Hugging Face's logo

kmfoda
/
gpt2-1b-miner-2

Safetensors
gpt_optimized
custom_code
Model card Files Community
kmfoda commited on
Commit
a7b2e53
·
verified ·
1 Parent(s): 02aed09

Outer Step 0. Inner Step 2. Batch Size 476

Browse files
Files changed (2) hide show
  1. config.json +260 -20
  2. model.safetensors +1 -1
config.json CHANGED
@@ -4,20 +4,260 @@
4
  "all_reduce_scores": {
5
  "0": "SUCCESS",
6
  "1": "SUCCESS",
7
- "10": "NON_PARTICIPATING",
8
- "11": "NON_PARTICIPATING",
9
- "12": "NON_PARTICIPATING",
10
- "13": "NON_PARTICIPATING",
11
- "14": "NON_PARTICIPATING",
12
- "15": "NON_PARTICIPATING",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
13
  "2": "SUCCESS",
14
- "3": "NON_PARTICIPATING",
15
- "4": "NON_PARTICIPATING",
16
- "5": "NON_PARTICIPATING",
17
- "6": "NON_PARTICIPATING",
18
- "7": "NON_PARTICIPATING",
19
- "8": "NON_PARTICIPATING",
20
- "9": "NON_PARTICIPATING"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
21
  },
22
  "architectures": [
23
  "GPTOptim"
@@ -28,20 +268,20 @@
28
  "AutoModelForCausalLM": "distributed/optimized-gpt2-500m--modeling_gpt_optimized.GPTOptim"
29
  },
30
  "block_list": [
31
- 4017966,
32
- 4017969,
33
- 4017973,
34
- 4017977,
35
- 4017981
36
  ],
37
  "block_size": 1024,
38
  "bos_token_id": 50256,
39
  "embd_pdrop": 0.1,
40
  "eos_token_id": 50256,
41
  "initializer_range": 0.02,
42
- "inner_step": 12,
43
  "inner_steps": 0,
44
- "last_allreduce_block": 4017198,
45
  "layer_norm_epsilon": 1e-05,
46
  "model_type": "gpt_optimized",
47
  "n_embd": 1280,
 
