hf-transformers-bot commited on
Commit
d4fdf31
·
verified ·
1 Parent(s): dcaefbf

Upload benchmark results for run 18014412306

Browse files
2025-09-25/18014412306/benchmark_results/Llama-2-7b-hf/Llama-2-7b-hf_benchmark_20250925_165323.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": "2025-09-25T16:47:23.759984",
8
+ "commit_id": "e691f84412563b6abca098f3e044980725d8daa3",
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
+ 3.649643798828125,
42
+ 3.645258544921875,
43
+ 3.644918701171875,
44
+ 3.643963134765625,
45
+ 3.6349375
46
+ ],
47
+ "mean": 3.6437443359375004,
48
+ "median": 3.644918701171875,
49
+ "std": 0.0048183785054137255,
50
+ "min": 3.6349375,
51
+ "max": 3.649643798828125,
52
+ "p25": 3.643963134765625,
53
+ "p75": 3.645258544921875,
54
+ "p90": 3.647889697265625,
55
+ "p95": 3.6487667480468753,
56
+ "p99": 3.6494683886718753,
57
+ "unit": "seconds"
58
+ },
59
+ "time_to_first_token_seconds": {
60
+ "name": "time_to_first_token_seconds",
61
+ "measurements": [
62
+ 0.03884719848632812,
63
+ 0.03726467132568359,
64
+ 0.037169055938720705,
65
+ 0.03730495834350586,
66
+ 0.03702259063720703
67
+ ],
68
+ "mean": 0.03752169494628906,
69
+ "median": 0.03726467132568359,
70
+ "std": 0.0006698439985838909,
71
+ "min": 0.03702259063720703,
72
+ "max": 0.03884719848632812,
73
+ "p25": 0.037169055938720705,
74
+ "p75": 0.03730495834350586,
75
+ "p90": 0.03823030242919922,
76
+ "p95": 0.03853875045776367,
77
+ "p99": 0.03878550888061523,
78
+ "unit": "seconds"
79
+ },
80
+ "tokens_per_second": {
81
+ "name": "tokens_per_second",
82
+ "measurements": [
83
+ 27.399934216075906,
84
+ 27.432896396143885,
85
+ 27.43545417565804,
86
+ 27.442648649745976,
87
+ 27.510789387712993
88
+ ],
89
+ "mean": 27.44434456506736,
90
+ "median": 27.43545417565804,
91
+ "std": 0.03633227016931222,
92
+ "min": 27.399934216075906,
93
+ "max": 27.510789387712993,
94
+ "p25": 27.432896396143885,
95
+ "p75": 27.442648649745976,
96
+ "p90": 27.483533092526187,
97
+ "p95": 27.49716124011959,
98
+ "p99": 27.50806375819431,
99
+ "unit": "tokens/sec"
100
+ },
101
+ "time_per_output_token_seconds": {
102
+ "name": "time_per_output_token_seconds",
103
+ "measurements": [
104
+ 0.036496437988281255,
105
+ 0.03645258544921875,
106
+ 0.03644918701171875,
107
+ 0.03643963134765625,
108
+ 0.036349374999999996
109
+ ],
110
+ "mean": 0.036437443359375,
111
+ "median": 0.03644918701171875,
112
+ "std": 4.8183785054139314e-05,
113
+ "min": 0.036349374999999996,
114
+ "max": 0.036496437988281255,
115
+ "p25": 0.03643963134765625,
116
+ "p75": 0.03645258544921875,
117
+ "p90": 0.03647889697265625,
118
+ "p95": 0.036487667480468756,
119
+ "p99": 0.036494683886718754,
120
+ "unit": "seconds/token"
121
+ }
122
+ },
123
+ "gpu_metrics": {
124
+ "gpu_utilization_mean": 91.22826086956522,
125
+ "gpu_utilization_max": 94,
126
+ "gpu_utilization_min": 89,
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": "2025-09-25T16:47:58.921754",
138
+ "commit_id": "e691f84412563b6abca098f3e044980725d8daa3",
