hf-transformers-bot commited on
Commit
53689b1
·
verified ·
1 Parent(s): cb4b818

Upload benchmark results for run 18169125332

Browse files
2025-10-01/18169125332/benchmark_results/Llama-2-7b-hf/Llama-2-7b-hf_benchmark_20251001_165313.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": "2025-10-01T16:47:12.801999",
8
+ "commit_id": "94bbf8e19911a3aed089d47094c836a18717c9c0",
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
+ 3.661740966796875,
42
+ 3.642160888671875,
43
+ 3.643485107421875,
44
+ 3.653228271484375,
45
+ 3.640120849609375
46
+ ],
47
+ "mean": 3.6481472167968754,
48
+ "median": 3.643485107421875,
49
+ "std": 0.008155993063668876,
50
+ "min": 3.640120849609375,
51
+ "max": 3.661740966796875,
52
+ "p25": 3.642160888671875,
53
+ "p75": 3.653228271484375,
54
+ "p90": 3.658335888671875,
55
+ "p95": 3.660038427734375,
56
+ "p99": 3.661400458984375,
57
+ "unit": "seconds"
58
+ },
59
+ "time_to_first_token_seconds": {
60
+ "name": "time_to_first_token_seconds",
61
+ "measurements": [
62
+ 0.03770483016967773,
63
+ 0.03731475067138672,
64
+ 0.037010398864746094,
65
+ 0.037335262298583985,
66
+ 0.03707673645019531
67
+ ],
68
+ "mean": 0.037288395690917964,
69
+ "median": 0.03731475067138672,
70
+ "std": 0.0002442910594166195,
71
+ "min": 0.037010398864746094,
72
+ "max": 0.03770483016967773,
73
+ "p25": 0.03707673645019531,
74
+ "p75": 0.037335262298583985,
75
+ "p90": 0.03755700302124023,
76
+ "p95": 0.03763091659545898,
77
+ "p99": 0.03769004745483398,
78
+ "unit": "seconds"
79
+ },
80
+ "tokens_per_second": {
81
+ "name": "tokens_per_second",
82
+ "measurements": [
83
+ 27.309413993714433,
84
+ 27.456228062584383,
85
+ 27.446249141048323,
86
+ 27.373049962565883,
87
+ 27.47161540274991
88
+ ],
89
+ "mean": 27.411311312532586,
90
+ "median": 27.446249141048323,
91
+ "std": 0.06119108444398242,
92
+ "min": 27.309413993714433,
93
+ "max": 27.47161540274991,
94
+ "p25": 27.373049962565883,
95
+ "p75": 27.456228062584383,
96
+ "p90": 27.4654604666837,
97
+ "p95": 27.468537934716807,
98
+ "p99": 27.47099990914329,
99
+ "unit": "tokens/sec"
100
+ },
101
+ "time_per_output_token_seconds": {
102
+ "name": "time_per_output_token_seconds",
103
+ "measurements": [
104
+ 0.03661740966796875,
105
+ 0.03642160888671875,
106
+ 0.03643485107421875,
107
+ 0.036532282714843746,
108
+ 0.03640120849609375
109
+ ],
110
+ "mean": 0.03648147216796875,
111
+ "median": 0.03643485107421875,
112
+ "std": 8.155993063668952e-05,
113
+ "min": 0.03640120849609375,
114
+ "max": 0.03661740966796875,
115
+ "p25": 0.03642160888671875,
116
+ "p75": 0.036532282714843746,
117
+ "p90": 0.03658335888671875,
118
+ "p95": 0.03660038427734375,
119
+ "p99": 0.036614004589843756,
120
+ "unit": "seconds/token"
121
+ }
122
+ },
123
+ "gpu_metrics": {
124
+ "gpu_utilization_mean": 91.03260869565217,
125
+ "gpu_utilization_max": 94,
126
+ "gpu_utilization_min": 88,
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": "2025-10-01T16:47:48.068116",
138
+ "commit_id": "94bbf8e19911a3aed089d47094c836a18717c9c0",
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
+ 7.810576171875,
172
+ 7.8097783203125,
