kmfoda commited on
Commit
79df68d
·
verified ·
1 Parent(s): 75f2f74

Run 4. Outer Step 79. Inner Step 0. Peers 96.

Browse files
Files changed (4) hide show
  1. config.json +71 -71
  2. inner_optimizer.pt +1 -1
  3. model.safetensors +1 -1
  4. outer_optimizer.pt +1 -1
config.json CHANGED
@@ -3,23 +3,23 @@
3
  "activation_function": "gelu_new",
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": "SUCCESS",
11
  "103": "NON_PARTICIPATING",
12
- "104": "NON_PARTICIPATING",
13
  "105": "NON_PARTICIPATING",
14
  "106": "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",
22
- "113": "NON_PARTICIPATING",
23
  "114": "NON_PARTICIPATING",
24
  "115": "NON_PARTICIPATING",
25
  "116": "NON_PARTICIPATING",
@@ -33,7 +33,7 @@
33
  "123": "NON_PARTICIPATING",
34
  "124": "NON_PARTICIPATING",
35
  "125": "NON_PARTICIPATING",
36
- "126": "NON_PARTICIPATING",
37
  "127": "NON_PARTICIPATING",
38
  "128": "NON_PARTICIPATING",
39
  "129": "NON_PARTICIPATING",
@@ -42,7 +42,7 @@
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",
@@ -52,18 +52,18 @@
52
  "140": "NON_PARTICIPATING",
53
  "141": "NON_PARTICIPATING",
54
  "142": "NON_PARTICIPATING",
55
- "143": "NON_PARTICIPATING",
56
  "144": "NON_PARTICIPATING",
57
- "145": "SUCCESS",
58
  "146": "NON_PARTICIPATING",
59
  "147": "NON_PARTICIPATING",
60
  "148": "NON_PARTICIPATING",
61
  "149": "NON_PARTICIPATING",
62
  "15": "SUCCESS",
63
- "150": "NON_PARTICIPATING",
64
  "151": "NON_PARTICIPATING",
65
- "152": "NON_PARTICIPATING",
66
- "153": "NON_PARTICIPATING",
67
  "154": "NON_PARTICIPATING",
68
  "155": "NON_PARTICIPATING",
69
  "156": "NON_PARTICIPATING",
@@ -75,15 +75,15 @@
75
  "161": "NON_PARTICIPATING",
76
  "162": "NON_PARTICIPATING",
77
  "163": "NON_PARTICIPATING",
78
- "164": "NON_PARTICIPATING",
79
  "165": "NON_PARTICIPATING",
80
  "166": "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",
88
  "173": "NON_PARTICIPATING",
89
  "174": "NON_PARTICIPATING",
@@ -92,18 +92,18 @@
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",
99
  "183": "NON_PARTICIPATING",
100
- "184": "NON_PARTICIPATING",
101
  "185": "NON_PARTICIPATING",
102
  "186": "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",
@@ -115,7 +115,7 @@
115
  "198": "NON_PARTICIPATING",
116
  "199": "NON_PARTICIPATING",
117
  "2": "SUCCESS",
118
- "20": "NON_PARTICIPATING",
119
  "200": "NON_PARTICIPATING",
120
  "201": "NON_PARTICIPATING",
121
  "202": "NON_PARTICIPATING",
@@ -126,7 +126,7 @@
126
  "207": "NON_PARTICIPATING",
127
  "208": "NON_PARTICIPATING",
128
  "209": "NON_PARTICIPATING",
129
- "21": "SUCCESS",
130
  "210": "NON_PARTICIPATING",
131
  "211": "NON_PARTICIPATING",
132
  "212": "SUCCESS",
@@ -137,11 +137,11 @@
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",
144
- "223": "SUCCESS",
145
  "224": "NON_PARTICIPATING",
146
  "225": "NON_PARTICIPATING",
147
  "226": "NON_PARTICIPATING",
@@ -155,21 +155,21 @@
155
  "233": "NON_PARTICIPATING",
156
  "234": "NON_PARTICIPATING",
157
  "235": "NON_PARTICIPATING",
158
- "236": "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",
166
  "243": "NON_PARTICIPATING",
167
  "244": "NON_PARTICIPATING",
168
- "245": "NON_PARTICIPATING",
169
  "246": "NON_PARTICIPATING",
170
  "247": "NON_PARTICIPATING",
171
  "248": "NON_PARTICIPATING",
172
- "249": "NON_PARTICIPATING",
173
  "25": "NON_PARTICIPATING",
174
  "250": "NON_PARTICIPATING",
175
  "251": "NON_PARTICIPATING",
@@ -177,87 +177,87 @@
177
  "253": "NON_PARTICIPATING",
178
  "254": "NON_PARTICIPATING",
179
  "255": "NON_PARTICIPATING",
180
- "26": "NON_PARTICIPATING",
181
  "27": "SUCCESS",
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": "SUCCESS",
190
- "35": "NON_PARTICIPATING",
191
  "36": "NON_PARTICIPATING",
192
  "37": "SUCCESS",
193
- "38": "NON_PARTICIPATING",
194
  "39": "SUCCESS",
195
- "4": "NON_PARTICIPATING",
196
- "40": "NON_PARTICIPATING",
197
  "41": "SUCCESS",
198
  "42": "NON_PARTICIPATING",
199
  "43": "NON_PARTICIPATING",
200
  "44": "SUCCESS",
201
- "45": "NON_PARTICIPATING",
202
- "46": "NON_PARTICIPATING",
203
- "47": "NON_PARTICIPATING",
204
  "48": "NON_PARTICIPATING",
205
  "49": "SUCCESS",
206
  "5": "NON_PARTICIPATING",
207
  "50": "SUCCESS",
208
- "51": "NON_PARTICIPATING",
209
- "52": "NON_PARTICIPATING",
210
  "53": "SUCCESS",
211
  "54": "SUCCESS",
212
- "55": "NON_PARTICIPATING",
213
- "56": "NON_PARTICIPATING",
214
  "57": "NON_PARTICIPATING",
215
  "58": "NON_PARTICIPATING",
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": "SUCCESS",
223
  "65": "SUCCESS",
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": "SUCCESS",
233
- "74": "NON_PARTICIPATING",
234
  "75": "NON_PARTICIPATING",
235
- "76": "NON_PARTICIPATING",
236
- "77": "NON_PARTICIPATING",
237
- "78": "NON_PARTICIPATING",
238
  "79": "NON_PARTICIPATING",
239
- "8": "NON_PARTICIPATING",
240
  "80": "SUCCESS",
241
  "81": "SUCCESS",
242
- "82": "NON_PARTICIPATING",
243
- "83": "NON_PARTICIPATING",
244
- "84": "NON_PARTICIPATING",
245
  "85": "NON_PARTICIPATING",
246
  "86": "NON_PARTICIPATING",
247
  "87": "SUCCESS",
248
- "88": "NON_PARTICIPATING",
249
- "89": "NON_PARTICIPATING",
250
  "9": "SUCCESS",
251
  "90": "SUCCESS",
252
- "91": "NON_PARTICIPATING",
253
- "92": "SUCCESS",
254
  "93": "NON_PARTICIPATING",
255
- "94": "NON_PARTICIPATING",
256
- "95": "NON_PARTICIPATING",
257
  "96": "NON_PARTICIPATING",
258
- "97": "NON_PARTICIPATING",
259
  "98": "SUCCESS",
260
- "99": "NON_PARTICIPATING"
261
  },
262
  "architectures": [
263
  "GPTOptim"
@@ -275,7 +275,7 @@
275
  "initializer_range": 0.02,
276
  "inner_step": 0,
277
  "inner_steps": 0,
278
- "last_allreduce_block": 5665748,
279
  "layer_norm_epsilon": 1e-05,
280
  "model_type": "gpt_optimized",
281
  "n_embd": 1280,
 