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
+ 7.7138740234375,
172
+ 7.75053271484375,
173
+ 7.81276806640625,
174
+ 7.8065888671875,
175
+ 7.86054931640625
176
+ ],
177
+ "mean": 7.788862597656251,
178
+ "median": 7.8065888671875,
179
+ "std": 0.05121941798297678,
180
+ "min": 7.7138740234375,
181
+ "max": 7.86054931640625,
182
+ "p25": 7.75053271484375,
183
+ "p75": 7.81276806640625,
184
+ "p90": 7.84143681640625,
185
+ "p95": 7.85099306640625,
186
+ "p99": 7.85863806640625,
187
+ "unit": "seconds"
188
+ },
189
+ "time_to_first_token_seconds": {
190
+ "name": "time_to_first_token_seconds",
191
+ "measurements": [
192
+ 0.03862262344360352,
193
+ 0.03801769638061524,
194
+ 0.03805795288085938,
195
+ 0.0380035514831543,
196
+ 0.03832931137084961
197
+ ],
198
+ "mean": 0.03820622711181641,
199
+ "median": 0.03805795288085938,
200
+ "std": 0.0002396422894959241,
201
+ "min": 0.0380035514831543,
202
+ "max": 0.03862262344360352,
203
+ "p25": 0.03801769638061524,
204
+ "p75": 0.03832931137084961,
205
+ "p90": 0.03850529861450196,
206
+ "p95": 0.03856396102905273,
207
+ "p99": 0.03861089096069336,
208
+ "unit": "seconds"
209
+ },
210
+ "tokens_per_second": {
211
+ "name": "tokens_per_second",
212
+ "measurements": [
213
+ 12.963654798634817,
214
+ 12.902338933229828,
215
+ 12.799560815069531,
216
+ 12.809692133310365,
217
+ 12.721757217562857
218
+ ],
219
+ "mean": 12.839400779561478,
220
+ "median": 12.809692133310365,
221
+ "std": 0.0845104804809146,
222
+ "min": 12.721757217562857,
223
+ "max": 12.963654798634817,
224
+ "p25": 12.799560815069531,
225
+ "p75": 12.902338933229828,
226
+ "p90": 12.939128452472822,
227
+ "p95": 12.95139162555382,
228
+ "p99": 12.961202164018617,
229
+ "unit": "tokens/sec"
230
+ },
231
+ "time_per_output_token_seconds": {
232
+ "name": "time_per_output_token_seconds",
233
+ "measurements": [
234
+ 0.077138740234375,
235
+ 0.0775053271484375,
236
+ 0.0781276806640625,
237
+ 0.078065888671875,
238
+ 0.0786054931640625
239
+ ],
240
+ "mean": 0.0778886259765625,
241
+ "median": 0.078065888671875,
242
+ "std": 0.000512194179829765,
243
+ "min": 0.077138740234375,
244
+ "max": 0.0786054931640625,
245
+ "p25": 0.0775053271484375,
246
+ "p75": 0.0781276806640625,
247
+ "p90": 0.0784143681640625,
248
+ "p95": 0.0785099306640625,
249
+ "p99": 0.0785863806640625,
250
+ "unit": "seconds/token"
251
+ }
252
+ },
253
+ "gpu_metrics": {
254
+ "gpu_utilization_mean": 44.790419161676645,
255
+ "gpu_utilization_max": 94,
256
+ "gpu_utilization_min": 0,
257
+ "gpu_memory_used_mean": 13629.718562874252,
258
+ "gpu_memory_used_max": 13777,
259
+ "gpu_memory_used_min": 13415,
260
+ "sample_count": 167,
261
+ "gpu_monitoring_status": "success"
262
+ }
263
+ },
264
+ {
265
+ "scenario_name": "eager_sdpa_default",
266
+ "metadata": {
267
+ "timestamp": "2025-09-25T16:49:25.914658",
268
+ "commit_id": "e691f84412563b6abca098f3e044980725d8daa3",
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
+ 3.4065419921875,
302