173
+ 7.88556787109375,
174
+ 7.87764697265625,
175
+ 7.90265869140625
176
+ ],
177
+ "mean": 7.8572456054687505,
178
+ "median": 7.87764697265625,
179
+ "std": 0.03927313096682077,
180
+ "min": 7.8097783203125,
181
+ "max": 7.90265869140625,
182
+ "p25": 7.810576171875,
183
+ "p75": 7.88556787109375,
184
+ "p90": 7.89582236328125,
185
+ "p95": 7.89924052734375,
186
+ "p99": 7.9019750585937505,
187
+ "unit": "seconds"
188
+ },
189
+ "time_to_first_token_seconds": {
190
+ "name": "time_to_first_token_seconds",
191
+ "measurements": [
192
+ 0.038833953857421874,
193
+ 0.0381328010559082,
194
+ 0.03802569580078125,
195
+ 0.03807100677490234,
196
+ 0.038037822723388674
197
+ ],
198
+ "mean": 0.03822025604248047,
199
+ "median": 0.03807100677490234,
200
+ "std": 0.0003090903848966267,
201
+ "min": 0.03802569580078125,
202
+ "max": 0.038833953857421874,
203
+ "p25": 0.038037822723388674,
204
+ "p75": 0.0381328010559082,
205
+ "p90": 0.03855349273681641,
206
+ "p95": 0.03869372329711914,
207
+ "p99": 0.03880590774536133,
208
+ "unit": "seconds"
209
+ },
210
+ "tokens_per_second": {
211
+ "name": "tokens_per_second",
212
+ "measurements": [
213
+ 12.803152776371181,
214
+ 12.80446075401518,
215
+ 12.681394876654549,
216
+ 12.694145897512994,
217
+ 12.65396924059811
218
+ ],
219
+ "mean": 12.727424709030403,
220
+ "median": 12.694145897512994,
221
+ "std": 0.06370435309652552,
222
+ "min": 12.65396924059811,
223
+ "max": 12.80446075401518,
224
+ "p25": 12.681394876654549,
225
+ "p75": 12.803152776371181,
226
+ "p90": 12.80393756295758,
227
+ "p95": 12.80419915848638,
228
+ "p99": 12.80440843490942,
229
+ "unit": "tokens/sec"
230
+ },
231
+ "time_per_output_token_seconds": {
232
+ "name": "time_per_output_token_seconds",
233
+ "measurements": [
234
+ 0.07810576171875,
235
+ 0.07809778320312499,
236
+ 0.0788556787109375,
237
+ 0.0787764697265625,
238
+ 0.0790265869140625
239
+ ],
240
+ "mean": 0.0785724560546875,
241
+ "median": 0.0787764697265625,
242
+ "std": 0.00039273130966820905,
243
+ "min": 0.07809778320312499,
244
+ "max": 0.0790265869140625,
245
+ "p25": 0.07810576171875,
246
+ "p75": 0.0788556787109375,
247
+ "p90": 0.0789582236328125,
248
+ "p95": 0.0789924052734375,
249
+ "p99": 0.0790197505859375,
250
+ "unit": "seconds/token"
251
+ }
252
+ },
253
+ "gpu_metrics": {
254
+ "gpu_utilization_mean": 44.07058823529412,
255
+ "gpu_utilization_max": 94,
256
+ "gpu_utilization_min": 0,
257
+ "gpu_memory_used_mean": 13630.482352941177,
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": "2025-10-01T16:49:15.697399",
268
+ "commit_id": "94bbf8e19911a3aed089d47094c836a18717c9c0",
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
+ 3.4123525390625,
302
+ 3.410125732421875,
303
+ 3.40988134765625,
304
+ 3.409494873046875,
305
+ 3.41013525390625
306
+ ],
307
+ "mean": 3.41039794921875,
308
+ "median": 3.410125732421875,
309
+ "std": 0.0010045850900333299,
310
+ "min": 3.409494873046875,
311
+ "max": 3.4123525390625,
312
+ "p25": 3.40988134765625,
313
+ "p75": 3.41013525390625,
314
+ "p90": 3.411465625,
315
+ "p95": 3.41190908203125,
316
+ "p99": 3.41226384765625,
317
