hf-transformers-bot commited on
Commit
617ad9e
·
verified ·
1 Parent(s): bd21c69

Upload benchmark results for run 18319664876

Browse files
2025-10-07/18319664876/benchmark_results/Llama-2-7b-hf/Llama-2-7b-hf_benchmark_20251007_164938.json ADDED
@@ -0,0 +1,1175 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_name": "Llama-2-7b-hf",
3
+ "benchmark_scenarios": [
4
+ {
5
+ "scenario_name": "eager_eager_attn",
6
+ "metadata": {
7
+ "timestamp": "2025-10-07T16:43:35.928056",
8
+ "commit_id": "50090c3fc82e1e0a06b4da366ea2fb6055d529e9",
9
+ "hardware_info": {
10
+ "gpu_name": "NVIDIA A10G",
11
+ "gpu_memory_total_mb": 23028,
12
+ "cpu_count": 16,
13
+ "memory_total_mb": 63607,
14
+ "python_version": "3.10.12",
15
+ "torch_version": "2.8.0+cu126",
16
+ "cuda_version": "12.6"
17
+ },
18
+ "config": {
19
+ "name": "eager",
20
+ "model_id": "meta-llama/Llama-2-7b-hf",
21
+ "variant": "eager",
22
+ "warmup_iterations": 3,
23
+ "measurement_iterations": 5,
24
+ "num_tokens_to_generate": 100,
25
+ "device": "cuda",
26
+ "torch_dtype": "float16",
27
+ "compile_mode": null,
28
+ "compile_options": {},
29
+ "use_cache": true,
30
+ "batch_size": 1,
31
+ "sequence_length": null,
32
+ "attn_implementation": "eager",
33
+ "sdpa_backend": null,
34
+ "custom_params": {}
35
+ }
36
+ },
37
+ "measurements": {
38
+ "latency_seconds": {
39
+ "name": "latency_seconds",
40
+ "measurements": [
41
+ 3.682453857421875,
42
+ 3.668287841796875,
43
+ 3.65483349609375,
44
+ 3.657760986328125,
45
+ 3.6610390625
46
+ ],
47
+ "mean": 3.6648750488281245,
48
+ "median": 3.6610390625,
49
+ "std": 0.009867310931595803,
50
+ "min": 3.65483349609375,
51
+ "max": 3.682453857421875,
52
+ "p25": 3.657760986328125,
53
+ "p75": 3.668287841796875,
54
+ "p90": 3.676787451171875,
55
+ "p95": 3.6796206542968752,
56
+ "p99": 3.6818872167968753,
57
+ "unit": "seconds"
58
+ },
59
+ "time_to_first_token_seconds": {
60
+ "name": "time_to_first_token_seconds",
61
+ "measurements": [
62
+ 0.03904627227783203,
63
+ 0.037419231414794925,
64
+ 0.037423583984375,
65
+ 0.03850604629516602,
66
+ 0.03713216018676758
67
+ ],
68
+ "mean": 0.03790545883178711,
69
+ "median": 0.037423583984375,
70
+ "std": 0.0007387519047470771,
71
+ "min": 0.03713216018676758,
72
+ "max": 0.03904627227783203,
73
+ "p25": 0.037419231414794925,
74
+ "p75": 0.03850604629516602,
75
+ "p90": 0.03883018188476563,
76
+ "p95": 0.03893822708129883,
77
+ "p99": 0.03902466323852539,
78
+ "unit": "seconds"
79
+ },
80
+ "tokens_per_second": {
81
+ "name": "tokens_per_second",
82
+ "measurements": [
83
+ 27.155805305870434,
84
+ 27.260674274409165,
85
+ 27.361027556215355,
86
+ 27.339129148617737,
87
+ 27.31464982832316
88
+ ],
89
+ "mean": 27.28625722268717,
90
+ "median": 27.31464982832316,
91
+ "std": 0.07329989526060698,
92
+ "min": 27.155805305870434,
93
+ "max": 27.361027556215355,
94
+ "p25": 27.260674274409165,
95
+ "p75": 27.339129148617737,
96
+ "p90": 27.352268193176307,
97
+ "p95": 27.35664787469583,
98
+ "p99": 27.36015161991145,
99
+ "unit": "tokens/sec"
100
+ },
101
+ "time_per_output_token_seconds": {
102
+ "name": "time_per_output_token_seconds",
103
+ "measurements": [
104
+ 0.03682453857421875,
105
+ 0.03668287841796875,
106
+ 0.0365483349609375,
107
+ 0.03657760986328125,
108
+ 0.036610390625
109
+ ],
110
+ "mean": 0.03664875048828124,
111
+ "median": 0.036610390625,
112
+ "std": 9.867310931595953e-05,
113
+ "min": 0.0365483349609375,
114
+ "max": 0.03682453857421875,
115
+ "p25": 0.03657760986328125,
116
+ "p75": 0.03668287841796875,
117
+ "p90": 0.03676787451171875,
118
+ "p95": 0.03679620654296875,
119
+ "p99": 0.03681887216796875,
120
+ "unit": "seconds/token"
121
+ }
122
+ },
123
+ "gpu_metrics": {
124
+ "gpu_utilization_mean": 90.71739130434783,
125
+ "gpu_utilization_max": 94,
126
+ "gpu_utilization_min": 86,
127
+ "gpu_memory_used_mean": 13253,
128
+ "gpu_memory_used_max": 13253,
129
+ "gpu_memory_used_min": 13253,
130
+ "sample_count": 92,
131
+ "gpu_monitoring_status": "success"
132
+ }
133
+ },
134
+ {
135
+ "scenario_name": "compiled_compile_max-autotune_eager_attn",
136
+ "metadata": {
137
+ "timestamp": "2025-10-07T16:44:11.403417",
138
+ "commit_id": "50090c3fc82e1e0a06b4da366ea2fb6055d529e9",
139
+ "hardware_info": {
140
+ "gpu_name": "NVIDIA A10G",
141