4
  "all_reduce_scores": {
5
  "0": "SUCCESS",
6
  "1": "SUCCESS",
7
+ "10": "SUCCESS",
8
+ "100": "SUCCESS",
9
+ "101": "SUCCESS",
10
+ "102": "SUCCESS",
11
+ "103": "SUCCESS",
12
+ "104": "SUCCESS",
13
+ "105": "SUCCESS",
14
+ "106": "SUCCESS",
15
+ "107": "SUCCESS",
16
+ "108": "SUCCESS",
17
+ "109": "SUCCESS",
18
+ "11": "SUCCESS",
19
+ "110": "SUCCESS",
20
+ "111": "SUCCESS",
21
+ "112": "SUCCESS",
22
+ "113": "SUCCESS",
23
+ "114": "SUCCESS",
24
+ "115": "SUCCESS",
25
+ "116": "SUCCESS",
26
+ "117": "SUCCESS",
27
+ "118": "SUCCESS",
28
+ "119": "SUCCESS",
29
+ "12": "SUCCESS",
30
+ "120": "SUCCESS",
31
+ "121": "SUCCESS",
32
+ "122": "SUCCESS",
33
+ "123": "SUCCESS",
34
+ "124": "SUCCESS",
35
+ "125": "SUCCESS",
36
+ "126": "SUCCESS",
37
+ "127": "SUCCESS",
38
+ "128": "SUCCESS",
39
+ "129": "SUCCESS",
40
+ "13": "SUCCESS",
41
+ "130": "SUCCESS",
42
+ "131": "SUCCESS",
43
+ "132": "SUCCESS",
44
+ "133": "SUCCESS",
45
+ "134": "SUCCESS",
46
+ "135": "SUCCESS",
47
+ "136": "SUCCESS",
48
+ "137": "SUCCESS",
49
+ "138": "SUCCESS",
50
+ "139": "SUCCESS",
51
+ "14": "SUCCESS",
52
+ "140": "SUCCESS",
53
+ "141": "SUCCESS",
54
+ "142": "SUCCESS",
55
+ "143": "SUCCESS",
56
+ "144": "SUCCESS",
57
+ "145": "SUCCESS",
58
+ "146": "SUCCESS",
59
+ "147": "SUCCESS",
60
+ "148": "SUCCESS",
61
+ "149": "SUCCESS",
62
+ "15": "SUCCESS",
63
+ "150": "SUCCESS",
64
+ "151": "SUCCESS",
65
+ "152": "SUCCESS",
66
+ "153": "SUCCESS",
67
+ "154": "SUCCESS",
68
+ "155": "SUCCESS",
69
+ "156": "SUCCESS",
70
+ "157": "SUCCESS",
71
+ "158": "SUCCESS",
72
+ "159": "SUCCESS",
73
+ "16": "SUCCESS",
74
+ "160": "SUCCESS",
75
+ "161": "SUCCESS",
76
+ "162": "SUCCESS",
77
+ "163": "SUCCESS",
78
+ "164": "SUCCESS",
79
+ "165": "SUCCESS",
80
+ "166": "SUCCESS",
81
+ "167": "SUCCESS",
82
+ "168": "SUCCESS",
83
+ "169": "SUCCESS",
84
+ "17": "SUCCESS",
85
+ "170": "SUCCESS",
86
+ "171": "SUCCESS",
87
+ "172": "SUCCESS",
88
+ "173": "SUCCESS",
89
+ "174": "SUCCESS",
90
+ "175": "SUCCESS",
91
+ "176": "SUCCESS",
92
+ "177": "SUCCESS",
93
+ "178": "SUCCESS",
94
+ "179": "SUCCESS",
95
+ "18": "SUCCESS",
96
+ "180": "SUCCESS",
97
+ "181": "SUCCESS",
98
+ "182": "SUCCESS",
99
+ "183": "SUCCESS",
100
+ "184": "SUCCESS",
101
+ "185": "SUCCESS",
102
+ "186": "SUCCESS",
103
+ "187": "SUCCESS",
104
+ "188": "SUCCESS",
105
+ "189": "SUCCESS",
106
+ "19": "SUCCESS",
107
+ "190": "SUCCESS",
108
+ "191": "SUCCESS",
109
+ "192": "SUCCESS",
110
+ "193": "SUCCESS",
111
+ "194": "SUCCESS",
112
+ "195": "SUCCESS",
113
+ "196": "SUCCESS",
114
+ "197": "SUCCESS",
115
+ "198": "SUCCESS",
116
+ "199": "SUCCESS",
117
  "2": "SUCCESS",
118
+ "20": "SUCCESS",
119
+ "200": "SUCCESS",
120
+ "201": "SUCCESS",
121
+ "202": "SUCCESS",
122
+ "203": "SUCCESS",
123
+ "204": "SUCCESS",
124
+ "205": "SUCCESS",
125
+ "206": "SUCCESS",
126
+ "207": "SUCCESS",
127
+ "208": "SUCCESS",
128
+ "209": "SUCCESS",
129
+ "21": "SUCCESS",
130
+ "210": "SUCCESS",
131
+ "211": "SUCCESS",
132
+ "212": "SUCCESS",
133
+ "213": "SUCCESS",
134
+ "214": "SUCCESS",
135
+ "215": "SUCCESS",
136
+ "216": "SUCCESS",
137
+ "217": "SUCCESS",
138
+ "218": "SUCCESS",
139
+ "219": "SUCCESS",
140
+ "22": "SUCCESS",
141
+ "220": "SUCCESS",
142
+ "221": "SUCCESS",
143
+ "222": "SUCCESS",
