hf-transformers-bot commited on
Commit
5b3724e
·
verified ·
1 Parent(s): d8c44f5

Upload benchmark results for run 17952921112

Browse files
2025-09-23/17952921112/benchmark_results/Llama-2-7b-hf/Llama-2-7b-hf_benchmark_20250923_164910.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": "2025-09-23T16:43:09.856077",
8
+ "commit_id": "869735d37d0f929311ac6611728c482a4414ba8c",
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
+ 3.646213134765625,
42
+ 3.637670654296875,
43
+ 3.657628662109375,
44
+ 3.638285400390625,
45
+ 3.63398291015625
46
+ ],
47
+ "mean": 3.6427561523437495,
48
+ "median": 3.638285400390625,
49
+ "std": 0.008437558486735863,
50
+ "min": 3.63398291015625,
51
+ "max": 3.657628662109375,
52
+ "p25": 3.637670654296875,
53
+ "p75": 3.646213134765625,
54
+ "p90": 3.6530624511718752,
55
+ "p95": 3.655345556640625,
56
+ "p99": 3.657172041015625,
57
+ "unit": "seconds"
58
+ },
59
+ "time_to_first_token_seconds": {
60
+ "name": "time_to_first_token_seconds",
61
+ "measurements": [
62
+ 0.03761471939086914,
63
+ 0.03713942337036133,
64
+ 0.03696854400634766,
65
+ 0.037187232971191406,
66
+ 0.03697103881835938
67
+ ],
68
+ "mean": 0.03717619171142579,
69
+ "median": 0.03713942337036133,
70
+ "std": 0.00023621415068430932,
71
+ "min": 0.03696854400634766,
72
+ "max": 0.03761471939086914,
73
+ "p25": 0.03697103881835938,
74
+ "p75": 0.037187232971191406,
75
+ "p90": 0.03744372482299805,
76
+ "p95": 0.037529222106933595,
77
+ "p99": 0.03759761993408203,
78
+ "unit": "seconds"
79
+ },
80
+ "tokens_per_second": {
81
+ "name": "tokens_per_second",
82
+ "measurements": [
83
+ 27.425714379263216,
84
+ 27.49011922832497,
85
+ 27.3401182126371,
86
+ 27.485474336142925,
87
+ 27.51801603703753
88
+ ],
89
+ "mean": 27.451888438681145,
90
+ "median": 27.485474336142925,
91
+ "std": 0.06346673123115289,
92
+ "min": 27.3401182126371,
93
+ "max": 27.51801603703753,
94
+ "p25": 27.425714379263216,
95
+ "p75": 27.49011922832497,
96
+ "p90": 27.506857313552505,
97
+ "p95": 27.512436675295017,
98
+ "p99": 27.516900164689027,
99
+ "unit": "tokens/sec"
100
+ },
101
+ "time_per_output_token_seconds": {
102
+ "name": "time_per_output_token_seconds",
103
+ "measurements": [
104
+ 0.036462131347656246,
105
+ 0.03637670654296875,
106
+ 0.03657628662109375,
107
+ 0.03638285400390625,
108
+ 0.0363398291015625
109
+ ],
110
+ "mean": 0.0364275615234375,
111
+ "median": 0.03638285400390625,
112
+ "std": 8.437558486735874e-05,
113
+ "min": 0.0363398291015625,
114
+ "max": 0.03657628662109375,
115
+ "p25": 0.03637670654296875,
116
+ "p75": 0.036462131347656246,
117
+ "p90": 0.03653062451171875,
118
+ "p95": 0.036553455566406254,
119
+ "p99": 0.03657172041015625,
120
+ "unit": "seconds/token"
121
+ }
122
+ },
123
+ "gpu_metrics": {
124
+ "gpu_utilization_mean": 91.20652173913044,
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": "2025-09-23T16:43:45.155576",
138
+ "commit_id": "869735d37d0f929311ac6611728c482a4414ba8c",
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
+ 7.7824306640625,
172
+ 7.82115087890625,
173
+ 7.86147509765625,