3
  "activation_function": "gelu_new",
4
  "all_reduce_scores": {
5
  "0": "NON_PARTICIPATING",
6
+ "1": "SUCCESS",
7
  "10": "NON_PARTICIPATING",
8
  "100": "NON_PARTICIPATING",
9
  "101": "NON_PARTICIPATING",
10
  "102": "SUCCESS",
11
  "103": "NON_PARTICIPATING",
12
+ "104": "SUCCESS",
13
  "105": "NON_PARTICIPATING",
14
  "106": "NON_PARTICIPATING",
15
  "107": "NON_PARTICIPATING",
16
  "108": "NON_PARTICIPATING",
17
  "109": "NON_PARTICIPATING",
18
+ "11": "SUCCESS",
19
  "110": "NON_PARTICIPATING",
20
  "111": "NON_PARTICIPATING",
21
  "112": "NON_PARTICIPATING",
22
+ "113": "SUCCESS",
23
  "114": "NON_PARTICIPATING",
24
  "115": "NON_PARTICIPATING",
25
  "116": "NON_PARTICIPATING",
 
33
  "123": "NON_PARTICIPATING",
34
  "124": "NON_PARTICIPATING",
35
  "125": "NON_PARTICIPATING",
36
+ "126": "SUCCESS",
37
  "127": "NON_PARTICIPATING",
38
  "128": "NON_PARTICIPATING",
39
  "129": "NON_PARTICIPATING",
 