+ "gpu_memory_total_mb": 23028,
142
+ "cpu_count": 16,
143
+ "memory_total_mb": 63607,
144
+ "python_version": "3.10.12",
145
+ "torch_version": "2.8.0+cu126",
146
+ "cuda_version": "12.6"
147
+ },
148
+ "config": {
149
+ "name": "compiled",
150
+ "model_id": "meta-llama/Llama-2-7b-hf",
151
+ "variant": "compiled",
152
+ "warmup_iterations": 3,
153
+ "measurement_iterations": 5,
154
+ "num_tokens_to_generate": 100,
155
+ "device": "cuda",
156
+ "torch_dtype": "float16",
157
+ "compile_mode": "max-autotune",
158
+ "compile_options": {},
159
+ "use_cache": true,
160
+ "batch_size": 1,
161
+ "sequence_length": null,
162
+ "attn_implementation": "eager",
163
+ "sdpa_backend": null,
164
+ "custom_params": {}
165
+ }
166
+ },
167
+ "measurements": {
168
+ "latency_seconds": {
169
+ "name": "latency_seconds",
170
+ "measurements": [
171
+ 7.81532958984375,
172
+ 7.8352919921875,
173
+ 7.8824296875,
174
+ 7.896419921875,
175
+ 7.91605615234375
176
+ ],
177
+ "mean": 7.86910546875,
178
+ "median": 7.8824296875,
179
+ "std": 0.03785017092198803,
180
+ "min": 7.81532958984375,
181
+ "max": 7.91605615234375,
182
+ "p25": 7.8352919921875,
183
+ "p75": 7.896419921875,
184
+ "p90": 7.908201660156251,
185
+ "p95": 7.91212890625,
186
+ "p99": 7.9152707031250005,
187
+ "unit": "seconds"
188
+ },
189
+ "time_to_first_token_seconds": {
190
+ "name": "time_to_first_token_seconds",
191
+ "measurements": [
192
+ 0.03947999954223633,
193
+ 0.03826505661010742,
194
+ 0.03817292785644531,
195
+ 0.03854867172241211,
196
+ 0.04225136184692383
197
+ ],
198
+ "mean": 0.039343603515625,
199
+ "median": 0.03854867172241211,
200
+ "std": 0.0015257341755083762,
201
+ "min": 0.03817292785644531,
202
+ "max": 0.04225136184692383,
203
+ "p25": 0.03826505661010742,
204
+ "p75": 0.03947999954223633,
205
+ "p90": 0.04114281692504883,
206
+ "p95": 0.04169708938598633,
207
+ "p99": 0.04214050735473633,
208
+ "unit": "seconds"
209
+ },
210
+ "tokens_per_second": {
211
+ "name": "tokens_per_second",
212
+ "measurements": [
213
+ 12.79536567849332,
214
+ 12.762766224884677,
215
+ 12.686443642951938,
216
+ 12.663966834258108,
217
+ 12.632553139531792
218
+ ],
219
+ "mean": 12.708219104023968,
220
+ "median": 12.686443642951938,
221
+ "std": 0.061200794280460755,
222
+ "min": 12.632553139531792,
223
+ "max": 12.79536567849332,
224
+ "p25": 12.663966834258108,
225
+ "p75": 12.762766224884677,
226
+ "p90": 12.782325897049864,
227
+ "p95": 12.788845787771592,
228
+ "p99": 12.794061700348974,
229
+ "unit": "tokens/sec"
230
+ },
231
+ "time_per_output_token_seconds": {
232
+ "name": "time_per_output_token_seconds",
233
+ "measurements": [
234
+ 0.0781532958984375,
235
+ 0.078352919921875,
236
+ 0.078824296875,
237
+ 0.07896419921875,
238
+ 0.0791605615234375
239
+ ],
240
+ "mean": 0.0786910546875,
241
+ "median": 0.078824296875,
242
+ "std": 0.0003785017092198808,
243
+ "min": 0.0781532958984375,
244
+ "max": 0.0791605615234375,
245
+ "p25": 0.078352919921875,
246
+ "p75": 0.07896419921875,
247
+ "p90": 0.0790820166015625,
248
+ "p95": 0.0791212890625,
249
+ "p99": 0.07915270703125,
250
+ "unit": "seconds/token"
251
+ }
252
+ },
253
+ "gpu_metrics": {
254
+ "gpu_utilization_mean": 43.65882352941176,
255
+ "gpu_utilization_max": 91,
256
+ "gpu_utilization_min": 0,
257
+ "gpu_memory_used_mean": 13629.505882352942,
258
+ "gpu_memory_used_max": 13777,
259
+ "gpu_memory_used_min": 13415,
260
+ "sample_count": 170,
261
+ "gpu_monitoring_status": "success"
262
+ }
263
+ },
264
+ {
265
+ "scenario_name": "eager_sdpa_default",
266
+ "metadata": {
267
+ "timestamp": "2025-10-07T16:45:39.408793",
268
+ "commit_id": "50090c3fc82e1e0a06b4da366ea2fb6055d529e9",
269
+ "hardware_info": {
270
+ "gpu_name": "NVIDIA A10G",
271
+ "gpu_memory_total_mb": 23028,
272
+ "cpu_count": 16,
273
+ "memory_total_mb": 63607,
274
+ "python_version": "3.10.12",
275
+ "torch_version": "2.8.0+cu126",
276
+ "cuda_version": "12.6"
277
+ },
278
+ "config": {
279
+ "name": "eager",
280
+ "model_id": "meta-llama/Llama-2-7b-hf",
281
+ "variant": "eager",
282
+ "warmup_iterations": 3,
283
+ "measurement_iterations": 5,
284
+ "num_tokens_to_generate": 100,
285
+ "device": "cuda",
286
+ "torch_dtype": "float16",
287
+ "compile_mode": null,
288
+ "compile_options": {},
289
+ "use_cache": true,
290
+ "batch_size": 1,