174
+ 7.9009521484375,
175
+ 7.91982177734375
176
+ ],
177
+ "mean": 7.8571661132812505,
178
+ "median": 7.86147509765625,
179
+ "std": 0.05048564337943163,
180
+ "min": 7.7824306640625,
181
+ "max": 7.91982177734375,
182
+ "p25": 7.82115087890625,
183
+ "p75": 7.9009521484375,
184
+ "p90": 7.91227392578125,
185
+ "p95": 7.9160478515625,
186
+ "p99": 7.9190669921875,
187
+ "unit": "seconds"
188
+ },
189
+ "time_to_first_token_seconds": {
190
+ "name": "time_to_first_token_seconds",
191
+ "measurements": [
192
+ 0.03892671966552735,
193
+ 0.03819084930419922,
194
+ 0.03806793594360352,
195
+ 0.03866739273071289,
196
+ 0.03806281661987305
197
+ ],
198
+ "mean": 0.0383831428527832,
199
+ "median": 0.03819084930419922,
200
+ "std": 0.00035077481219955456,
201
+ "min": 0.03806281661987305,
202
+ "max": 0.03892671966552735,
203
+ "p25": 0.03806793594360352,
204
+ "p75": 0.03866739273071289,
205
+ "p90": 0.038822988891601566,
206
+ "p95": 0.03887485427856446,
207
+ "p99": 0.03891634658813477,
208
+ "unit": "seconds"
209
+ },
210
+ "tokens_per_second": {
211
+ "name": "tokens_per_second",
212
+ "measurements": [
213
+ 12.849455949768666,
214
+ 12.78584207724484,
215
+ 12.7202590808706,
216
+ 12.656702397542821,
217
+ 12.626546759684693
218
+ ],
219
+ "mean": 12.727761253022324,
220
+ "median": 12.7202590808706,
221
+ "std": 0.08189383194013498,
222
+ "min": 12.626546759684693,
223
+ "max": 12.849455949768666,
224
+ "p25": 12.656702397542821,
225
+ "p75": 12.78584207724484,
226
+ "p90": 12.824010400759136,
227
+ "p95": 12.836733175263902,
228
+ "p99": 12.846911394867714,
229
+ "unit": "tokens/sec"
230
+ },
231
+ "time_per_output_token_seconds": {
232
+ "name": "time_per_output_token_seconds",
233
+ "measurements": [
234
+ 0.077824306640625,
235
+ 0.07821150878906251,
236
+ 0.0786147509765625,
237
+ 0.079009521484375,
238
+ 0.0791982177734375
239
+ ],
240
+ "mean": 0.0785716611328125,
241
+ "median": 0.0786147509765625,
242
+ "std": 0.0005048564337943156,
243
+ "min": 0.077824306640625,
244
+ "max": 0.0791982177734375,
245
+ "p25": 0.07821150878906251,
246
+ "p75": 0.079009521484375,
247
+ "p90": 0.07912273925781249,
248
+ "p95": 0.079160478515625,
249
+ "p99": 0.079190669921875,
250
+ "unit": "seconds/token"
251
+ }
252
+ },
253
+ "gpu_metrics": {
254
+ "gpu_utilization_mean": 45.10059171597633,
255
+ "gpu_utilization_max": 93,
256
+ "gpu_utilization_min": 0,
257
+ "gpu_memory_used_mean": 13630.68047337278,
258
+ "gpu_memory_used_max": 13777,
259
+ "gpu_memory_used_min": 13415,
260
+ "sample_count": 169,
261
+ "gpu_monitoring_status": "success"
262
+ }
263
+ },
264
+ {
265
+ "scenario_name": "eager_sdpa_default",
266
+ "metadata": {
267
+ "timestamp": "2025-09-23T16:45:12.762480",
268
+ "commit_id": "869735d37d0f929311ac6611728c482a4414ba8c",
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
+ 3.40424609375,
302
+ 3.404853759765625,
303
+ 3.408125244140625,
304
+ 3.40607568359375,
305
+ 3.40654150390625
306
+ ],
307
+ "mean": 3.40596845703125,
308
+ "median": 3.40607568359375,
309
+ "std": 0.0013565173171330685,
310