42
  "131": "NON_PARTICIPATING",
43
  "132": "NON_PARTICIPATING",
44
  "133": "NON_PARTICIPATING",
45
+ "134": "SUCCESS",
46
  "135": "NON_PARTICIPATING",
47
  "136": "NON_PARTICIPATING",
48
  "137": "NON_PARTICIPATING",
 
52
  "140": "NON_PARTICIPATING",
53
  "141": "NON_PARTICIPATING",
54
  "142": "NON_PARTICIPATING",
55
+ "143": "SUCCESS",
56
  "144": "NON_PARTICIPATING",
57
+ "145": "NON_PARTICIPATING",
58
  "146": "NON_PARTICIPATING",
59
  "147": "NON_PARTICIPATING",
60
  "148": "NON_PARTICIPATING",
61
  "149": "NON_PARTICIPATING",
62
  "15": "SUCCESS",
63
+ "150": "SUCCESS",
64
  "151": "NON_PARTICIPATING",
65
+ "152": "SUCCESS",
66
+ "153": "SUCCESS",
67
  "154": "NON_PARTICIPATING",
68
  "155": "NON_PARTICIPATING",
69
  "156": "NON_PARTICIPATING",
 
75
  "161": "NON_PARTICIPATING",
76
  "162": "NON_PARTICIPATING",
77
  "163": "NON_PARTICIPATING",
78
+ "164": "SUCCESS",
79
  "165": "NON_PARTICIPATING",
80
  "166": "NON_PARTICIPATING",
81
  "167": "NON_PARTICIPATING",
82
  "168": "NON_PARTICIPATING",
83
  "169": "NON_PARTICIPATING",
84
+ "17": "SUCCESS",
85
  "170": "NON_PARTICIPATING",
86
+ "171": "SUCCESS",
87
  "172": "NON_PARTICIPATING",
88
  "173": "NON_PARTICIPATING",
89
  "174": "NON_PARTICIPATING",
 
92
  "177": "NON_PARTICIPATING",
93
  "178": "NON_PARTICIPATING",
94
  "179": "NON_PARTICIPATING",
95
+ "18": "SUCCESS",
96
  "180": "NON_PARTICIPATING",
97
+ "181": "SUCCESS",
98
  "182": "NON_PARTICIPATING",
99
  "183": "NON_PARTICIPATING",
100
+ "184": "SUCCESS",
101
  "185": "NON_PARTICIPATING",
102
  "186": "NON_PARTICIPATING",
103
  "187": "NON_PARTICIPATING",
104
  "188": "NON_PARTICIPATING",
105
  "189": "NON_PARTICIPATING",
106
+ "19": "SUCCESS",
107
  "190": "NON_PARTICIPATING",
108
  "191": "NON_PARTICIPATING",
109
  "192": "NON_PARTICIPATING",
 
115
  "198": "NON_PARTICIPATING",
116
  "199": "NON_PARTICIPATING",
117
  "2": "SUCCESS",
118
+ "20": "SUCCESS",
119
  "200": "NON_PARTICIPATING",
120
  "201": "NON_PARTICIPATING",
121
  "202": "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": "SUCCESS",
 
137
  "217": "NON_PARTICIPATING",
138
  "218": "NON_PARTICIPATING",
139
  "219": "NON_PARTICIPATING",
140
+ "22": "SUCCESS",
141
  "220": "NON_PARTICIPATING",
142
  "221": "NON_PARTICIPATING",
143
  "222": "NON_PARTICIPATING",
144
+ "223": "NON_PARTICIPATING",
145
  "224": "NON_PARTICIPATING",
146
  "225": "NON_PARTICIPATING",
147
  "226": "NON_PARTICIPATING",
 
155
  "233": "NON_PARTICIPATING",
156
  "234": "NON_PARTICIPATING",
157
  "235": "NON_PARTICIPATING",
158
+ "236": "SUCCESS",
159
  "237": "NON_PARTICIPATING",
160
  "238": "NON_PARTICIPATING",
161
  "239": "NON_PARTICIPATING",
162
+ "24": "SUCCESS",
163
  "240": "NON_PARTICIPATING",
164
  "241": "NON_PARTICIPATING",
165
  "242": "NON_PARTICIPATING",
166
  "243": "NON_PARTICIPATING",
167
  "244": "NON_PARTICIPATING",
168
+ "245": "SUCCESS",
169
  "246": "NON_PARTICIPATING",
170
  "247": "NON_PARTICIPATING",
171
  "248": "NON_PARTICIPATING",
172
+ "249": "SUCCESS",
173
  "25": "NON_PARTICIPATING",
174
  "250": "NON_PARTICIPATING",
175
  "251": "NON_PARTICIPATING",
 