291
+ "sequence_length": null,
292
+ "attn_implementation": "sdpa",
293
+ "sdpa_backend": null,
294
+ "custom_params": {}
295
+ }
296
+ },
297
+ "measurements": {
298
+ "latency_seconds": {
299
+ "name": "latency_seconds",
300
+ "measurements": [
301
+ 3.41564697265625,
302
+ 3.41465869140625,
303
+ 3.415184814453125,
304
+ 3.41537744140625,
305
+ 3.4139794921875
306
+ ],
307
+ "mean": 3.414969482421875,
308
+ "median": 3.415184814453125,
309
+ "std": 0.0005913481468372594,
310
+ "min": 3.4139794921875,
311
+ "max": 3.41564697265625,
312
+ "p25": 3.41465869140625,
313
+ "p75": 3.41537744140625,
314
+ "p90": 3.41553916015625,
315
+ "p95": 3.41559306640625,
316
+ "p99": 3.41563619140625,
317
+ "unit": "seconds"
318
+ },
319
+ "time_to_first_token_seconds": {
320
+ "name": "time_to_first_token_seconds",
321
+ "measurements": [
322
+ 0.035205440521240236,
323
+ 0.034635616302490235,
324
+ 0.03470412826538086,
325
+ 0.03464031982421875,
326
+ 0.03463283157348633
327
+ ],
328
+ "mean": 0.034763667297363285,
329
+ "median": 0.03464031982421875,
330
+ "std": 0.00022245814020869866,
331
+ "min": 0.03463283157348633,
332
+ "max": 0.035205440521240236,
333
+ "p25": 0.034635616302490235,
334
+ "p75": 0.03470412826538086,
335
+ "p90": 0.03500491561889649,
336
+ "p95": 0.03510517807006836,
337
+ "p99": 0.03518538803100586,
338
+ "unit": "seconds"
339
+ },
340
+ "tokens_per_second": {
341
+ "name": "tokens_per_second",
342
+ "measurements": [
343
+ 29.277030325599746,
344
+ 29.285503775727953,
345
+ 29.280992225310374,
346
+ 29.27934078021723,
347
+ 29.2913300237563
348
+ ],
349
+ "mean": 29.28283942612232,
350
+ "median": 29.280992225310374,
351
+ "std": 0.005071239096974063,
352
+ "min": 29.277030325599746,
353
+ "max": 29.2913300237563,
354
+ "p25": 29.27934078021723,
355
+ "p75": 29.285503775727953,
356
+ "p90": 29.28899952454496,
357
+ "p95": 29.29016477415063,
358
+ "p99": 29.291096973835163,
359
+ "unit": "tokens/sec"
360
+ },
361
+ "time_per_output_token_seconds": {
362
+ "name": "time_per_output_token_seconds",
363
+ "measurements": [
364
+ 0.0341564697265625,
365
+ 0.0341465869140625,
366
+ 0.03415184814453125,
367
+ 0.034153774414062504,
368
+ 0.034139794921875
369
+ ],
370
+ "mean": 0.03414969482421875,
371
+ "median": 0.03415184814453125,
372
+ "std": 5.91348146837227e-06,
373
+ "min": 0.034139794921875,
374
+ "max": 0.0341564697265625,
375
+ "p25": 0.0341465869140625,
376
+ "p75": 0.034153774414062504,
377
+ "p90": 0.0341553916015625,
378
+ "p95": 0.0341559306640625,
379
+ "p99": 0.0341563619140625,
380
+ "unit": "seconds/token"
381
+ }
382
+ },
383
+ "gpu_metrics": {
384
+ "gpu_utilization_mean": 96.84883720930233,
385
+ "gpu_utilization_max": 98,
386
+ "gpu_utilization_min": 95,
387
+ "gpu_memory_used_mean": 13849,
388
+ "gpu_memory_used_max": 13849,
389
+ "gpu_memory_used_min": 13849,
390
+ "sample_count": 86,
391
+ "gpu_monitoring_status": "success"
392
+ }
393
+ },
394
+ {
395
+ "scenario_name": "eager_sdpa_math",
396
+ "metadata": {
397
+ "timestamp": "2025-10-07T16:46:12.013443",
398
+ "commit_id": "50090c3fc82e1e0a06b4da366ea2fb6055d529e9",
399
+ "hardware_info": {
400
+ "gpu_name": "NVIDIA A10G",
401
+ "gpu_memory_total_mb": 23028,
402
+ "cpu_count": 16,
403
+ "memory_total_mb": 63607,
404
+ "python_version": "3.10.12",
405
+ "torch_version": "2.8.0+cu126",
406
+ "cuda_version": "12.6"
407
+ },
408
+ "config": {
409
+ "name": "eager",
410
+ "model_id": "meta-llama/Llama-2-7b-hf",
411
+ "variant": "eager",
412
+ "warmup_iterations": 3,
413
+ "measurement_iterations": 5,
414
+ "num_tokens_to_generate": 100,
415
+ "device": "cuda",
416
+ "torch_dtype": "float16",
417
+ "compile_mode": null,
418
+ "compile_options": {},
419
+ "use_cache": true,
420
+ "batch_size": 1,
421
+ "sequence_length": null,
422
+ "attn_implementation": "sdpa",
423
+ "sdpa_backend": "math",
424
+ "custom_params": {}
425
+ }
426
+ },
427
+ "measurements": {
428
+ "latency_seconds": {
429
+ "name": "latency_seconds",
430
+ "measurements": [
431
+ 3.579625,
432
+ 3.60485791015625,
433
+ 3.562582275390625,
434
+ 3.537706298828125,
435
+ 3.54537744140625
436
+ ],
437
+ "mean": 3.5660297851562497,
438
+ "median": 3.562582275390625,
439
+ "std": 0.024220147239003613,
440
+ "min": 3.537706298828125,
441
+ "max": 3.60485791015625,
442