+ "min": 3.40424609375,
311
+ "max": 3.408125244140625,
312
+ "p25": 3.404853759765625,
313
+ "p75": 3.40654150390625,
314
+ "p90": 3.407491748046875,
315
+ "p95": 3.4078084960937503,
316
+ "p99": 3.40806189453125,
317
+ "unit": "seconds"
318
+ },
319
+ "time_to_first_token_seconds": {
320
+ "name": "time_to_first_token_seconds",
321
+ "measurements": [
322
+ 0.03507455825805664,
323
+ 0.03457807922363281,
324
+ 0.034545951843261716,
325
+ 0.03452083206176758,
326
+ 0.034538272857666016
327
+ ],
328
+ "mean": 0.03465153884887695,
329
+ "median": 0.034545951843261716,
330
+ "std": 0.00021232230302065884,
331
+ "min": 0.03452083206176758,
332
+ "max": 0.03507455825805664,
333
+ "p25": 0.034538272857666016,
334
+ "p75": 0.03457807922363281,
335
+ "p90": 0.034875966644287104,
336
+ "p95": 0.03497526245117187,
337
+ "p99": 0.03505469909667969,
338
+ "unit": "seconds"
339
+ },
340
+ "tokens_per_second": {
341
+ "name": "tokens_per_second",
342
+ "measurements": [
343
+ 29.375079605318266,
344
+ 29.369837019632687,
345
+ 29.341644698041453,
346
+ 29.359300640815476,
347
+ 29.355285965349584
348
+ ],
349
+ "mean": 29.360229585831497,
350
+ "median": 29.359300640815476,
351
+ "std": 0.011692101945178128,
352
+ "min": 29.341644698041453,
353
+ "max": 29.375079605318266,
354
+ "p25": 29.355285965349584,
355
+ "p75": 29.369837019632687,
356
+ "p90": 29.372982571044034,
357
+ "p95": 29.37403108818115,
358
+ "p99": 29.374869901890843,
359
+ "unit": "tokens/sec"
360
+ },
361
+ "time_per_output_token_seconds": {
362
+ "name": "time_per_output_token_seconds",
363
+ "measurements": [
364
+ 0.0340424609375,
365
+ 0.034048537597656255,
366
+ 0.03408125244140625,
367
+ 0.0340607568359375,
368
+ 0.0340654150390625
369
+ ],
370
+ "mean": 0.0340596845703125,
371
+ "median": 0.0340607568359375,
372
+ "std": 1.3565173171330372e-05,
373
+ "min": 0.0340424609375,
374
+ "max": 0.03408125244140625,
375
+ "p25": 0.034048537597656255,
376
+ "p75": 0.0340654150390625,
377
+ "p90": 0.03407491748046875,
378
+ "p95": 0.0340780849609375,
379
+ "p99": 0.0340806189453125,
380
+ "unit": "seconds/token"
381
+ }
382
+ },
383
+ "gpu_metrics": {
384
+ "gpu_utilization_mean": 96.87209302325581,
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": "2025-09-23T16:45:45.127135",
398
+ "commit_id": "869735d37d0f929311ac6611728c482a4414ba8c",
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
+ 3.521896728515625,
432
+ 3.52783203125,
433
+ 3.531720703125,
434
+ 3.52489208984375,
435
+ 3.52904736328125
436
+ ],
437
+ "mean": 3.527077783203125,
438
+ "median": 3.52783203125,
439
+ "std": 0.003394861538778413,
440
+ "min": 3.521896728515625,
441
+ "max": 3.531720703125,
442
+ "p25": 3.52489208984375,
443
+ "p75": 3.52904736328125,
444
+ "p90": 3.5306513671874997,
445
+ "p95": 3.53118603515625,
446
+ "p99": 3.53161376953125,
447
+ "unit": "seconds"
448
+ },
449
+ "time_to_first_token_seconds": {
450
+ "name": "time_to_first_token_seconds",
451
+ "measurements": [
452
+ 0.03917308807373047,
453
+ 0.038193023681640625,
454
+ 0.0372498550415039,
455
+ 0.03809782409667969,
456
