Hugging Face's logo

Datasets:
optimum-benchmark
/
cpu

Dataset card Viewer Files Community
IlyasMoutawwakil HF staff commited on
Commit
a832a5e
•
1 Parent(s): f955fa6

Upload cpu_inference_transformers_text-generation_openai-community/gpt2/benchmark.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. cpu_inference_transformers_text-generation_openai-community/gpt2/benchmark.json +69 -93
cpu_inference_transformers_text-generation_openai-community/gpt2/benchmark.json CHANGED
@@ -44,8 +44,8 @@
44
  "duration": 1,
45
  "warmup_runs": 1,
46
  "input_shapes": {
47
- "batch_size": 1,
48
- "sequence_length": 2
49
  },
50
  "new_tokens": null,
51
  "memory": true,
@@ -79,7 +79,7 @@
79
  "processor": "x86_64",
80
  "python_version": "3.10.15",
81
  "optimum_benchmark_version": "0.5.0.dev0",
82
- "optimum_benchmark_commit": "6807ba28334c8c98abf72a03d78f86133328d180",
83
  "transformers_version": "4.46.3",
84
  "transformers_commit": null,
85
  "accelerate_version": "1.1.1",
@@ -100,7 +100,7 @@
100
  "load": {
101
  "memory": {
102
  "unit": "MB",
103
- "max_ram": 1130.385408,
104
  "max_global_vram": null,
105
  "max_process_vram": null,
106
  "max_reserved": null,
@@ -109,15 +109,15 @@
109
  "latency": {
110
  "unit": "s",
111
  "values": [
112
- 1.7419912789999898
113
  ],
114
  "count": 1,
115
- "total": 1.7419912789999898,
116
- "mean": 1.7419912789999898,
117
- "p50": 1.7419912789999898,
118
- "p90": 1.7419912789999898,
119
- "p95": 1.7419912789999898,
120
- "p99": 1.7419912789999898,
121
  "stdev": 0,
122
  "stdev_": 0
123
  },
@@ -128,7 +128,7 @@
128
  "prefill": {
129
  "memory": {
130
  "unit": "MB",
131
- "max_ram": 985.68192,
132
  "max_global_vram": null,
133
  "max_process_vram": null,
134
  "max_reserved": null,
@@ -137,51 +137,43 @@
137
  "latency": {
138
  "unit": "s",
139
  "values": [
140
- 0.047665391999998974,
141
- 0.049731436999991274,
142
- 0.04895271399999501,
143
- 0.04833657500000754,
144
- 0.04391997499999434,
145
- 0.043644501000017044,
146
- 0.04382332400001587,
147
- 0.04319213599995919,
148
- 0.04416296799996644,
149
- 0.044141969000008885,
150
- 0.04314894600003072,
151
- 0.04354990399997405,
152
- 0.043227575999992496,
153
- 0.04241360399998939
154
  ],
155
- "count": 14,
156
- "total": 0.6299110209999412,
157
- "mean": 0.04499364435713866,
158
- "p50": 0.043871649500005105,
159
- "p90": 0.04876787229999877,
160
- "p95": 0.04922526704999371,
161
- "p99": 0.04963020300999176,
162
- "stdev": 0.0024009023029775032,
163
- "stdev_": 5.336092102076141
164
  },
165
  "throughput": {
166
  "unit": "tokens/s",
167
- "value": 44.45072250927105
168
  },
169
  "energy": {
170
  "unit": "kWh",
171
- "cpu": 1.6854151983796455e-06,
172
- "ram": 7.043250781543132e-08,
173
  "gpu": 0.0,
174
- "total": 1.7558477061950768e-06
175
  },
176
  "efficiency": {
177
  "unit": "tokens/kWh",
178
- "value": 1139050.9512547653
179
  }
180
  },
181
  "decode": {
182
  "memory": {
183
  "unit": "MB",
184
- "max_ram": 985.68192,
185
  "max_global_vram": null,
186
  "max_process_vram": null,
187
  "max_reserved": null,
@@ -190,45 +182,37 @@
190
  "latency": {
191
  "unit": "s",
192
  "values": [
193
- 0.030765811999970083,
194
- 0.030026732999999695,
195
- 0.03023100500001874,
196
- 0.02618368299999929,
197
- 0.026192708999985825,
198
- 0.026696138000033898,
199
- 0.026371824000023025,
200
- 0.026651877000006152,
201
- 0.026894829000013942,
202
- 0.026896673000010196,
203
- 0.027024220999976478,
204
- 0.02588880300004348,
205
- 0.026431090000016866,
206
- 0.025466253999979926
207
  ],
208
- "count": 14,
209
- "total": 0.3817216510000776,
210
- "mean": 0.02726583221429126,
211
- "p50": 0.026674007500020025,
212
- "p90": 0.030169723400013025,
213
- "p95": 0.03041818745000171,
214
- "p99": 0.030696287089976407,
215
- "stdev": 0.0016615794659438356,
216
- "stdev_": 6.093998719294277
217
  },
218
  "throughput": {
219
  "unit": "tokens/s",
220
- "value": 36.67593903390393
221
  },
222
  "energy": {
223
  "unit": "kWh",
224
- "cpu": 1.0257383568054851e-06,
225
- "ram": 4.287120110748343e-08,
226
  "gpu": 0.0,
227
- "total": 1.068609557912968e-06
228
  },
229
  "efficiency": {
230
  "unit": "tokens/kWh",
231
- "value": 935795.4854465602
232
  }
233
  },
234
  "per_token": {
@@ -236,34 +220,26 @@
236
  "latency": {
237
  "unit": "s",
238
  "values": [
239
- 0.030299631999980647,
240
- 0.029594156000030125,
241
- 0.029819386999974995,
242
- 0.02586457799998243,
243
- 0.025855570000032913,
244
- 0.02636103300000059,
245
- 0.026048771000034776,
246
- 0.02631907600004979,
247
- 0.02656132700002445,
248
- 0.02656453299999839,
249
- 0.02669703099996923,
250
- 0.02555920800000422,
251
- 0.026109781000002386,
252
- 0.02515429199996788
253
  ],
254
- "count": 14,
255
- "total": 0.3768083750000528,
256
- "mean": 0.0269148839285752,
257
- "p50": 0.02634005450002519,
258
- "p90": 0.029751817699991533,
259
- "p95": 0.029987472749976974,
260
- "p99": 0.030237200149979913,
261
- "stdev": 0.001616638367978842,
262
- "stdev_": 6.006484636043616
263
  },
264
  "throughput": {
265
  "unit": "tokens/s",
266
- "value": 37.154163571863386
267
  },
268
  "energy": null,
269
  "efficiency": null
 