+ "p25": 3.54537744140625,
443
+ "p75": 3.579625,
444
+ "p90": 3.59476474609375,
445
+ "p95": 3.599811328125,
446
+ "p99": 3.60384859375,
447
+ "unit": "seconds"
448
+ },
449
+ "time_to_first_token_seconds": {
450
+ "name": "time_to_first_token_seconds",
451
+ "measurements": [
452
+ 0.03865302276611328,
453
+ 0.03896352005004883,
454
+ 0.03810022354125977,
455
+ 0.038313854217529295,
456
+ 0.038325054168701175
457
+ ],
458
+ "mean": 0.038471134948730466,
459
+ "median": 0.038325054168701175,
460
+ "std": 0.00030304401033698445,
461
+ "min": 0.03810022354125977,
462
+ "max": 0.03896352005004883,
463
+ "p25": 0.038313854217529295,
464
+ "p75": 0.03865302276611328,
465
+ "p90": 0.038839321136474605,
466
+ "p95": 0.03890142059326172,
467
+ "p99": 0.0389511001586914,
468
+ "unit": "seconds"
469
+ },
470
+ "tokens_per_second": {
471
+ "name": "tokens_per_second",
472
+ "measurements": [
473
+ 27.935887139015957,
474
+ 27.740344416422666,
475
+ 28.069527177175253,
476
+ 28.266902776277746,
477
+ 28.20574160373054
478
+ ],
479
+ "mean": 28.04368062252443,
480
+ "median": 28.069527177175253,
481
+ "std": 0.1899432678260422,
482
+ "min": 27.740344416422666,
483
+ "max": 28.266902776277746,
484
+ "p25": 27.935887139015957,
485
+ "p75": 28.20574160373054,
486
+ "p90": 28.242438307258865,
487
+ "p95": 28.254670541768306,
488
+ "p99": 28.264456329375857,
489
+ "unit": "tokens/sec"
490
+ },
491
+ "time_per_output_token_seconds": {
492
+ "name": "time_per_output_token_seconds",
493
+ "measurements": [
494
+ 0.03579625,
495
+ 0.0360485791015625,
496
+ 0.03562582275390625,
497
+ 0.03537706298828125,
498
+ 0.0354537744140625
499
+ ],
500
+ "mean": 0.0356602978515625,
501
+ "median": 0.03562582275390625,
502
+ "std": 0.00024220147239003607,
503
+ "min": 0.03537706298828125,
504
+ "max": 0.0360485791015625,
505
+ "p25": 0.0354537744140625,
506
+ "p75": 0.03579625,
507
+ "p90": 0.0359476474609375,
508
+ "p95": 0.03599811328125,
509
+ "p99": 0.036038485937499996,
510
+ "unit": "seconds/token"
511
+ }
512
+ },
513
+ "gpu_metrics": {
514
+ "gpu_utilization_mean": 95.13333333333334,
515
+ "gpu_utilization_max": 98,
516
+ "gpu_utilization_min": 90,
517
+ "gpu_memory_used_mean": 13849,
518
+ "gpu_memory_used_max": 13849,
519
+ "gpu_memory_used_min": 13849,
520
+ "sample_count": 90,
521
+ "gpu_monitoring_status": "success"
522
+ }
523
+ },
524
+ {
525
+ "scenario_name": "eager_sdpa_flash_attention",
526
+ "metadata": {
527
+ "timestamp": "2025-10-07T16:46:45.652740",
528
+ "commit_id": "50090c3fc82e1e0a06b4da366ea2fb6055d529e9",
529
+ "hardware_info": {
530
+ "gpu_name": "NVIDIA A10G",
531
+ "gpu_memory_total_mb": 23028,
532
+ "cpu_count": 16,
533
+ "memory_total_mb": 63607,
534
+ "python_version": "3.10.12",
535
+ "torch_version": "2.8.0+cu126",
536
+ "cuda_version": "12.6"
537
+ },
538
+ "config": {
539
+ "name": "eager",
540
+ "model_id": "meta-llama/Llama-2-7b-hf",
541
+ "variant": "eager",
542
+ "warmup_iterations": 3,
543
+ "measurement_iterations": 5,
544
+ "num_tokens_to_generate": 100,
545
+ "device": "cuda",
546
+ "torch_dtype": "float16",
547
+ "compile_mode": null,
548
+ "compile_options": {},
549
+ "use_cache": true,
550
+ "batch_size": 1,
551
+ "sequence_length": null,
552
+ "attn_implementation": "sdpa",
553
+ "sdpa_backend": "flash_attention",
554
+ "custom_params": {}
555
+ }
556
+ },
557
+ "measurements": {
558
+ "latency_seconds": {
559
+ "name": "latency_seconds",
560
+ "measurements": [
561
+ 3.41335400390625,
562
+ 3.41447216796875,
563
+ 3.41110400390625,
564
+ 3.415123046875,
565
+ 3.414615966796875
566
+ ],
567
+ "mean": 3.413733837890625,
568
+ "median": 3.41447216796875,
569
+ "std": 0.0014362092434484467,
570
+ "min": 3.41110400390625,
571
+ "max": 3.415123046875,
572
+ "p25": 3.41335400390625,
573
+ "p75": 3.414615966796875,
574
+ "p90": 3.41492021484375,
575
+ "p95": 3.415021630859375,
576
+ "p99": 3.4151027636718747,
577
+ "unit": "seconds"
578
+ },
579
+ "time_to_first_token_seconds": {
580
+ "name": "time_to_first_token_seconds",
581
+ "measurements": [
582
+ 0.03521440124511719,
583
+ 0.03465532684326172,
584
+ 0.034645729064941407,
585
+ 0.03463238525390625,
586
+ 0.034645824432373046
587
+ ],
588
+ "mean": 0.03475873336791992,
589
+ "median": 0.034645824432373046,
590
+ "std": 0.0002279510341209938,
591