+ "unit": "seconds"
318
+ },
319
+ "time_to_first_token_seconds": {
320
+ "name": "time_to_first_token_seconds",
321
+ "measurements": [
322
+ 0.03502592086791992,
323
+ 0.03460681533813476,
324
+ 0.03460809707641602,
325
+ 0.03502409744262695,
326
+ 0.03460617446899414
327
+ ],
328
+ "mean": 0.03477422103881836,
329
+ "median": 0.03460809707641602,
330
+ "std": 0.00020476938749987858,
331
+ "min": 0.03460617446899414,
332
+ "max": 0.03502592086791992,
333
+ "p25": 0.03460681533813476,
334
+ "p75": 0.03502409744262695,
335
+ "p90": 0.035025191497802734,
336
+ "p95": 0.03502555618286132,
337
+ "p99": 0.0350258479309082,
338
+ "unit": "seconds"
339
+ },
340
+ "tokens_per_second": {
341
+ "name": "tokens_per_second",
342
+ "measurements": [
343
+ 29.305295644357344,
344
+ 29.324431955469247,
345
+ 29.326533625204895,
346
+ 29.329857859746713,
347
+ 29.32435007832952
348
+ ],
349
+ "mean": 29.322093832621544,
350
+ "median": 29.324431955469247,
351
+ "std": 0.008633984798327839,
352
+ "min": 29.305295644357344,
353
+ "max": 29.329857859746713,
354
+ "p25": 29.32435007832952,
355
+ "p75": 29.326533625204895,
356
+ "p90": 29.328528165929985,
357
+ "p95": 29.32919301283835,
358
+ "p99": 29.32972489036504,
359
+ "unit": "tokens/sec"
360
+ },
361
+ "time_per_output_token_seconds": {
362
+ "name": "time_per_output_token_seconds",
363
+ "measurements": [
364
+ 0.034123525390625,
365
+ 0.03410125732421875,
366
+ 0.034098813476562496,
367
+ 0.03409494873046875,
368
+ 0.034101352539062504
369
+ ],
370
+ "mean": 0.0341039794921875,
371
+ "median": 0.03410125732421875,
372
+ "std": 1.0045850900333952e-05,
373
+ "min": 0.03409494873046875,
374
+ "max": 0.034123525390625,
375
+ "p25": 0.034098813476562496,
376
+ "p75": 0.034101352539062504,
377
+ "p90": 0.03411465625,
378
+ "p95": 0.0341190908203125,
379
+ "p99": 0.0341226384765625,
380
+ "unit": "seconds/token"
381
+ }
382
+ },
383
+ "gpu_metrics": {
384
+ "gpu_utilization_mean": 96.83720930232558,
385
+ "gpu_utilization_max": 97,
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": "2025-10-01T16:49:48.180899",
398
+ "commit_id": "94bbf8e19911a3aed089d47094c836a18717c9c0",
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
+ 3.5281689453125,
432
+ 3.532262939453125,
433
+ 3.53406494140625,
434
+ 3.52518505859375,
435
+ 3.5300654296875
436
+ ],
437
+ "mean": 3.529949462890625,
438
+ "median": 3.5300654296875,
439
+ "std": 0.0031039515674662487,
440
+ "min": 3.52518505859375,
441
+ "max": 3.53406494140625,
442
+ "p25": 3.5281689453125,
443
+ "p75": 3.532262939453125,
444
+ "p90": 3.533344140625,
445
+ "p95": 3.533704541015625,
446
+ "p99": 3.5339928613281253,
447
+ "unit": "seconds"
448
+ },
449
+ "time_to_first_token_seconds": {
450
+ "name": "time_to_first_token_seconds",
451
+ "measurements": [
452
+ 0.0381168327331543,
453
+ 0.03922467041015625,
454
+ 0.03775148773193359,
455
+ 0.037580223083496095,
456
+ 0.037741470336914065
457
+ ],
458
+ "mean": 0.038082936859130855,
459
+ "median": 0.03775148773193359,
460
+ "std": 0.00059730787400477,
461
+ "min": 0.037580223083496095,
462
+ "max": 0.03922467041015625,
463
+ "p25": 0.037741470336914065,