44
  "duration": 1,
45
  "warmup_runs": 1,
46
  "input_shapes": {
47
+ "batch_size": 2,
48
+ "sequence_length": 16
49
  },
50
  "new_tokens": null,
51
  "memory": true,
 
79
  "processor": "x86_64",
80
  "python_version": "3.10.15",
81
  "optimum_benchmark_version": "0.5.0.dev0",
82
+ "optimum_benchmark_commit": "04bb8676c2e912cf262caaa2d4796a0191a83273",
83
  "transformers_version": "4.46.3",
84
  "transformers_commit": null,
85
  "accelerate_version": "1.1.1",
 
100
  "load": {
101
  "memory": {
102
  "unit": "MB",
103
+ "max_ram": 1130.233856,
104
  "max_global_vram": null,
105
  "max_process_vram": null,
106
  "max_reserved": null,
 
109
  "latency": {
110
  "unit": "s",
111
  "values": [
112
+ 1.7327888100000166
113
  ],
114
  "count": 1,
115
+ "total": 1.7327888100000166,
116
+ "mean": 1.7327888100000166,
117
+ "p50": 1.7327888100000166,
118
+ "p90": 1.7327888100000166,
119
+ "p95": 1.7327888100000166,
120
+ "p99": 1.7327888100000166,
121
  "stdev": 0,
122
  "stdev_": 0
123
  },
 