+ "min": 0.03463238525390625,
592
+ "max": 0.03521440124511719,
593
+ "p25": 0.034645729064941407,
594
+ "p75": 0.03465532684326172,
595
+ "p90": 0.034990771484375,
596
+ "p95": 0.03510258636474609,
597
+ "p99": 0.03519203826904297,
598
+ "unit": "seconds"
599
+ },
600
+ "tokens_per_second": {
601
+ "name": "tokens_per_second",
602
+ "measurements": [
603
+ 29.296697584123937,
604
+ 29.28710356408892,
605
+ 29.31602199331486,
606
+ 29.28152181559161,
607
+ 29.285870203965075
608
+ ],
609
+ "mean": 29.293443032216885,
610
+ "median": 29.28710356408892,
611
+ "std": 0.012329262606361715,
612
+ "min": 29.28152181559161,
613
+ "max": 29.31602199331486,
614
+ "p25": 29.285870203965075,
615
+ "p75": 29.296697584123937,
616
+ "p90": 29.30829222963849,
617
+ "p95": 29.312157111476676,
618
+ "p99": 29.315249016947224,
619
+ "unit": "tokens/sec"
620
+ },
621
+ "time_per_output_token_seconds": {
622
+ "name": "time_per_output_token_seconds",
623
+ "measurements": [
624
+ 0.0341335400390625,
625
+ 0.0341447216796875,
626
+ 0.0341110400390625,
627
+ 0.034151230468749995,
628
+ 0.03414615966796875
629
+ ],
630
+ "mean": 0.03413733837890624,
631
+ "median": 0.0341447216796875,
632
+ "std": 1.4362092434484057e-05,
633
+ "min": 0.0341110400390625,
634
+ "max": 0.034151230468749995,
635
+ "p25": 0.0341335400390625,
636
+ "p75": 0.03414615966796875,
637
+ "p90": 0.034149202148437495,
638
+ "p95": 0.034150216308593745,
639
+ "p99": 0.03415102763671875,
640
+ "unit": "seconds/token"
641
+ }
642
+ },
643
+ "gpu_metrics": {
644
+ "gpu_utilization_mean": 96.82558139534883,
645
+ "gpu_utilization_max": 98,
646
+ "gpu_utilization_min": 95,
647
+ "gpu_memory_used_mean": 13849,
648
+ "gpu_memory_used_max": 13849,
649
+ "gpu_memory_used_min": 13849,
650
+ "sample_count": 86,
651
+ "gpu_monitoring_status": "success"
652
+ }
653
+ },
654
+ {
655
+ "scenario_name": "eager_sdpa_efficient_attention",
656
+ "metadata": {
657
+ "timestamp": "2025-10-07T16:47:18.084276",
658
+ "commit_id": "50090c3fc82e1e0a06b4da366ea2fb6055d529e9",
659
+ "hardware_info": {
660
+ "gpu_name": "NVIDIA A10G",
661
+ "gpu_memory_total_mb": 23028,
662
+ "cpu_count": 16,
663
+ "memory_total_mb": 63607,
664
+ "python_version": "3.10.12",
665
+ "torch_version": "2.8.0+cu126",
666
+ "cuda_version": "12.6"
667
+ },
668
+ "config": {
669
+ "name": "eager",
670
+ "model_id": "meta-llama/Llama-2-7b-hf",
671
+ "variant": "eager",
672
+ "warmup_iterations": 3,
673
+ "measurement_iterations": 5,
674
+ "num_tokens_to_generate": 100,
675
+ "device": "cuda",
676
+ "torch_dtype": "float16",
677
+ "compile_mode": null,
678
+ "compile_options": {},
679
+ "use_cache": true,
680
+ "batch_size": 1,
681
+ "sequence_length": null,
682
+ "attn_implementation": "sdpa",
683
+ "sdpa_backend": "efficient_attention",
684
+ "custom_params": {}
685
+ }
686
+ },
687
+ "measurements": {
688
+ "latency_seconds": {
689
+ "name": "latency_seconds",
690
+ "measurements": [
691
+ 3.4146220703125,
692
+ 3.41341796875,
693
+ 3.414650634765625,
694
+ 3.412226806640625,
695
+ 3.4114990234375
696
+ ],
697
+ "mean": 3.41328330078125,
698
+ "median": 3.41341796875,
699
+ "std": 0.0012633184122010962,
700
+ "min": 3.4114990234375,
701
+ "max": 3.414650634765625,
702
+ "p25": 3.412226806640625,
703
+ "p75": 3.4146220703125,
704
+ "p90": 3.414639208984375,
705
+ "p95": 3.414644921875,
706
+ "p99": 3.4146494921875,
707
+ "unit": "seconds"
708
+ },
709
+ "time_to_first_token_seconds": {
710
+ "name": "time_to_first_token_seconds",
711
+ "measurements": [
712
+ 0.035950462341308595,
713
+ 0.03488723373413086,
714
+ 0.034802398681640624,
715
+ 0.03477193450927735,
716
+ 0.0347264633178711
717
+ ],
718
+ "mean": 0.03502769851684571,
719
+ "median": 0.034802398681640624,
720
+ "std": 0.0004643579049252084,
721
+ "min": 0.0347264633178711,
722
+ "max": 0.035950462341308595,
723
+ "p25": 0.03477193450927735,
724
+ "p75": 0.03488723373413086,
725
+ "p90": 0.0355251708984375,
726
+ "p95": 0.03573781661987305,
727
+ "p99": 0.035907933197021485,
728
+ "unit": "seconds"
729
+ },
730
+ "tokens_per_second": {
731
+ "name": "tokens_per_second",
732
+ "measurements": [
733
+ 29.28581785651265,
734
+ 29.29614858640361,
735
+ 29.285572872922565,
736
+ 29.3063754746277,
737
+ 29.312627473431853
738
+ ],
739
+ "mean": 29.297308452779674,