464
+ "p75": 0.0381168327331543,
465
+ "p90": 0.03878153533935547,
466
+ "p95": 0.03900310287475586,
467
+ "p99": 0.039180356903076176,
468
+ "unit": "seconds"
469
+ },
470
+ "tokens_per_second": {
471
+ "name": "tokens_per_second",
472
+ "measurements": [
473
+ 28.343313925728893,
474
+ 28.310463211292614,
475
+ 28.296027848375843,
476
+ 28.367305074160143,
477
+ 28.32808682779926
478
+ ],
479
+ "mean": 28.32903937747135,
480
+ "median": 28.32808682779926,
481
+ "std": 0.024914929761237284,
482
+ "min": 28.296027848375843,
483
+ "max": 28.367305074160143,
484
+ "p25": 28.310463211292614,
485
+ "p75": 28.343313925728893,
486
+ "p90": 28.357708614787644,
487
+ "p95": 28.36250684447389,
488
+ "p99": 28.366345428222893,
489
+ "unit": "tokens/sec"
490
+ },
491
+ "time_per_output_token_seconds": {
492
+ "name": "time_per_output_token_seconds",
493
+ "measurements": [
494
+ 0.035281689453125,
495
+ 0.03532262939453125,
496
+ 0.0353406494140625,
497
+ 0.0352518505859375,
498
+ 0.035300654296875
499
+ ],
500
+ "mean": 0.03529949462890625,
501
+ "median": 0.035300654296875,
502
+ "std": 3.1039515674661405e-05,
503
+ "min": 0.0352518505859375,
504
+ "max": 0.0353406494140625,
505
+ "p25": 0.035281689453125,
506
+ "p75": 0.03532262939453125,
507
+ "p90": 0.03533344140625,
508
+ "p95": 0.035337045410156254,
509
+ "p99": 0.03533992861328125,
510
+ "unit": "seconds/token"
511
+ }
512
+ },
513
+ "gpu_metrics": {
514
+ "gpu_utilization_mean": 96.2247191011236,
515
+ "gpu_utilization_max": 98,
516
+ "gpu_utilization_min": 88,
517
+ "gpu_memory_used_mean": 13849,
518
+ "gpu_memory_used_max": 13849,
519
+ "gpu_memory_used_min": 13849,
520
+ "sample_count": 89,
521
+ "gpu_monitoring_status": "success"
522
+ }
523
+ },
524
+ {
525
+ "scenario_name": "eager_sdpa_flash_attention",
526
+ "metadata": {
527
+ "timestamp": "2025-10-01T16:50:21.651445",
528
+ "commit_id": "94bbf8e19911a3aed089d47094c836a18717c9c0",
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
+ 3.408355712890625,
562
+ 3.409234619140625,
563
+ 3.408414794921875,
564
+ 3.40861572265625,
565
+ 3.4106962890625
566
+ ],
567
+ "mean": 3.409063427734375,
568
+ "median": 3.40861572265625,
569
+ "std": 0.0008737925700020731,
570
+ "min": 3.408355712890625,
571
+ "max": 3.4106962890625,
572
+ "p25": 3.408414794921875,
573
+ "p75": 3.409234619140625,
574
+ "p90": 3.41011162109375,
575
+ "p95": 3.410403955078125,
576
+ "p99": 3.410637822265625,
577
+ "unit": "seconds"
578
+ },
579
+ "time_to_first_token_seconds": {
580
+ "name": "time_to_first_token_seconds",
581
+ "measurements": [
582
+ 0.03529158401489258,
583
+ 0.034644287109375,
584
+ 0.034611263275146485,
585
+ 0.034654014587402346,
586
+ 0.03466099166870117
587
+ ],
588
+ "mean": 0.03477242813110352,
589
+ "median": 0.034654014587402346,
590
+ "std": 0.00026013723931905716,
591
+ "min": 0.034611263275146485,
592
+ "max": 0.03529158401489258,
593
+ "p25": 0.034644287109375,
594
+ "p75": 0.03466099166870117,
595
+ "p90": 0.03503934707641602,
596
+ "p95": 0.0351654655456543,
597
+ "p99": 0.03526636032104492,
598
+ "unit": "seconds"
599
+ },
600
+ "tokens_per_second": {
601