+ 3.404058349609375,
303
+ 3.404971435546875,
304
+ 3.4098759765625,
305
+ 3.405879150390625
306
+ ],
307
+ "mean": 3.406265380859375,
308
+ "median": 3.405879150390625,
309
+ "std": 0.001990345223496749,
310
+ "min": 3.404058349609375,
311
+ "max": 3.4098759765625,
312
+ "p25": 3.404971435546875,
313
+ "p75": 3.4065419921875,
314
+ "p90": 3.4085423828125,
315
+ "p95": 3.4092091796875,
316
+ "p99": 3.4097426171875003,
317
+ "unit": "seconds"
318
+ },
319
+ "time_to_first_token_seconds": {
320
+ "name": "time_to_first_token_seconds",
321
+ "measurements": [
322
+ 0.035141056060791015,
323
+ 0.03457379150390625,
324
+ 0.034568031311035155,
325
+ 0.034942913055419925,
326
+ 0.03458019256591797
327
+ ],
328
+ "mean": 0.03476119689941406,
329
+ "median": 0.03458019256591797,
330
+ "std": 0.0002377015100411115,
331
+ "min": 0.034568031311035155,
332
+ "max": 0.035141056060791015,
333
+ "p25": 0.03457379150390625,
334
+ "p75": 0.034942913055419925,
335
+ "p90": 0.03506179885864258,
336
+ "p95": 0.0351014274597168,
337
+ "p99": 0.03513313034057617,
338
+ "unit": "seconds"
339
+ },
340
+ "tokens_per_second": {
341
+ "name": "tokens_per_second",
342
+ "measurements": [
343
+ 29.35528175767043,
344
+ 29.376699730037025,
345
+ 29.368821998337534,
346
+ 29.326579819131755,
347
+ 29.360994792939398
348
+ ],
349
+ "mean": 29.35767561962323,
350
+ "median": 29.360994792939398,
351
+ "std": 0.01714558153132399,
352
+ "min": 29.326579819131755,
353
+ "max": 29.376699730037025,
354
+ "p25": 29.35528175767043,
355
+ "p75": 29.368821998337534,
356
+ "p90": 29.373548637357228,
357
+ "p95": 29.375124183697128,
358
+ "p99": 29.376384620769045,
359
+ "unit": "tokens/sec"
360
+ },
361
+ "time_per_output_token_seconds": {
362
+ "name": "time_per_output_token_seconds",
363
+ "measurements": [
364
+ 0.034065419921875,
365
+ 0.03404058349609375,
366
+ 0.03404971435546875,
367
+ 0.034098759765625,
368
+ 0.03405879150390625
369
+ ],
370
+ "mean": 0.03406265380859375,
371
+ "median": 0.03405879150390625,
372
+ "std": 1.990345223496663e-05,
373
+ "min": 0.03404058349609375,
374
+ "max": 0.034098759765625,
375
+ "p25": 0.03404971435546875,
376
+ "p75": 0.034065419921875,
377
+ "p90": 0.034085423828125,
378
+ "p95": 0.034092091796875,
379
+ "p99": 0.034097426171875,
380
+ "unit": "seconds/token"
381
+ }
382
+ },
383
+ "gpu_metrics": {
384
+ "gpu_utilization_mean": 96.96511627906976,
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": "2025-09-25T16:49:58.282857",
398
+ "commit_id": "e691f84412563b6abca098f3e044980725d8daa3",
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
+ 3.52410986328125,
432
+ 3.52261962890625,
433
+ 3.52538037109375,
434
+ 3.52180029296875,
435
+ 3.52892529296875
436
+ ],
437
+ "mean": 3.5245670898437504,
438
+ "median": 3.52410986328125,
439
+ "std": 0.002502486205772723,
440
+ "min": 3.52180029296875,
441
+ "max": 3.52892529296875,
442
+ "p25": 3.52261962890625,
443
+ "p75": 3.52538037109375,
444
+ "p90": 3.52750732421875,
445
+ "p95": 3.5282163085937497,