740
+ "median": 29.29614858640361,
741
+ "std": 0.010844226169709864,
742
+ "min": 29.285572872922565,
743
+ "max": 29.312627473431853,
744
+ "p25": 29.28581785651265,
745
+ "p75": 29.3063754746277,
746
+ "p90": 29.31012667391019,
747
+ "p95": 29.31137707367102,
748
+ "p99": 29.312377393479686,
749
+ "unit": "tokens/sec"
750
+ },
751
+ "time_per_output_token_seconds": {
752
+ "name": "time_per_output_token_seconds",
753
+ "measurements": [
754
+ 0.034146220703125,
755
+ 0.0341341796875,
756
+ 0.03414650634765625,
757
+ 0.03412226806640625,
758
+ 0.034114990234375
759
+ ],
760
+ "mean": 0.0341328330078125,
761
+ "median": 0.0341341796875,
762
+ "std": 1.263318412201031e-05,
763
+ "min": 0.034114990234375,
764
+ "max": 0.03414650634765625,
765
+ "p25": 0.03412226806640625,
766
+ "p75": 0.034146220703125,
767
+ "p90": 0.03414639208984375,
768
+ "p95": 0.03414644921875,
769
+ "p99": 0.034146494921875,
770
+ "unit": "seconds/token"
771
+ }
772
+ },
773
+ "gpu_metrics": {
774
+ "gpu_utilization_mean": 96.82558139534883,
775
+ "gpu_utilization_max": 98,
776
+ "gpu_utilization_min": 95,
777
+ "gpu_memory_used_mean": 13849,
778
+ "gpu_memory_used_max": 13849,
779
+ "gpu_memory_used_min": 13849,
780
+ "sample_count": 86,
781
+ "gpu_monitoring_status": "success"
782
+ }
783
+ },
784
+ {
785
+ "scenario_name": "compiled_compile_max-autotune_sdpa_default",
786
+ "metadata": {
787
+ "timestamp": "2025-10-07T16:47:50.526967",
788
+ "commit_id": "50090c3fc82e1e0a06b4da366ea2fb6055d529e9",
789
+ "hardware_info": {
790
+ "gpu_name": "NVIDIA A10G",
791
+ "gpu_memory_total_mb": 23028,
792
+ "cpu_count": 16,
793
+ "memory_total_mb": 63607,
794
+ "python_version": "3.10.12",
795
+ "torch_version": "2.8.0+cu126",
796
+ "cuda_version": "12.6"
797
+ },
798
+ "config": {
799
+ "name": "compiled",
800
+ "model_id": "meta-llama/Llama-2-7b-hf",
801
+ "variant": "compiled",
802
+ "warmup_iterations": 3,
803
+ "measurement_iterations": 5,
804
+ "num_tokens_to_generate": 100,
805
+ "device": "cuda",
806
+ "torch_dtype": "float16",
807
+ "compile_mode": "max-autotune",
808
+ "compile_options": {},
809
+ "use_cache": true,
810
+ "batch_size": 1,
811
+ "sequence_length": null,
812
+ "attn_implementation": "sdpa",
813
+ "sdpa_backend": null,
814
+ "custom_params": {}
815
+ }
816
+ },
817
+ "measurements": {
818
+ "latency_seconds": {
819
+ "name": "latency_seconds",
820
+ "measurements": [
821
+ 3.708273193359375,
822
+ 3.701499755859375,
823
+ 3.699865234375,
824
+ 3.69852490234375,
825
+ 3.7069599609375
826
+ ],
827
+ "mean": 3.7030246093750003,
828
+ "median": 3.701499755859375,
829
+ "std": 0.003888156832313487,
830
+ "min": 3.69852490234375,
831
+ "max": 3.708273193359375,
832
+ "p25": 3.699865234375,
833
+ "p75": 3.7069599609375,
834
+ "p90": 3.707747900390625,
835
+ "p95": 3.7080105468749998,
836
+ "p99": 3.7082206640625,
837
+ "unit": "seconds"
838
+ },
839
+ "time_to_first_token_seconds": {
840
+ "name": "time_to_first_token_seconds",
841
+ "measurements": [
842
+ 0.03847145462036133,
843
+ 0.03749955368041992,
844
+ 0.037446048736572264,
845
+ 0.03802300643920899,
846
+ 0.037486270904541014
847
+ ],
848
+ "mean": 0.0377852668762207,
849
+ "median": 0.03749955368041992,
850
+ "std": 0.0004033542580959285,
851
+ "min": 0.037446048736572264,
852
+ "max": 0.03847145462036133,
853
+ "p25": 0.037486270904541014,
854
+ "p75": 0.03802300643920899,
855
+ "p90": 0.03829207534790039,
856
+ "p95": 0.03838176498413086,
857
+ "p99": 0.03845351669311523,
858
+ "unit": "seconds"
859
+ },
860
+ "tokens_per_second": {
861
+ "name": "tokens_per_second",
862
+ "measurements": [
863
+ 26.966729468334734,
864
+ 27.016076346270907,
865
+ 27.028011472123932,
866
+ 27.03780632560569,
867
+ 26.976282736733346
868
+ ],
869
+ "mean": 27.00498126981372,
870
+ "median": 27.016076346270907,
871
+ "std": 0.028347068085215355,
872
+ "min": 26.966729468334734,
873
+ "max": 27.03780632560569,
874
+ "p25": 26.976282736733346,
875
+ "p75": 27.028011472123932,
876
+ "p90": 27.033888384212986,
877
+ "p95": 27.03584735490934,
878
+ "p99": 27.03741453146642,
879
+ "unit": "tokens/sec"
880
+ },
881
+ "time_per_output_token_seconds": {
882
+ "name": "time_per_output_token_seconds",
883
+ "measurements": [
884
+ 0.03708273193359375,
885
+ 0.03701499755859375,
886