+ "name": "tokens_per_second",
602
+ "measurements": [
603
+ 29.339660652728657,
604
+ 29.33209684031874,
605
+ 29.33915207415127,
606
+ 29.337422618608493,
607
+ 29.31952643238342
608
+ ],
609
+ "mean": 29.333571723638112,
610
+ "median": 29.337422618608493,
611
+ "std": 0.00751651845773265,
612
+ "min": 29.31952643238342,
613
+ "max": 29.339660652728657,
614
+ "p25": 29.33209684031874,
615
+ "p75": 29.33915207415127,
616
+ "p90": 29.339457221297703,
617
+ "p95": 29.33955893701318,
618
+ "p99": 29.339640309585562,
619
+ "unit": "tokens/sec"
620
+ },
621
+ "time_per_output_token_seconds": {
622
+ "name": "time_per_output_token_seconds",
623
+ "measurements": [
624
+ 0.03408355712890625,
625
+ 0.03409234619140625,
626
+ 0.03408414794921875,
627
+ 0.034086157226562504,
628
+ 0.034106962890625
629
+ ],
630
+ "mean": 0.034090634277343754,
631
+ "median": 0.034086157226562504,
632
+ "std": 8.737925700022388e-06,
633
+ "min": 0.03408355712890625,
634
+ "max": 0.034106962890625,
635
+ "p25": 0.03408414794921875,
636
+ "p75": 0.03409234619140625,
637
+ "p90": 0.0341011162109375,
638
+ "p95": 0.03410403955078125,
639
+ "p99": 0.034106378222656254,
640
+ "unit": "seconds/token"
641
+ }
642
+ },
643
+ "gpu_metrics": {
644
+ "gpu_utilization_mean": 96.87209302325581,
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": "2025-10-01T16:50:54.127209",
658
+ "commit_id": "94bbf8e19911a3aed089d47094c836a18717c9c0",
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
+ 3.4062568359375,
692
+ 3.40503564453125,
693
+ 3.4069375,
694
+ 3.40852978515625,
695
+ 3.404775634765625
696
+ ],
697
+ "mean": 3.4063070800781245,
698
+ "median": 3.4062568359375,
699
+ "std": 0.001363982749958574,
700
+ "min": 3.404775634765625,
701
+ "max": 3.40852978515625,
702
+ "p25": 3.40503564453125,
703
+ "p75": 3.4069375,
704
+ "p90": 3.40789287109375,
705
+ "p95": 3.408211328125,
706
+ "p99": 3.40846609375,
707
+ "unit": "seconds"
708
+ },
709
+ "time_to_first_token_seconds": {
710
+ "name": "time_to_first_token_seconds",
711
+ "measurements": [
712
+ 0.03557033538818359,
713
+ 0.03469891357421875,
714
+ 0.035000446319580075,
715
+ 0.03500492858886719,
716
+ 0.03476831817626953
717
+ ],
718
+ "mean": 0.03500858840942382,
719
+ "median": 0.035000446319580075,
720
+ "std": 0.00030635523320438496,
721
+ "min": 0.03469891357421875,
722
+ "max": 0.03557033538818359,
723
+ "p25": 0.03476831817626953,
724
+ "p75": 0.03500492858886719,
725
+ "p90": 0.03534417266845703,
726
+ "p95": 0.03545725402832031,
727
+ "p99": 0.035547719116210935,
728
+ "unit": "seconds"
729
+ },
730
+ "tokens_per_second": {
731
+ "name": "tokens_per_second",
732
+ "measurements": [
733
+ 29.357739247656326,
734
+ 29.36826818850126,
735
+ 29.35187393370145,
736
+ 29.33816228788387,
737
+ 29.370510931444596
738
+ ],
739
+ "mean": 29.3573109178375,
740
+ "median": 29.357739247656326,
741
+ "std": 0.011753437213486857,
742
+ "min": 29.33816228788387,
743
+ "max": 29.370510931444596,
744
+ "p25": 29.35187393370145,
745
+ "p75": 29.36826818850126,
746
+ "p90": 29.369613834267263,
747
+ "p95": 29.370062382855927,
748
+ "p99": 29.37042122172686,
749