446
+ "p99": 3.52878349609375,
447
+ "unit": "seconds"
448
+ },
449
+ "time_to_first_token_seconds": {
450
+ "name": "time_to_first_token_seconds",
451
+ "measurements": [
452
+ 0.03744518280029297,
453
+ 0.036663360595703125,
454
+ 0.036929920196533204,
455
+ 0.036744094848632815,
456
+ 0.036948192596435545
457
+ ],
458
+ "mean": 0.03694615020751953,
459
+ "median": 0.036929920196533204,
460
+ "std": 0.0002720652296289977,
461
+ "min": 0.036663360595703125,
462
+ "max": 0.03744518280029297,
463
+ "p25": 0.036744094848632815,
464
+ "p75": 0.036948192596435545,
465
+ "p90": 0.03724638671875,
466
+ "p95": 0.03734578475952149,
467
+ "p99": 0.03742530319213867,
468
+ "unit": "seconds"
469
+ },
470
+ "tokens_per_second": {
471
+ "name": "tokens_per_second",
472
+ "measurements": [
473
+ 28.375959853558985,
474
+ 28.387964223957198,
475
+ 28.36573347374002,
476
+ 28.394568595967613,
477
+ 28.3372391586884
478
+ ],
479
+ "mean": 28.372293061182443,
480
+ "median": 28.375959853558985,
481
+ "std": 0.020134718406212813,
482
+ "min": 28.3372391586884,
483
+ "max": 28.394568595967613,
484
+ "p25": 28.36573347374002,
485
+ "p75": 28.387964223957198,
486
+ "p90": 28.391926847163447,
487
+ "p95": 28.39324772156553,
488
+ "p99": 28.394304421087195,
489
+ "unit": "tokens/sec"
490
+ },
491
+ "time_per_output_token_seconds": {
492
+ "name": "time_per_output_token_seconds",
493
+ "measurements": [
494
+ 0.0352410986328125,
495
+ 0.0352261962890625,
496
+ 0.035253803710937504,
497
+ 0.0352180029296875,
498
+ 0.0352892529296875
499
+ ],
500
+ "mean": 0.0352456708984375,
501
+ "median": 0.0352410986328125,
502
+ "std": 2.5024862057728776e-05,
503
+ "min": 0.0352180029296875,
504
+ "max": 0.0352892529296875,
505
+ "p25": 0.0352261962890625,
506
+ "p75": 0.035253803710937504,
507
+ "p90": 0.0352750732421875,
508
+ "p95": 0.0352821630859375,
509
+ "p99": 0.0352878349609375,
510
+ "unit": "seconds/token"
511
+ }
512
+ },
513
+ "gpu_metrics": {
514
+ "gpu_utilization_mean": 96.51685393258427,
515
+ "gpu_utilization_max": 98,
516
+ "gpu_utilization_min": 93,
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": "2025-09-25T16:50:31.623162",
528
+ "commit_id": "e691f84412563b6abca098f3e044980725d8daa3",
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
+ 3.406857666015625,
562
+ 3.40596044921875,
563
+ 3.40547119140625,
564
+ 3.4057822265625,
565
+ 3.40776318359375
566
+ ],
567
+ "mean": 3.4063669433593753,
568
+ "median": 3.40596044921875,
569
+ "std": 0.0008366359181725522,
570
+ "min": 3.40547119140625,
571
+ "max": 3.40776318359375,
572
+ "p25": 3.4057822265625,
573
+ "p75": 3.406857666015625,
574
+ "p90": 3.4074009765625,
575
+ "p95": 3.407582080078125,
576
+ "p99": 3.4077269628906253,
577
+ "unit": "seconds"
578
+ },
579
+ "time_to_first_token_seconds": {
580
+ "name": "time_to_first_token_seconds",
581
+ "measurements": [
582
+ 0.03509411239624023,
583
+ 0.03458736038208008,
584
+ 0.034702560424804685,
585
+ 0.034568161010742185,
586
+ 0.03471043014526367
587
+ ],