+ 0.03699865234375,
887
+ 0.0369852490234375,
888
+ 0.037069599609375004
889
+ ],
890
+ "mean": 0.037030246093750004,
891
+ "median": 0.03701499755859375,
892
+ "std": 3.888156832313566e-05,
893
+ "min": 0.0369852490234375,
894
+ "max": 0.03708273193359375,
895
+ "p25": 0.03699865234375,
896
+ "p75": 0.037069599609375004,
897
+ "p90": 0.037077479003906254,
898
+ "p95": 0.03708010546875,
899
+ "p99": 0.037082206640625004,
900
+ "unit": "seconds/token"
901
+ }
902
+ },
903
+ "gpu_metrics": {
904
+ "gpu_utilization_mean": 90.47311827956989,
905
+ "gpu_utilization_max": 93,
906
+ "gpu_utilization_min": 88,
907
+ "gpu_memory_used_mean": 13845,
908
+ "gpu_memory_used_max": 13845,
909
+ "gpu_memory_used_min": 13845,
910
+ "sample_count": 93,
911
+ "gpu_monitoring_status": "success"
912
+ }
913
+ },
914
+ {
915
+ "scenario_name": "compiled_compile_max-autotune_sdpa_math",
916
+ "metadata": {
917
+ "timestamp": "2025-10-07T16:48:25.306048",
918
+ "commit_id": "50090c3fc82e1e0a06b4da366ea2fb6055d529e9",
919
+ "hardware_info": {
920
+ "gpu_name": "NVIDIA A10G",
921
+ "gpu_memory_total_mb": 23028,
922
+ "cpu_count": 16,
923
+ "memory_total_mb": 63607,
924
+ "python_version": "3.10.12",
925
+ "torch_version": "2.8.0+cu126",
926
+ "cuda_version": "12.6"
927
+ },
928
+ "config": {
929
+ "name": "compiled",
930
+ "model_id": "meta-llama/Llama-2-7b-hf",
931
+ "variant": "compiled",
932
+ "warmup_iterations": 3,
933
+ "measurement_iterations": 5,
934
+ "num_tokens_to_generate": 100,
935
+ "device": "cuda",
936
+ "torch_dtype": "float16",
937
+ "compile_mode": "max-autotune",
938
+ "compile_options": {},
939
+ "use_cache": true,
940
+ "batch_size": 1,
941
+ "sequence_length": null,
942
+ "attn_implementation": "sdpa",
943
+ "sdpa_backend": "math",
944
+ "custom_params": {}
945
+ }
946
+ },
947
+ "measurements": {
948
+ "latency_seconds": {
949
+ "name": "latency_seconds",
950
+ "measurements": [
951
+ 3.8983623046875,
952
+ 3.943302734375,
953
+ 3.923885986328125,
954
+ 3.88124853515625,
955
+ 3.91166455078125
956
+ ],
957
+ "mean": 3.911692822265625,
958
+ "median": 3.91166455078125,
959
+ "std": 0.021224598508301884,
960
+ "min": 3.88124853515625,
961
+ "max": 3.943302734375,
962
+ "p25": 3.8983623046875,
963
+ "p75": 3.923885986328125,
964
+ "p90": 3.93553603515625,
965
+ "p95": 3.939419384765625,
966
+ "p99": 3.942526064453125,
967
+ "unit": "seconds"
968
+ },
969
+ "time_to_first_token_seconds": {
970
+ "name": "time_to_first_token_seconds",
971
+ "measurements": [
972
+ 0.04123049545288086,
973
+ 0.04034326553344727,
974
+ 0.04047270584106445,
975
+ 0.0422825927734375,
976
+ 0.041080768585205076
977
+ ],
978
+ "mean": 0.04108196563720703,
979
+ "median": 0.041080768585205076,
980
+ "std": 0.0006899917973165999,
981
+ "min": 0.04034326553344727,
982
+ "max": 0.0422825927734375,
983
+ "p25": 0.04047270584106445,
984
+ "p75": 0.04123049545288086,
985
+ "p90": 0.041861753845214844,
986
+ "p95": 0.04207217330932617,
987
+ "p99": 0.042240508880615237,
988
+ "unit": "seconds"
989
+ },
990
+ "tokens_per_second": {
991
+ "name": "tokens_per_second",
992
+ "measurements": [
993
+ 25.651797391883562,
994
+ 25.359452909427624,
995
+ 25.484940273093287,
996
+ 25.764905054186187,
997
+ 25.56456431828432
998
+ ],
999
+ "mean": 25.565131989374994,
1000
+ "median": 25.56456431828432,
1001
+ "std": 0.13867594380764017,
1002
+ "min": 25.359452909427624,
1003
+ "max": 25.764905054186187,
1004
+ "p25": 25.484940273093287,
1005
+ "p75": 25.651797391883562,
1006
+ "p90": 25.719661989265138,
1007
+ "p95": 25.74228352172566,
1008
+ "p99": 25.760380747694082,
1009
+ "unit": "tokens/sec"
1010
+ },
1011
+ "time_per_output_token_seconds": {
1012
+ "name": "time_per_output_token_seconds",
1013
+ "measurements": [
1014
+ 0.038983623046875,
1015
+ 0.03943302734375,
1016
+ 0.039238859863281246,
1017
+ 0.038812485351562495,
1018
+ 0.0391166455078125
1019
+ ],
1020
+ "mean": 0.03911692822265625,
1021
+ "median": 0.0391166455078125,
1022
+ "std": 0.00021224598508302019,
1023
+ "min": 0.038812485351562495,
1024
+ "max": 0.03943302734375,
1025
+ "p25": 0.038983623046875,
1026
+ "p75": 0.039238859863281246,
1027
+ "p90": 0.0393553603515625,
1028
+ "p95": 0.03939419384765625,
1029
+ "p99": 0.03942526064453125,
1030
+ "unit": "seconds/token"
1031
+ }
1032
+ },
1033