+ "unit": "tokens/sec"
750
+ },
751
+ "time_per_output_token_seconds": {
752
+ "name": "time_per_output_token_seconds",
753
+ "measurements": [
754
+ 0.034062568359375,
755
+ 0.0340503564453125,
756
+ 0.034069375,
757
+ 0.0340852978515625,
758
+ 0.03404775634765625
759
+ ],
760
+ "mean": 0.034063070800781255,
761
+ "median": 0.034062568359375,
762
+ "std": 1.3639827499586041e-05,
763
+ "min": 0.03404775634765625,
764
+ "max": 0.0340852978515625,
765
+ "p25": 0.0340503564453125,
766
+ "p75": 0.034069375,
767
+ "p90": 0.0340789287109375,
768
+ "p95": 0.03408211328125,
769
+ "p99": 0.034084660937500004,
770
+ "unit": "seconds/token"
771
+ }
772
+ },
773
+ "gpu_metrics": {
774
+ "gpu_utilization_mean": 97.04651162790698,
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": "2025-10-01T16:51:26.596680",
788
+ "commit_id": "94bbf8e19911a3aed089d47094c836a18717c9c0",
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
+ 3.683592529296875,
822
+ 3.6912255859375,
823
+ 3.699804931640625,
824
+ 3.690763671875,
825
+ 3.693473876953125
826
+ ],
827
+ "mean": 3.691772119140625,
828
+ "median": 3.6912255859375,
829
+ "std": 0.00520852014979685,
830
+ "min": 3.683592529296875,
831
+ "max": 3.699804931640625,
832
+ "p25": 3.690763671875,
833
+ "p75": 3.693473876953125,
834
+ "p90": 3.6972725097656247,
835
+ "p95": 3.698538720703125,
836
+ "p99": 3.699551689453125,
837
+ "unit": "seconds"
838
+ },
839
+ "time_to_first_token_seconds": {
840
+ "name": "time_to_first_token_seconds",
841
+ "measurements": [
842
+ 0.03793596649169922,
843
+ 0.03736118316650391,
844
+ 0.037761886596679686,
845
+ 0.037478206634521484,
846
+ 0.03744326400756836
847
+ ],
848
+ "mean": 0.03759610137939453,
849
+ "median": 0.037478206634521484,
850
+ "std": 0.00021699678922753677,
851
+ "min": 0.03736118316650391,
852
+ "max": 0.03793596649169922,
853
+ "p25": 0.03744326400756836,
854
+ "p75": 0.037761886596679686,
855
+ "p90": 0.037866334533691405,
856
+ "p95": 0.03790115051269531,
857
+ "p99": 0.037929003295898434,
858
+ "unit": "seconds"
859
+ },
860
+ "tokens_per_second": {
861
+ "name": "tokens_per_second",
862
+ "measurements": [
863
+ 27.14741090516003,
864
+ 27.091272985582627,
865
+ 27.028451998861584,
866
+ 27.09466356842011,
867
+ 27.074781988845007
868
+ ],
869
+ "mean": 27.087316289373877,
870
+ "median": 27.091272985582627,
871
+ "std": 0.03821809568259638,
872
+ "min": 27.028451998861584,
873
+ "max": 27.14741090516003,
874
+ "p25": 27.074781988845007,
875
+ "p75": 27.09466356842011,
876
+ "p90": 27.126311970464062,
877
+ "p95": 27.136861437812044,
878
+ "p99": 27.14530101169043,
879
+ "unit": "tokens/sec"
880
+ },
881
+ "time_per_output_token_seconds": {
882
+ "name": "time_per_output_token_seconds",
883
+ "measurements": [
884
+ 0.03683592529296875,
885
+ 0.036912255859375,
886
+ 0.03699804931640625,
887
+ 0.03690763671875,
888
+ 0.036934738769531246
889
+ ],
890
+ "mean": 0.036917721191406246,
891
+ "median": 0.036912255859375,
892
+ "std": 5.208520149796758e-05,
893
+ "min": 0.03683592529296875,
894
+ "max": 0.03699804931640625,
895
+ "p25": 0.03690763671875,
896
+ "p75": 0.036934738769531246,
897
