
189
+ "34": "NON_PARTICIPATING",
190
+ "35": "NON_PARTICIPATING",
191
+ "36": "NON_PARTICIPATING",
192
+ "37": "NON_PARTICIPATING",
193
+ "38": "NON_PARTICIPATING",
194
+ "39": "NON_PARTICIPATING",
195
+ "4": "NON_PARTICIPATING",
196
+ "40": "NON_PARTICIPATING",
197
+ "41": "NON_PARTICIPATING",
198
+ "42": "NON_PARTICIPATING",
199
+ "43": "NON_PARTICIPATING",
200
+ "44": "NON_PARTICIPATING",
201
+ "45": "SUCCESS",
202
+ "46": "NON_PARTICIPATING",
203
+ "47": "NON_PARTICIPATING",
204
+ "48": "NON_PARTICIPATING",
205
+ "49": "NON_PARTICIPATING",
206
+ "5": "NON_PARTICIPATING",
207
+ "50": "NON_PARTICIPATING",
208
+ "51": "NON_PARTICIPATING",
209
+ "52": "NON_PARTICIPATING",
210
+ "53": "NON_PARTICIPATING",
211
+ "54": "NON_PARTICIPATING",
212
+ "55": "NON_PARTICIPATING",
213
+ "56": "NON_PARTICIPATING",
214
+ "57": "NON_PARTICIPATING",
215
+ "58": "SUCCESS",
216
+ "59": "NON_PARTICIPATING",
217
+ "6": "SUCCESS",
218
+ "60": "NON_PARTICIPATING",
219
+ "61": "NON_PARTICIPATING",
220
+ "62": "NON_PARTICIPATING",
221
+ "63": "NON_PARTICIPATING",
222
+ "64": "NON_PARTICIPATING",
223
+ "65": "NON_PARTICIPATING",
224
+ "66": "NON_PARTICIPATING",
225
+ "67": "NON_PARTICIPATING",
226
+ "68": "NON_PARTICIPATING",
227
+ "69": "NON_PARTICIPATING",
228
+ "7": "NON_PARTICIPATING",
229
+ "70": "NON_PARTICIPATING",
230
+ "71": "NON_PARTICIPATING",
231
+ "72": "NON_PARTICIPATING",
232
+ "73": "NON_PARTICIPATING",
233
+ "74": "SUCCESS",
234
+ "75": "NON_PARTICIPATING",
235
+ "76": "NON_PARTICIPATING",
236
+ "77": "NON_PARTICIPATING",
237
+ "78": "SUCCESS",
238
+ "79": "NON_PARTICIPATING",
239
+ "8": "NON_PARTICIPATING",
240
+ "80": "NON_PARTICIPATING",
241
+ "81": "NON_PARTICIPATING",
242
+ "82": "NON_PARTICIPATING",
243
+ "83": "NON_PARTICIPATING",
244
+ "84": "NON_PARTICIPATING",
245
+ "85": "SUCCESS",
246
+ "86": "NON_PARTICIPATING",
247
+ "87": "NON_PARTICIPATING",
248
+ "88": "NON_PARTICIPATING",
249
+ "89": "NON_PARTICIPATING",
250
+ "9": "NON_PARTICIPATING",
251
+ "90": "NON_PARTICIPATING",
252
+ "91": "NON_PARTICIPATING",
253
+ "92": "NON_PARTICIPATING",
254
+ "93": "NON_PARTICIPATING",
255
+ "94": "NON_PARTICIPATING",
256
+ "95": "NON_PARTICIPATING",
257
+ "96": "NON_PARTICIPATING",
258
+ "97": "NON_PARTICIPATING",
259
+ "98": "NON_PARTICIPATING",
260
+ "99": "NON_PARTICIPATING"
261
  },
262
  "architectures": [
263
  "GPTOptim"
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:202fbe1122f23297247d2b39eeeb0491cb9c8696b8b64c313ccfaa91d5d2dd25
3
  size 4040701744
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:805dab43124032164a1763547559dbd61e1f20177e71e22434129ae78cfcff65
3
  size 4040701744