128
  "prefill": {
129
  "memory": {
130
  "unit": "MB",
131
+ "max_ram": 1012.641792,
132
  "max_global_vram": null,
133
  "max_process_vram": null,
134
  "max_reserved": null,
 
137
  "latency": {
138
  "unit": "s",
139
  "values": [
140
+ 0.15366975499998148,
141
+ 0.16072087899999588,
142
+ 0.11341367500000388,
143
+ 0.10321070399999144,
144
+ 0.10557986899999605,
145
+ 0.10479721300001188
 
 
 
 
 
 
 
 
146
  ],
147
+ "count": 6,
148
+ "total": 0.7413920949999806,
149
+ "mean": 0.12356534916666344,
150
+ "p50": 0.10949677199999996,
151
+ "p90": 0.15719531699998868,
152
+ "p95": 0.15895809799999228,
153
+ "p99": 0.16036832279999516,
154
+ "stdev": 0.024082833263604584,
155
+ "stdev_": 19.489956873849763
156
  },
157
  "throughput": {
158
  "unit": "tokens/s",
159
+ "value": 258.9722783596782
160
  },
161
  "energy": {
162
  "unit": "kWh",
163
+ "cpu": 4.218644694444484e-06,
164
+ "ram": 1.7630452732304486e-07,
165
  "gpu": 0.0,
166
+ "total": 4.3949492217675285e-06
167
  },
168
  "efficiency": {
169
  "unit": "tokens/kWh",
170
+ "value": 7281085.260668945
171
  }
172
  },
173
  "decode": {
174
  "memory": {
175
  "unit": "MB",
176
+ "max_ram": 1020.64128,
177
  "max_global_vram": null,
178
  "max_process_vram": null,
179
  "max_reserved": null,
 
182
  "latency": {
183
  "unit": "s",
184
  "values": [
185
+ 0.05179725900001131,
186
+ 0.052174370000017234,
187
+ 0.04633807199999751,
188
+ 0.044985338000003594,
189
+ 0.04457143900000915,
190
+ 0.04629236700000661
 
 
 
 
 
 
 
 
191
  ],
192
+ "count": 6,
193
+ "total": 0.2861588450000454,
194
+ "mean": 0.0476931408333409,
195
+ "p50": 0.04631521950000206,
196
+ "p90": 0.05198581450001427,
197
+ "p95": 0.05208009225001575,
198
+ "p99": 0.05215551445001694,
199
+ "stdev": 0.0031037825891959037,
200
+ "stdev_": 6.507817549785145
201
  },
202
  "throughput": {
203
  "unit": "tokens/s",
204
+ "value": 41.93475130918318
205
  },
206
  "energy": {
207
  "unit": "kWh",
208
+ "cpu": 1.6928337666666454e-06,
209
+ "ram": 7.074798652428004e-08,
210
  "gpu": 0.0,
211
+ "total": 1.7635817531909286e-06
212
  },
213
  "efficiency": {
214
  "unit": "tokens/kWh",
215
+ "value": 1134055.7342359147
216
  }
217
  },
218
  "per_token": {
 
220
  "latency": {
221
  "unit": "s",
222
  "values": [
223
+ 0.05067520299999728,
224
+ 0.05078066000001513,
225
+ 0.04527658899999665,
226
+ 0.043886185000019395,
227
+ 0.043664963999987094,
228
+ 0.04519753200000309
 
 
 
 
 
 
 
 
229
  ],
230
+ "count": 6,
231
+ "total": 0.27948113300001864,
232
+ "mean": 0.04658018883333644,
233
+ "p50": 0.04523706049999987,
234
+ "p90": 0.05072793150000621,
235
+ "p95": 0.05075429575001067,
236
+ "p99": 0.05077538715001424,
237
+ "stdev": 0.002993893896316374,
238
+ "stdev_": 6.427397508044683
239
  },
240
  "throughput": {
241
  "unit": "tokens/s",
242
+ "value": 42.93670871872127
243
  },
244
  "energy": null,
245
  "efficiency": null