+ 0.03847139358520508
457
+ ],
458
+ "mean": 0.038237036895751954,
459
+ "median": 0.038193023681640625,
460
+ "std": 0.0006207995240792244,
461
+ "min": 0.0372498550415039,
462
+ "max": 0.03917308807373047,
463
+ "p25": 0.03809782409667969,
464
+ "p75": 0.03847139358520508,
465
+ "p90": 0.038892410278320313,
466
+ "p95": 0.03903274917602539,
467
+ "p99": 0.03914502029418945,
468
+ "unit": "seconds"
469
+ },
470
+ "tokens_per_second": {
471
+ "name": "tokens_per_second",
472
+ "measurements": [
473
+ 28.393791104189198,
474
+ 28.346020761245676,
475
+ 28.314809806878618,
476
+ 28.369662801346283,
477
+ 28.336258969055507
478
+ ],
479
+ "mean": 28.35210868854306,
480
+ "median": 28.346020761245676,
481
+ "std": 0.027294884833741782,
482
+ "min": 28.314809806878618,
483
+ "max": 28.393791104189198,
484
+ "p25": 28.336258969055507,
485
+ "p75": 28.369662801346283,
486
+ "p90": 28.38413978305203,
487
+ "p95": 28.388965443620616,
488
+ "p99": 28.39282597207548,
489
+ "unit": "tokens/sec"
490
+ },
491
+ "time_per_output_token_seconds": {
492
+ "name": "time_per_output_token_seconds",
493
+ "measurements": [
494
+ 0.03521896728515625,
495
+ 0.0352783203125,
496
+ 0.03531720703125,
497
+ 0.0352489208984375,
498
+ 0.0352904736328125
499
+ ],
500
+ "mean": 0.03527077783203125,
501
+ "median": 0.0352783203125,
502
+ "std": 3.3948615387784467e-05,
503
+ "min": 0.03521896728515625,
504
+ "max": 0.03531720703125,
505
+ "p25": 0.0352489208984375,
506
+ "p75": 0.0352904736328125,
507
+ "p90": 0.035306513671875,
508
+ "p95": 0.0353118603515625,
509
+ "p99": 0.0353161376953125,
510
+ "unit": "seconds/token"
511
+ }
512
+ },
513
+ "gpu_metrics": {
514
+ "gpu_utilization_mean": 96.20224719101124,
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": 89,
521
+ "gpu_monitoring_status": "success"
522
+ }
523
+ },
524
+ {
525
+ "scenario_name": "eager_sdpa_flash_attention",
526
+ "metadata": {
527
+ "timestamp": "2025-09-23T16:46:18.464792",
528
+ "commit_id": "869735d37d0f929311ac6611728c482a4414ba8c",
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
+ 3.406349609375,
562
+ 3.40407275390625,
563
+ 3.40154443359375,
564
+ 3.40117724609375,
565
+ 3.403020751953125
566
+ ],
567
+ "mean": 3.4032329589843746,
568
+ "median": 3.403020751953125,
569
+ "std": 0.0018730120381763807,
570
+ "min": 3.40117724609375,
571
+ "max": 3.406349609375,
572
+ "p25": 3.40154443359375,
573
+ "p75": 3.40407275390625,
574
+ "p90": 3.4054388671874998,
575
+ "p95": 3.40589423828125,
576
+ "p99": 3.4062585351562498,
577
+ "unit": "seconds"
578
+ },
579
+ "time_to_first_token_seconds": {
580
+ "name": "time_to_first_token_seconds",
581
+ "measurements": [
582
+ 0.03515059280395508,
583
+ 0.03457772827148437,
584
+ 0.03460761642456055,
585
+ 0.034565406799316405,
586
+ 0.03472812652587891
587
+ ],
588
+ "mean": 0.03472589416503906,
589
+ "median": 0.03460761642456055,
590
+ "std": 0.0002200331192317175,
591
+ "min": 0.034565406799316405,
592
+ "max": 0.03515059280395508,
593
+ "p25": 0.03457772827148437,
594
+ "p75": 0.03472812652587891,
595
+ "p90": 0.03498160629272461,
596