+ "p90": 0.03697272509765625,
898
+ "p95": 0.03698538720703125,
899
+ "p99": 0.03699551689453125,
900
+ "unit": "seconds/token"
901
+ }
902
+ },
903
+ "gpu_metrics": {
904
+ "gpu_utilization_mean": 90.54838709677419,
905
+ "gpu_utilization_max": 94,
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": "2025-10-01T16:52:01.366285",
918
+ "commit_id": "94bbf8e19911a3aed089d47094c836a18717c9c0",
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
+ 3.826435302734375,
952
+ 3.82772607421875,
953
+ 3.820493408203125,
954
+ 3.840164794921875,
955
+ 3.8404189453125
956
+ ],
957
+ "mean": 3.831047705078125,
958
+ "median": 3.82772607421875,
959
+ "std": 0.007932744042531534,
960
+ "min": 3.820493408203125,
961
+ "max": 3.8404189453125,
962
+ "p25": 3.826435302734375,
963
+ "p75": 3.840164794921875,
964
+ "p90": 3.84031728515625,
965
+ "p95": 3.840368115234375,
966
+ "p99": 3.840408779296875,
967
+ "unit": "seconds"
968
+ },
969
+ "time_to_first_token_seconds": {
970
+ "name": "time_to_first_token_seconds",
971
+ "measurements": [
972
+ 0.040234432220458985,
973
+ 0.039510879516601566,
974
+ 0.03980179214477539,
975
+ 0.03963321685791016,
976
+ 0.03985980987548828
977
+ ],
978
+ "mean": 0.039808026123046875,
979
+ "median": 0.03980179214477539,
980
+ "std": 0.00024633187544929316,
981
+ "min": 0.039510879516601566,
982
+ "max": 0.040234432220458985,
983
+ "p25": 0.03963321685791016,
984
+ "p75": 0.03985980987548828,
985
+ "p90": 0.04008458328247071,
986
+ "p95": 0.040159507751464846,
987
+ "p99": 0.040219447326660156,
988
+ "unit": "seconds"
989
+ },
990
+ "tokens_per_second": {
991
+ "name": "tokens_per_second",
992
+ "measurements": [
993
+ 26.133984266907607,
994
+ 26.12517146238326,
995
+ 26.17462963953458,
996
+ 26.040549127536703,
997
+ 26.038825821869512
998
+ ],
999
+ "mean": 26.102632063646336,
1000
+ "median": 26.12517146238326,
1001
+ "std": 0.05403732695871759,
1002
+ "min": 26.038825821869512,
1003
+ "max": 26.17462963953458,
1004
+ "p25": 26.040549127536703,
1005
+ "p75": 26.133984266907607,
1006
+ "p90": 26.15837149048379,
1007
+ "p95": 26.166500565009184,
1008
+ "p99": 26.1730038246295,
1009
+ "unit": "tokens/sec"
1010
+ },
1011
+ "time_per_output_token_seconds": {
1012
+ "name": "time_per_output_token_seconds",
1013
+ "measurements": [
1014
+ 0.038264353027343746,
1015
+ 0.0382772607421875,
1016
+ 0.03820493408203125,
1017
+ 0.03840164794921875,
1018
+ 0.038404189453125
1019
+ ],
1020
+ "mean": 0.03831047705078126,
1021
+ "median": 0.0382772607421875,
1022
+ "std": 7.932744042531495e-05,
1023
+ "min": 0.03820493408203125,
1024
+ "max": 0.038404189453125,
1025
+ "p25": 0.038264353027343746,
1026
+ "p75": 0.03840164794921875,
1027
+ "p90": 0.0384031728515625,
1028
+ "p95": 0.03840368115234375,
1029
+ "p99": 0.03840408779296875,
1030
+ "unit": "seconds/token"
1031
+ }
1032
+ },
1033
+ "gpu_metrics": {
1034
+ "gpu_utilization_mean": 90.4020618556701,
1035
+ "gpu_utilization_max": 93,
1036
+ "gpu_utilization_min": 85,
1037
+ "gpu_memory_used_mean": 13845,
1038
+ "gpu_memory_used_max": 13845,
1039
+ "gpu_memory_used_min": 13845,
1040
+ "sample_count": 97,
1041