144
+ "223": "SUCCESS",
145
+ "224": "SUCCESS",
146
+ "225": "SUCCESS",
147
+ "226": "SUCCESS",
148
+ "227": "SUCCESS",
149
+ "228": "SUCCESS",
150
+ "229": "SUCCESS",
151
+ "23": "SUCCESS",
152
+ "230": "SUCCESS",
153
+ "231": "SUCCESS",
154
+ "232": "SUCCESS",
155
+ "233": "SUCCESS",
156
+ "234": "SUCCESS",
157
+ "235": "SUCCESS",
158
+ "236": "SUCCESS",
159
+ "237": "SUCCESS",
160
+ "238": "SUCCESS",
161
+ "239": "SUCCESS",
162
+ "24": "SUCCESS",
163
+ "240": "SUCCESS",
164
+ "241": "SUCCESS",
165
+ "242": "SUCCESS",
166
+ "243": "SUCCESS",
167
+ "244": "SUCCESS",
168
+ "245": "SUCCESS",
169
+ "246": "SUCCESS",
170
+ "247": "SUCCESS",
171
+ "248": "SUCCESS",
172
+ "249": "SUCCESS",
173
+ "25": "SUCCESS",
174
+ "250": "SUCCESS",
175
+ "251": "SUCCESS",
176
+ "252": "SUCCESS",
177
+ "253": "SUCCESS",
178
+ "254": "SUCCESS",
179
+ "255": "SUCCESS",
180
+ "26": "SUCCESS",
181
+ "27": "SUCCESS",
182
+ "28": "SUCCESS",
183
+ "29": "SUCCESS",
184
+ "3": "SUCCESS",
185
+ "30": "SUCCESS",
186
+ "31": "SUCCESS",
187
+ "32": "SUCCESS",
188
+ "33": "SUCCESS",
189
+ "34": "SUCCESS",
190
+ "35": "SUCCESS",
191
+ "36": "SUCCESS",
192
+ "37": "SUCCESS",
193
+ "38": "SUCCESS",
194
+ "39": "SUCCESS",
195
+ "4": "SUCCESS",
196
+ "40": "SUCCESS",
197
+ "41": "SUCCESS",
198
+ "42": "SUCCESS",
199
+ "43": "SUCCESS",
200
+ "44": "SUCCESS",
201
+ "45": "SUCCESS",
202
+ "46": "SUCCESS",
203
+ "47": "SUCCESS",
204
+ "48": "SUCCESS",
205
+ "49": "SUCCESS",
206
+ "5": "SUCCESS",
207
+ "50": "SUCCESS",
208
+ "51": "SUCCESS",
209
+ "52": "SUCCESS",
210
+ "53": "SUCCESS",
211
+ "54": "SUCCESS",
212
+ "55": "SUCCESS",
213
+ "56": "SUCCESS",
214
+ "57": "SUCCESS",
215
+ "58": "SUCCESS",
216
+ "59": "SUCCESS",
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": "SUCCESS",
235
+ "76": "SUCCESS",
236
+ "77": "SUCCESS",
237
+ "78": "SUCCESS",
238
+ "79": "SUCCESS",
239
+ "8": "SUCCESS",
240
+ "80": "SUCCESS",
241
+ "81": "SUCCESS",
242
+ "82": "SUCCESS",
243
+ "83": "SUCCESS",
244
+ "84": "SUCCESS",
245
+ "85": "SUCCESS",
246
+ "86": "SUCCESS",
247
+ "87": "SUCCESS",
248
+ "88": "SUCCESS",
249
+ "89": "SUCCESS",
250
+ "9": "SUCCESS",
251
+ "90": "SUCCESS",
252
+ "91": "SUCCESS",
253
+ "92": "SUCCESS",
254
+ "93": "SUCCESS",
255
+ "94": "SUCCESS",
256
+ "95": "SUCCESS",
257
+ "96": "SUCCESS",
258
+ "97": "SUCCESS",
259
+ "98": "SUCCESS",
260
+ "99": "SUCCESS"
261
  },
262
  "architectures": [
263
  "GPTOptim"
 
268
  "AutoModelForCausalLM": "distributed/optimized-gpt2-500m--modeling_gpt_optimized.GPTOptim"
269
  },
270
  "block_list": [
271
+ 5048291,
272
+ 5048294,
273
+ 5048297,
274
+ 5048304,
275
+ 5048307
276
  ],
277
  "block_size": 1024,
278
  "bos_token_id": 50256,
279
  "embd_pdrop": 0.1,
280
  "eos_token_id": 50256,
281
  "initializer_range": 0.02,
282
+ "inner_step": 2,
283
  "inner_steps": 0,
284
+ "last_allreduce_block": 4016775,
285
  "layer_norm_epsilon": 1e-05,
286
  "model_type": "gpt_optimized",
287
  "n_embd": 1280,
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6a2c3c325d500575e01ed856c21131ead8e7298a5d6f10267421b7e99afaa364
3
  size 4040701744
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:188d0fa688dc204878bb8535c61e5784d3160145cd8129454abd1891618cb7dd
3
  size 4040701744