+ "p95": 0.03506609954833984,
597
+ "p99": 0.03513369415283203,
598
+ "unit": "seconds"
599
+ },
600
+ "tokens_per_second": {
601
+ "name": "tokens_per_second",
602
+ "measurements": [
603
+ 29.356939676649365,
604
+ 29.376575422851275,
605
+ 29.398410619716486,
606
+ 29.401584441049035,
607
+ 29.38565682933498
608
+ ],
609
+ "mean": 29.383833397920228,
610
+ "median": 29.38565682933498,
611
+ "std": 0.016167066991132448,
612
+ "min": 29.356939676649365,
613
+ "max": 29.401584441049035,
614
+ "p25": 29.376575422851275,
615
+ "p75": 29.398410619716486,
616
+ "p90": 29.400314912516016,
617
+ "p95": 29.400949676782524,
618
+ "p99": 29.40145748819573,
619
+ "unit": "tokens/sec"
620
+ },
621
+ "time_per_output_token_seconds": {
622
+ "name": "time_per_output_token_seconds",
623
+ "measurements": [
624
+ 0.03406349609375,
625
+ 0.0340407275390625,
626
+ 0.0340154443359375,
627
+ 0.0340117724609375,
628
+ 0.03403020751953125
629
+ ],
630
+ "mean": 0.03403232958984375,
631
+ "median": 0.03403020751953125,
632
+ "std": 1.8730120381764702e-05,
633
+ "min": 0.0340117724609375,
634
+ "max": 0.03406349609375,
635
+ "p25": 0.0340154443359375,
636
+ "p75": 0.0340407275390625,
637
+ "p90": 0.034054388671875,
638
+ "p95": 0.0340589423828125,
639
+ "p99": 0.0340625853515625,
640
+ "unit": "seconds/token"
641
+ }
642
+ },
643
+ "gpu_metrics": {
644
+ "gpu_utilization_mean": 96.84883720930233,
645
+ "gpu_utilization_max": 98,
646
+ "gpu_utilization_min": 94,
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": "2025-09-23T16:46:50.820078",
658
+ "commit_id": "869735d37d0f929311ac6611728c482a4414ba8c",
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
+ 3.40277001953125,
692
+ 3.402108154296875,
693
+ 3.401876953125,
694
+ 3.40280419921875,
695
+ 3.403853515625
696
+ ],
697
+ "mean": 3.402682568359375,
698
+ "median": 3.40277001953125,
699
+ "std": 0.0006888434790783525,
700
+ "min": 3.401876953125,
701
+ "max": 3.403853515625,
702
+ "p25": 3.402108154296875,
703
+ "p75": 3.40280419921875,
704
+ "p90": 3.4034337890625,
705
+ "p95": 3.40364365234375,
706
+ "p99": 3.4038115429687497,
707
+ "unit": "seconds"
708
+ },
709
+ "time_to_first_token_seconds": {
710
+ "name": "time_to_first_token_seconds",
711
+ "measurements": [
712
+ 0.035182048797607425,
713
+ 0.0346794548034668,
714
+ 0.03526700973510742,
715
+ 0.0346583366394043,
716
+ 0.03467619323730469
717
+ ],
718
+ "mean": 0.03489260864257813,
719
+ "median": 0.0346794548034668,
720
+ "std": 0.00027243543810200445,
721
+ "min": 0.0346583366394043,
722
+ "max": 0.03526700973510742,
723
+ "p25": 0.03467619323730469,
724
+ "p75": 0.035182048797607425,
725
+ "p90": 0.035233025360107424,
726
+ "p95": 0.03525001754760742,
727
+ "p99": 0.03526361129760742,
728
+ "unit": "seconds"
729
+ },
730
+ "tokens_per_second": {
731
+ "name": "tokens_per_second",
732
+ "measurements": [
733
+ 29.38782210552553,
734
+ 29.393539377547313,
735
+ 29.395537045553322,
736
+ 29.387526917640162,
737
+ 29.378467534211286
738
+ ],
739
+ "mean": 29.38857859609552,
740
+ "median": 29.38782210552553,
741