+ "gpu_monitoring_status": "success"
1042
+ }
1043
+ },
1044
+ {
1045
+ "scenario_name": "compiled_compile_max-autotune_sdpa_efficient_attention",
1046
+ "metadata": {
1047
+ "timestamp": "2025-10-01T16:52:42.128119",
1048
+ "commit_id": "94bbf8e19911a3aed089d47094c836a18717c9c0",
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
+ 3.690845947265625,
1082
+ 3.68854541015625,
1083
+ 3.697932861328125,
1084
+ 3.701249755859375,
1085
+ 3.693765380859375
1086
+ ],
1087
+ "mean": 3.6944678710937504,
1088
+ "median": 3.693765380859375,
1089
+ "std": 0.004619252896321583,
1090
+ "min": 3.68854541015625,
1091
+ "max": 3.701249755859375,
1092
+ "p25": 3.690845947265625,
1093
+ "p75": 3.697932861328125,
1094
+ "p90": 3.699922998046875,
1095
+ "p95": 3.700586376953125,
1096
+ "p99": 3.701117080078125,
1097
+ "unit": "seconds"
1098
+ },
1099
+ "time_to_first_token_seconds": {
1100
+ "name": "time_to_first_token_seconds",
1101
+ "measurements": [
1102
+ 0.038093536376953126,
1103
+ 0.03814828872680664,
1104
+ 0.03771395111083985,
1105
+ 0.037588062286376955,
1106
+ 0.037897342681884764
1107
+ ],
1108
+ "mean": 0.03788823623657227,
1109
+ "median": 0.037897342681884764,
1110
+ "std": 0.00021463367773541528,
1111
+ "min": 0.037588062286376955,
1112
+ "max": 0.03814828872680664,
1113
+ "p25": 0.03771395111083985,
1114
+ "p75": 0.038093536376953126,
1115
+ "p90": 0.03812638778686523,
1116
+ "p95": 0.038137338256835934,
1117
+ "p99": 0.0381460986328125,
1118
+ "unit": "seconds"
1119
+ },
1120
+ "tokens_per_second": {
1121
+ "name": "tokens_per_second",
1122
+ "measurements": [
1123
+ 27.09405958113351,
1124
+ 27.110958082461,
1125
+ 27.042135092762248,
1126
+ 27.017901140470727,
1127
+ 27.072645306111575
1128
+ ],
1129
+ "mean": 27.067539840587813,
1130
+ "median": 27.072645306111575,
1131
+ "std": 0.033834621042886204,
1132
+ "min": 27.017901140470727,
1133
+ "max": 27.110958082461,
1134
+ "p25": 27.042135092762248,
1135
+ "p75": 27.09405958113351,
1136
+ "p90": 27.104198681930004,
1137
+ "p95": 27.107578382195502,
1138
+ "p99": 27.1102821424079,
1139
+ "unit": "tokens/sec"
1140
+ },
1141
+ "time_per_output_token_seconds": {
1142
+ "name": "time_per_output_token_seconds",
1143
+ "measurements": [
1144
+ 0.03690845947265625,
1145
+ 0.0368854541015625,
1146
+ 0.03697932861328125,
1147
+ 0.037012497558593746,
1148
+ 0.03693765380859375
1149
+ ],
1150
+ "mean": 0.0369446787109375,
1151
+ "median": 0.03693765380859375,
1152
+ "std": 4.619252896321448e-05,
1153
+ "min": 0.0368854541015625,
1154
+ "max": 0.037012497558593746,
1155
+ "p25": 0.03690845947265625,
1156
+ "p75": 0.03697932861328125,
1157
+ "p90": 0.03699922998046875,
1158
+ "p95": 0.037005863769531244,
1159
+ "p99": 0.03701117080078124,
1160
+ "unit": "seconds/token"
1161
+ }
1162
+ },
1163
+ "gpu_metrics": {
1164
+ "gpu_utilization_mean": 90.45161290322581,
1165
+ "gpu_utilization_max": 93,
1166
+ "gpu_utilization_min": 89,
1167
+ "gpu_memory_used_mean": 13845,
1168
+ "gpu_memory_used_max": 13845,
1169
+ "gpu_memory_used_min": 13845,
1170
+ "sample_count": 93,
1171
+ "gpu_monitoring_status": "success"
1172
+ }
1173
+ }
1174
+ ]
1175
+ }