177
  "253": "NON_PARTICIPATING",
178
  "254": "NON_PARTICIPATING",
179
  "255": "NON_PARTICIPATING",
180
+ "26": "SUCCESS",
181
  "27": "SUCCESS",
182
  "28": "NON_PARTICIPATING",
183
+ "29": "SUCCESS",
184
+ "3": "SUCCESS",
185
  "30": "NON_PARTICIPATING",
186
+ "31": "SUCCESS",
187
+ "32": "SUCCESS",
188
  "33": "NON_PARTICIPATING",
189
  "34": "SUCCESS",
190
+ "35": "SUCCESS",
191
  "36": "NON_PARTICIPATING",
192
  "37": "SUCCESS",
193
+ "38": "SUCCESS",
194
  "39": "SUCCESS",
195
+ "4": "SUCCESS",
196
+ "40": "SUCCESS",
197
  "41": "SUCCESS",
198
  "42": "NON_PARTICIPATING",
199
  "43": "NON_PARTICIPATING",
200
  "44": "SUCCESS",
201
+ "45": "SUCCESS",
202
+ "46": "SUCCESS",
203
+ "47": "SUCCESS",
204
  "48": "NON_PARTICIPATING",
205
  "49": "SUCCESS",
206
  "5": "NON_PARTICIPATING",
207
  "50": "SUCCESS",
208
+ "51": "SUCCESS",
209
+ "52": "SUCCESS",
210
  "53": "SUCCESS",
211
  "54": "SUCCESS",
212
+ "55": "SUCCESS",
213
+ "56": "SUCCESS",
214
  "57": "NON_PARTICIPATING",
215
  "58": "NON_PARTICIPATING",
216
  "59": "NON_PARTICIPATING",
217
  "6": "SUCCESS",
218
+ "60": "SUCCESS",
219
+ "61": "SUCCESS",
220
+ "62": "SUCCESS",
221
+ "63": "SUCCESS",
222
  "64": "SUCCESS",
223
  "65": "SUCCESS",
224
+ "66": "SUCCESS",
225
+ "67": "SUCCESS",
226
+ "68": "SUCCESS",
227
+ "69": "SUCCESS",
228
+ "7": "SUCCESS",
229
+ "70": "SUCCESS",
230
+ "71": "SUCCESS",
231
+ "72": "SUCCESS",
232
  "73": "SUCCESS",
233
+ "74": "SUCCESS",
234
  "75": "NON_PARTICIPATING",
235
+ "76": "SUCCESS",
236
+ "77": "SUCCESS",
237
+ "78": "SUCCESS",
238
  "79": "NON_PARTICIPATING",
239
+ "8": "SUCCESS",
240
  "80": "SUCCESS",
241
  "81": "SUCCESS",
242
+ "82": "SUCCESS",
243
+ "83": "SUCCESS",
244
+ "84": "SUCCESS",
245
  "85": "NON_PARTICIPATING",
246
  "86": "NON_PARTICIPATING",
247
  "87": "SUCCESS",
248
+ "88": "SUCCESS",
249
+ "89": "SUCCESS",
250
  "9": "SUCCESS",
251
  "90": "SUCCESS",
252
+ "91": "SUCCESS",
253
+ "92": "NON_PARTICIPATING",
254
  "93": "NON_PARTICIPATING",
255
+ "94": "SUCCESS",
256
+ "95": "SUCCESS",
257
  "96": "NON_PARTICIPATING",
258
+ "97": "SUCCESS",
259
  "98": "SUCCESS",
260
+ "99": "SUCCESS"
261
  },
262
  "architectures": [
263
  "GPTOptim"
 
275
  "initializer_range": 0.02,
276
  "inner_step": 0,
277
  "inner_steps": 0,
278
+ "last_allreduce_block": 5667101,
279
  "layer_norm_epsilon": 1e-05,
280
  "model_type": "gpt_optimized",
281
  "n_embd": 1280,
inner_optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:58d247a2ff38a1668c26ed4b2c5eee13664bcfaf79f27b9167d3ca85fc7cf999
3
  size 8081782026
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0c0fc66a2b1eba2734a8df554859ca43aefc75886716f50fbe5b3cbaf864ac32
3
  size 8081782026
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:011617872d27740638fa544a890b2f4ffb1ef4a2ba59167afbe52cfa4244321d
3
  size 4040701744
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df7b42b4b73da07a580253d738c8dc62f4e350c42e3514c51b3e2c4b1a1defbd
3
  size 4040701744
outer_optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f5fe374380cebeaef48d7a2fdc9b6414bc91a7cd2a5cf288a8c2acf097b6335b
3
  size 4040805354
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c02fde4d7e55ab47d256c136ce06c3f328237980d3b3f1fa08fe0af77c4b2031
3
  size 4040805354