+ "std": 0.005948802391271814,
742
+ "min": 29.378467534211286,
743
+ "max": 29.395537045553322,
744
+ "p25": 29.387526917640162,
745
+ "p75": 29.393539377547313,
746
+ "p90": 29.394737978350918,
747
+ "p95": 29.39513751195212,
748
+ "p99": 29.39545713883308,
749
+ "unit": "tokens/sec"
750
+ },
751
+ "time_per_output_token_seconds": {
752
+ "name": "time_per_output_token_seconds",
753
+ "measurements": [
754
+ 0.034027700195312496,
755
+ 0.03402108154296875,
756
+ 0.03401876953125,
757
+ 0.0340280419921875,
758
+ 0.03403853515625
759
+ ],
760
+ "mean": 0.03402682568359375,
761
+ "median": 0.034027700195312496,
762
+ "std": 6.888434790784534e-06,
763
+ "min": 0.03401876953125,
764
+ "max": 0.03403853515625,
765
+ "p25": 0.03402108154296875,
766
+ "p75": 0.0340280419921875,
767
+ "p90": 0.034034337890625,
768
+ "p95": 0.0340364365234375,
769
+ "p99": 0.0340381154296875,
770
+ "unit": "seconds/token"
771
+ }
772
+ },
773
+ "gpu_metrics": {
774
+ "gpu_utilization_mean": 97.01162790697674,
775
+ "gpu_utilization_max": 98,
776
+ "gpu_utilization_min": 94,
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": "2025-09-23T16:47:23.199928",
788
+ "commit_id": "869735d37d0f929311ac6611728c482a4414ba8c",
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
+ 3.690652587890625,
822
+ 3.679438232421875,
823
+ 3.6811142578125,
824
+ 3.688697509765625,
825
+ 3.680710693359375
826
+ ],
827
+ "mean": 3.6841226562499996,
828
+ "median": 3.6811142578125,
829
+ "std": 0.004608797466756947,
830
+ "min": 3.679438232421875,
831
+ "max": 3.690652587890625,
832
+ "p25": 3.680710693359375,
833
+ "p75": 3.688697509765625,
834
+ "p90": 3.689870556640625,
835
+ "p95": 3.690261572265625,
836
+ "p99": 3.690574384765625,
837
+ "unit": "seconds"
838
+ },
839
+ "time_to_first_token_seconds": {
840
+ "name": "time_to_first_token_seconds",
841
+ "measurements": [
842
+ 0.0381126708984375,
843
+ 0.037270687103271485,
844
+ 0.03757932662963867,
845
+ 0.03728675079345703,
846
+ 0.03736966323852539
847
+ ],
848
+ "mean": 0.03752381973266601,
849
+ "median": 0.03736966323852539,
850
+ "std": 0.00031428291258977446,
851
+ "min": 0.037270687103271485,
852
+ "max": 0.0381126708984375,
853
+ "p25": 0.03728675079345703,
854
+ "p75": 0.03757932662963867,
855
+ "p90": 0.03789933319091797,
856
+ "p95": 0.03800600204467773,
857
+ "p99": 0.038091337127685546,
858
+ "unit": "seconds"
859
+ },
860
+ "tokens_per_second": {
861
+ "name": "tokens_per_second",
862
+ "measurements": [
863
+ 27.09547908359332,
864
+ 27.178061889675515,
865
+ 27.16568761422389,
866
+ 27.109840190271896,
867
+ 27.168666143855564
868
+ ],
869
+ "mean": 27.143546984324036,
870
+ "median": 27.16568761422389,
871
+ "std": 0.033938560385748405,
872
+ "min": 27.09547908359332,
873
+ "max": 27.178061889675515,
874
+ "p25": 27.109840190271896,
875
+ "p75": 27.168666143855564,
876
+ "p90": 27.174303591347535,
877
+ "p95": 27.176182740511525,
878
+ "p99": 27.177686059842717,
879
+ "unit": "tokens/sec"
880
+ },
881
+ "time_per_output_token_seconds": {
882
+ "name": "time_per_output_token_seconds",
883
+ "measurements": [