+ "gpu_metrics": {
1034
+ "gpu_utilization_mean": 88.31313131313131,
1035
+ "gpu_utilization_max": 92,
1036
+ "gpu_utilization_min": 83,
1037
+ "gpu_memory_used_mean": 13845,
1038
+ "gpu_memory_used_max": 13845,
1039
+ "gpu_memory_used_min": 13845,
1040
+ "sample_count": 99,
1041
+ "gpu_monitoring_status": "success"
1042
+ }
1043
+ },
1044
+ {
1045
+ "scenario_name": "compiled_compile_max-autotune_sdpa_efficient_attention",
1046
+ "metadata": {
1047
+ "timestamp": "2025-10-07T16:49:06.504836",
1048
+ "commit_id": "50090c3fc82e1e0a06b4da366ea2fb6055d529e9",
1049
+ "hardware_info": {
1050
+ "gpu_name": "NVIDIA A10G",
1051
+ "gpu_memory_total_mb": 23028,
1052
+ "cpu_count": 16,
1053
+ "memory_total_mb": 63607,
1054
+ "python_version": "3.10.12",
1055
+ "torch_version": "2.8.0+cu126",
1056
+ "cuda_version": "12.6"
1057
+ },
1058
+ "config": {
1059
+ "name": "compiled",
1060
+ "model_id": "meta-llama/Llama-2-7b-hf",
1061
+ "variant": "compiled",
1062
+ "warmup_iterations": 3,
1063
+ "measurement_iterations": 5,
1064
+ "num_tokens_to_generate": 100,
1065
+ "device": "cuda",
1066
+ "torch_dtype": "float16",
1067
+ "compile_mode": "max-autotune",
1068
+ "compile_options": {},
1069
+ "use_cache": true,
1070
+ "batch_size": 1,
1071
+ "sequence_length": null,
1072
+ "attn_implementation": "sdpa",
1073
+ "sdpa_backend": "efficient_attention",
1074
+ "custom_params": {}
1075
+ }
1076
+ },
1077
+ "measurements": {
1078
+ "latency_seconds": {
1079
+ "name": "latency_seconds",
1080
+ "measurements": [
1081
+ 3.699159423828125,
1082
+ 3.708882080078125,
1083
+ 3.71523876953125,
1084
+ 3.7134765625,
1085
+ 3.710626220703125
1086
+ ],
1087
+ "mean": 3.7094766113281255,
1088
+ "median": 3.710626220703125,
1089
+ "std": 0.005609304833358554,
1090
+ "min": 3.699159423828125,
1091
+ "max": 3.71523876953125,
1092
+ "p25": 3.708882080078125,
1093
+ "p75": 3.7134765625,
1094
+ "p90": 3.71453388671875,
1095
+ "p95": 3.714886328125,
1096
+ "p99": 3.71516828125,
1097
+ "unit": "seconds"
1098
+ },
1099
+ "time_to_first_token_seconds": {
1100
+ "name": "time_to_first_token_seconds",
1101
+ "measurements": [
1102
+ 0.03812707138061523,
1103
+ 0.037854496002197265,
1104
+ 0.03822515106201172,
1105
+ 0.03748479843139649,
1106
+ 0.03754169464111328
1107
+ ],
1108
+ "mean": 0.0378466423034668,
1109
+ "median": 0.037854496002197265,
1110
+ "std": 0.0002986285628398596,
1111
+ "min": 0.03748479843139649,
1112
+ "max": 0.03822515106201172,
1113
+ "p25": 0.03754169464111328,
1114
+ "p75": 0.03812707138061523,
1115
+ "p90": 0.038185919189453126,
1116
+ "p95": 0.03820553512573242,
1117
+ "p99": 0.03822122787475586,
1118
+ "unit": "seconds"
1119
+ },
1120
+ "tokens_per_second": {
1121
+ "name": "tokens_per_second",
1122
+ "measurements": [
1123
+ 27.033168496564457,
1124
+ 26.962302343646787,
1125
+ 26.916170454535,
1126
+ 26.928943354546888,
1127
+ 26.94962899848507
1128
+ ],
1129
+ "mean": 26.958042729555636,
1130
+ "median": 26.94962899848507,
1131
+ "std": 0.04082371147433942,
1132
+ "min": 26.916170454535,
1133
+ "max": 27.033168496564457,
1134
+ "p25": 26.928943354546888,
1135
+ "p75": 26.962302343646787,
1136
+ "p90": 27.00482203539739,
1137
+ "p95": 27.018995265980923,
1138
+ "p99": 27.03033385044775,
1139
+ "unit": "tokens/sec"
1140
+ },
1141
+ "time_per_output_token_seconds": {
1142
+ "name": "time_per_output_token_seconds",
1143
+ "measurements": [
1144
+ 0.03699159423828125,
1145
+ 0.03708882080078125,
1146
+ 0.037152387695312504,
1147
+ 0.037134765624999996,
1148
+ 0.037106262207031254
1149
+ ],
1150
+ "mean": 0.03709476611328125,
1151
+ "median": 0.037106262207031254,
1152
+ "std": 5.6093048333585025e-05,
1153
+ "min": 0.03699159423828125,
1154
+ "max": 0.037152387695312504,
1155
+ "p25": 0.03708882080078125,
1156
+ "p75": 0.037134765624999996,
1157
+ "p90": 0.0371453388671875,
1158
+ "p95": 0.03714886328125,
1159
+ "p99": 0.0371516828125,
1160
+ "unit": "seconds/token"
1161
+ }
1162
+ },
1163
+ "gpu_metrics": {
1164
+ "gpu_utilization_mean": 90.17021276595744,
1165
+ "gpu_utilization_max": 93,
1166
+ "gpu_utilization_min": 86,
1167
+ "gpu_memory_used_mean": 13845,
1168
+ "gpu_memory_used_max": 13845,
1169
+ "gpu_memory_used_min": 13845,
1170
+ "sample_count": 94,
1171
+ "gpu_monitoring_status": "success"
1172
+ }
1173
+ }
1174
+ ]
1175
+ }