IlyasMoutawwakil HF staff commited on
Commit
a5a705e
·
verified ·
1 Parent(s): 50300cb

Upload cuda_training_transformers_fill-mask_google-bert/bert-base-uncased/benchmark.json with huggingface_hub

Browse files
cuda_training_transformers_fill-mask_google-bert/bert-base-uncased/benchmark.json CHANGED
@@ -3,7 +3,7 @@
3
  "name": "cuda_training_transformers_fill-mask_google-bert/bert-base-uncased",
4
  "backend": {
5
  "name": "pytorch",
6
- "version": "2.3.1+rocm5.7",
7
  "_target_": "optimum_benchmark.backends.pytorch.backend.PyTorchBackend",
8
  "task": "fill-mask",
9
  "library": "transformers",
@@ -11,7 +11,7 @@
11
  "model": "google-bert/bert-base-uncased",
12
  "processor": "google-bert/bert-base-uncased",
13
  "device": "cuda",
14
- "device_ids": "5",
15
  "seed": 42,
16
  "inter_op_num_threads": null,
17
  "intra_op_num_threads": null,
@@ -117,33 +117,33 @@
117
  "overall": {
118
  "memory": {
119
  "unit": "MB",
120
- "max_ram": 1282.10944,
121
  "max_global_vram": 68702.69952,
122
- "max_process_vram": 305992.740864,
123
  "max_reserved": 2497.708032,
124
  "max_allocated": 2195.345408
125
  },
126
  "latency": {
127
  "unit": "s",
128
  "count": 5,
129
- "total": 0.703011920928955,
130
- "mean": 0.14060238418579102,
131
- "stdev": 0.1981068943026263,
132
- "p50": 0.041936981201171876,
133
- "p90": 0.33906457977294924,
134
- "p95": 0.43793948593139637,
135
- "p99": 0.5170394108581542,
136
  "values": [
137
- 0.5368143920898437,
138
- 0.04086610412597656,
139
- 0.04095458221435547,
140
- 0.04243986129760742,
141
- 0.041936981201171876
142
  ]
143
  },
144
  "throughput": {
145
  "unit": "samples/s",
146
- "value": 71.12254929323296
147
  },
148
  "energy": null,
149
  "efficiency": null
@@ -151,30 +151,30 @@
151
  "warmup": {
152
  "memory": {
153
  "unit": "MB",
154
- "max_ram": 1282.10944,
155
  "max_global_vram": 68702.69952,
156
- "max_process_vram": 305992.740864,
157
  "max_reserved": 2497.708032,
158
  "max_allocated": 2195.345408
159
  },
160
  "latency": {
161
  "unit": "s",
162
  "count": 2,
163
- "total": 0.5776804962158203,
164
- "mean": 0.28884024810791015,
165
- "stdev": 0.24797414398193357,
166
- "p50": 0.28884024810791015,
167
- "p90": 0.487219563293457,
168
- "p95": 0.5120169776916503,
169
- "p99": 0.5318549092102051,
170
  "values": [
171
- 0.5368143920898437,
172
- 0.04086610412597656
173
  ]
174
  },
175
  "throughput": {
176
  "unit": "samples/s",
177
- "value": 13.848485542449776
178
  },
179
  "energy": null,
180
  "efficiency": null
@@ -182,31 +182,31 @@
182
  "train": {
183
  "memory": {
184
  "unit": "MB",
185
- "max_ram": 1282.10944,
186
  "max_global_vram": 68702.69952,
187
- "max_process_vram": 305992.740864,
188
  "max_reserved": 2497.708032,
189
  "max_allocated": 2195.345408
190
  },
191
  "latency": {
192
  "unit": "s",
193
  "count": 3,
194
- "total": 0.12533142471313477,
195
- "mean": 0.04177714157104492,
196
- "stdev": 0.0006168063009328952,
197
- "p50": 0.041936981201171876,
198
- "p90": 0.04233928527832031,
199
- "p95": 0.04238957328796387,
200
- "p99": 0.04242980369567871,
201
  "values": [
202
- 0.04095458221435547,
203
- 0.04243986129760742,
204
- 0.041936981201171876
205
  ]
206
  },
207
  "throughput": {
208
  "unit": "samples/s",
209
- "value": 143.6192083605477
210
  },
211
  "energy": null,
212
  "efficiency": null
 