884
+ 0.03690652587890625,
885
+ 0.03679438232421875,
886
+ 0.036811142578125,
887
+ 0.036886975097656254,
888
+ 0.03680710693359375
889
+ ],
890
+ "mean": 0.036841226562499996,
891
+ "median": 0.036811142578125,
892
+ "std": 4.608797466756932e-05,
893
+ "min": 0.03679438232421875,
894
+ "max": 0.03690652587890625,
895
+ "p25": 0.03680710693359375,
896
+ "p75": 0.036886975097656254,
897
+ "p90": 0.03689870556640625,
898
+ "p95": 0.03690261572265625,
899
+ "p99": 0.03690574384765625,
900
+ "unit": "seconds/token"
901
+ }
902
+ },
903
+ "gpu_metrics": {
904
+ "gpu_utilization_mean": 90.58064516129032,
905
+ "gpu_utilization_max": 94,
906
+ "gpu_utilization_min": 87,
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": "2025-09-23T16:47:57.826847",
918
+ "commit_id": "869735d37d0f929311ac6611728c482a4414ba8c",
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
+ 3.8264716796875,
952
+ 3.8284970703125,
953
+ 3.846236572265625,
954
+ 3.90469140625,
955
+ 3.898036865234375
956
+ ],
957
+ "mean": 3.86078671875,
958
+ "median": 3.846236572265625,
959
+ "std": 0.033902935160174855,
960
+ "min": 3.8264716796875,
961
+ "max": 3.90469140625,
962
+ "p25": 3.8284970703125,
963
+ "p75": 3.898036865234375,
964
+ "p90": 3.90202958984375,
965
+ "p95": 3.903360498046875,
966
+ "p99": 3.904425224609375,
967
+ "unit": "seconds"
968
+ },
969
+ "time_to_first_token_seconds": {
970
+ "name": "time_to_first_token_seconds",
971
+ "measurements": [
972
+ 0.040523681640625,
973
+ 0.040086624145507815,
974
+ 0.03949939346313477,
975
+ 0.040279327392578126,
976
+ 0.041201057434082033
977
+ ],
978
+ "mean": 0.040318016815185556,
979
+ "median": 0.040279327392578126,
980
+ "std": 0.0005562809338040034,
981
+ "min": 0.03949939346313477,
982
+ "max": 0.041201057434082033,
983
+ "p25": 0.040086624145507815,
984
+ "p75": 0.040523681640625,
985
+ "p90": 0.04093010711669922,
986
+ "p95": 0.04106558227539063,
987
+ "p99": 0.04117396240234375,
988
+ "unit": "seconds"
989
+ },
990
+ "tokens_per_second": {
991
+ "name": "tokens_per_second",
992
+ "measurements": [
993
+ 26.133735820087605,
994
+ 26.119910284230027,
995
+ 25.999440783512444,
996
+ 25.610218477172392,
997
+ 25.653939010140007
998
+ ],
999
+ "mean": 25.903448875028495,
1000
+ "median": 25.999440783512444,
1001
+ "std": 0.22686617382769125,
1002
+ "min": 25.610218477172392,
1003
+ "max": 26.133735820087605,
1004
+ "p25": 25.653939010140007,
1005
+ "p75": 26.119910284230027,
1006
+ "p90": 26.128205605744572,
1007
+ "p95": 26.13097071291609,
1008
+ "p99": 26.1331827986533,
1009
+ "unit": "tokens/sec"
1010
+ },
1011
+ "time_per_output_token_seconds": {
1012
+ "name": "time_per_output_token_seconds",
1013
+ "measurements": [
1014
+ 0.038264716796875,
1015
+ 0.038284970703125,
1016
+ 0.038462365722656246,
1017
+ 0.0390469140625,
1018
+ 0.03898036865234375
1019
+ ],
1020
+ "mean": 0.03860786718749999,
1021
+ "median": 0.038462365722656246,
1022
+ "std": 0.00033902935160174834,
1023
+ "min": 0.038264716796875,
1024
+ "max": 0.0390469140625,
1025
+ "p25": 0.038284970703125,
1026
+ "p75": 0.03898036865234375,