3
  "name": "cuda_training_transformers_fill-mask_google-bert/bert-base-uncased",
4
  "backend": {
5
  "name": "pytorch",
6
+ "version": "2.2.0.dev20231010+rocm5.7",
7
  "_target_": "optimum_benchmark.backends.pytorch.backend.PyTorchBackend",
8
  "task": "fill-mask",
9
  "library": "transformers",
 
11
  "model": "google-bert/bert-base-uncased",
12
  "processor": "google-bert/bert-base-uncased",
13
  "device": "cuda",
14
+ "device_ids": "6",
15
  "seed": 42,
16
  "inter_op_num_threads": null,
17
  "intra_op_num_threads": null,
 
117
  "overall": {
118
  "memory": {
119
  "unit": "MB",
120
+ "max_ram": 1203.769344,
121
  "max_global_vram": 68702.69952,
122
+ "max_process_vram": 287044.284416,
123
  "max_reserved": 2497.708032,
124
  "max_allocated": 2195.345408
125
  },
126
  "latency": {
127
  "unit": "s",
128
  "count": 5,
129
+ "total": 0.7220148658752441,
130
+ "mean": 0.14440297317504883,
131
+ "stdev": 0.20283927152458436,
132
+ "p50": 0.0409623031616211,
133
+ "p90": 0.3501985549926758,
134
+ "p95": 0.4501030616760253,
135
+ "p99": 0.5300266670227051,
136
  "values": [
137
+ 0.550007568359375,
138
+ 0.050485034942626954,
139
+ 0.0409623031616211,
140
+ 0.0403690185546875,
141
+ 0.040190940856933595
142
  ]
143
  },
144
  "throughput": {
145
  "unit": "samples/s",
146
+ "value": 69.250651701456
147
  },
148
  "energy": null,
149
  "efficiency": null
 
151
  "warmup": {
152
  "memory": {
153
  "unit": "MB",
154
+ "max_ram": 1203.769344,
155
  "max_global_vram": 68702.69952,
156
+ "max_process_vram": 287044.284416,
157
  "max_reserved": 2497.708032,
158
  "max_allocated": 2195.345408
159
  },
160
  "latency": {
161
  "unit": "s",
162
  "count": 2,
163
+ "total": 0.6004926033020019,
164
+ "mean": 0.30024630165100097,
165
+ "stdev": 0.24976126670837404,
166
+ "p50": 0.30024630165100097,
167
+ "p90": 0.5000553150177003,
168
+ "p95": 0.5250314416885377,
169
+ "p99": 0.5450123430252075,
170
  "values": [
171
+ 0.550007568359375,
172
+ 0.050485034942626954
173
  ]
174
  },
175
  "throughput": {
176
  "unit": "samples/s",
177
+ "value": 13.322395573250068
178
  },
179
  "energy": null,
180
  "efficiency": null
 
182
  "train": {
183
  "memory": {
184
  "unit": "MB",
185
+ "max_ram": 1203.769344,
186
  "max_global_vram": 68702.69952,
187
+ "max_process_vram": 287044.284416,
188
  "max_reserved": 2497.708032,
189
  "max_allocated": 2195.345408
190
  },
191
  "latency": {
192
  "unit": "s",
193
  "count": 3,
194
+ "total": 0.1215222625732422,
195
+ "mean": 0.040507420857747396,
196
+ "stdev": 0.0003297639048293156,
197
+ "p50": 0.0403690185546875,
198
+ "p90": 0.04084364624023438,
199
+ "p95": 0.040902974700927734,
200
+ "p99": 0.040950437469482424,
201
  "values": [
202
+ 0.0409623031616211,
203
+ 0.0403690185546875,
204
+ 0.040190940856933595
205
  ]
206
  },
207
  "throughput": {
208
  "unit": "samples/s",
209
+ "value": 148.12100777955226
210
  },
211
  "energy": null,
212
  "efficiency": null