1027
+ "p90": 0.0390202958984375,
1028
+ "p95": 0.03903360498046875,
1029
+ "p99": 0.03904425224609375,
1030
+ "unit": "seconds/token"
1031
+ }
1032
+ },
1033
+ "gpu_metrics": {
1034
+ "gpu_utilization_mean": 89.58762886597938,
1035
+ "gpu_utilization_max": 94,
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": "2025-09-23T16:48:38.789921",
1048
+ "commit_id": "869735d37d0f929311ac6611728c482a4414ba8c",
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
+ 3.68554443359375,
1082
+ 3.67249755859375,
1083
+ 3.68795458984375,
1084
+ 3.6894404296875,
1085
+ 3.6837666015625
1086
+ ],
1087
+ "mean": 3.6838407226562504,
1088
+ "median": 3.68554443359375,
1089
+ "std": 0.005997601485862664,
1090
+ "min": 3.67249755859375,
1091
+ "max": 3.6894404296875,
1092
+ "p25": 3.6837666015625,
1093
+ "p75": 3.68795458984375,
1094
+ "p90": 3.68884609375,
1095
+ "p95": 3.68914326171875,
1096
+ "p99": 3.68938099609375,
1097
+ "unit": "seconds"
1098
+ },
1099
+ "time_to_first_token_seconds": {
1100
+ "name": "time_to_first_token_seconds",
1101
+ "measurements": [
1102
+ 0.03821184158325195,
1103
+ 0.03734479904174805,
1104
+ 0.03731033706665039,
1105
+ 0.037346817016601565,
1106
+ 0.037362560272216794
1107
+ ],
1108
+ "mean": 0.03751527099609375,
1109
+ "median": 0.037346817016601565,
1110
+ "std": 0.0003487023068301199,
1111
+ "min": 0.03731033706665039,
1112
+ "max": 0.03821184158325195,
1113
+ "p25": 0.03734479904174805,
1114
+ "p75": 0.037362560272216794,
1115
+ "p90": 0.03787212905883789,
1116
+ "p95": 0.03804198532104492,
1117
+ "p99": 0.03817787033081054,
1118
+ "unit": "seconds"
1119
+ },
1120
+ "tokens_per_second": {
1121
+ "name": "tokens_per_second",
1122
+ "measurements": [
1123
+ 27.13303334196697,
1124
+ 27.229425861971542,
1125
+ 27.115301331363945,
1126
+ 27.10438124853262,
1127
+ 27.146128084657743
1128
+ ],
1129
+ "mean": 27.145653973698565,
1130
+ "median": 27.13303334196697,
1131
+ "std": 0.04427630321208719,
1132
+ "min": 27.10438124853262,
1133
+ "max": 27.229425861971542,
1134
+ "p25": 27.115301331363945,
1135
+ "p75": 27.146128084657743,
1136
+ "p90": 27.196106751046024,
1137
+ "p95": 27.212766306508783,
1138
+ "p99": 27.22609395087899,
1139
+ "unit": "tokens/sec"
1140
+ },
1141
+ "time_per_output_token_seconds": {
1142
+ "name": "time_per_output_token_seconds",
1143
+ "measurements": [
1144
+ 0.0368554443359375,
1145
+ 0.0367249755859375,
1146
+ 0.0368795458984375,
1147
+ 0.036894404296875,
1148
+ 0.036837666015625
1149
+ ],
1150
+ "mean": 0.036838407226562495,
1151
+ "median": 0.0368554443359375,
1152
+ "std": 5.997601485862639e-05,
1153
+ "min": 0.0367249755859375,
1154
+ "max": 0.036894404296875,
1155
+ "p25": 0.036837666015625,
1156
+ "p75": 0.0368795458984375,
1157
+ "p90": 0.0368884609375,
1158
+ "p95": 0.036891432617187504,
1159
+ "p99": 0.0368938099609375,
1160
+ "unit": "seconds/token"
1161
+ }
1162
+ },
1163
+ "gpu_metrics": {
1164
+ "gpu_utilization_mean": 90.66666666666667,
1165
+ "gpu_utilization_max": 94,
1166
+ "gpu_utilization_min": 88,
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
+ }