588
+ "mean": 0.03473252487182617,
589
+ "median": 0.034702560424804685,
590
+ "std": 0.000189852464167702,
591
+ "min": 0.034568161010742185,
592
+ "max": 0.03509411239624023,
593
+ "p25": 0.03458736038208008,
594
+ "p75": 0.03471043014526367,
595
+ "p90": 0.03494063949584961,
596
+ "p95": 0.03501737594604492,
597
+ "p99": 0.03507876510620117,
598
+ "unit": "seconds"
599
+ },
600
+ "tokens_per_second": {
601
+ "name": "tokens_per_second",
602
+ "measurements": [
603
+ 29.352561745542957,
604
+ 29.360293958474394,
605
+ 29.364512098164646,
606
+ 29.36183036603937,
607
+ 29.344762124738452
608
+ ],
609
+ "mean": 29.356792058591964,
610
+ "median": 29.360293958474394,
611
+ "std": 0.007209184596574077,
612
+ "min": 29.344762124738452,
613
+ "max": 29.364512098164646,
614
+ "p25": 29.352561745542957,
615
+ "p75": 29.36183036603937,
616
+ "p90": 29.363439405314537,
617
+ "p95": 29.36397575173959,
618
+ "p99": 29.364404828879636,
619
+ "unit": "tokens/sec"
620
+ },
621
+ "time_per_output_token_seconds": {
622
+ "name": "time_per_output_token_seconds",
623
+ "measurements": [
624
+ 0.03406857666015625,
625
+ 0.0340596044921875,
626
+ 0.0340547119140625,
627
+ 0.034057822265625,
628
+ 0.034077631835937504
629
+ ],
630
+ "mean": 0.034063669433593756,
631
+ "median": 0.0340596044921875,
632
+ "std": 8.366359181726534e-06,
633
+ "min": 0.0340547119140625,
634
+ "max": 0.034077631835937504,
635
+ "p25": 0.034057822265625,
636
+ "p75": 0.03406857666015625,
637
+ "p90": 0.034074009765625,
638
+ "p95": 0.034075820800781254,
639
+ "p99": 0.034077269628906254,
640
+ "unit": "seconds/token"
641
+ }
642
+ },
643
+ "gpu_metrics": {
644
+ "gpu_utilization_mean": 96.96511627906976,
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": "2025-09-25T16:51:04.006964",
658
+ "commit_id": "e691f84412563b6abca098f3e044980725d8daa3",
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
+ 3.40701025390625,
692
+ 3.405124267578125,
693
+ 3.402571044921875,
694
+ 3.4063701171875,
695
+ 3.40653466796875
696
+ ],
697
+ "mean": 3.4055220703124993,
698
+ "median": 3.4063701171875,
699
+ "std": 0.0016016257862519216,
700
+ "min": 3.402571044921875,
701
+ "max": 3.40701025390625,
702
+ "p25": 3.405124267578125,
703
+ "p75": 3.40653466796875,
704
+ "p90": 3.4068200195312497,
705
+ "p95": 3.40691513671875,
706
+ "p99": 3.40699123046875,
707
+ "unit": "seconds"
708
+ },
709
+ "time_to_first_token_seconds": {
710
+ "name": "time_to_first_token_seconds",
711
+ "measurements": [
712
+ 0.03523107147216797,
713
+ 0.03480876922607422,
714
+ 0.034697792053222654,
715
+ 0.034726303100585935,
716
+ 0.03470809555053711
717
+ ],
718
+ "mean": 0.03483440628051758,
719
+ "median": 0.034726303100585935,
720
+ "std": 0.00020214088054942904,
721
+ "min": 0.034697792053222654,
722
+ "max": 0.03523107147216797,
723
+ "p25": 0.03470809555053711,
724
+ "p75": 0.03480876922607422,
725
+ "p90": 0.035062150573730466,
726
+ "p95": 0.03514661102294922,
727
+ "p99": 0.03521417938232422,
728
+ "unit": "seconds"
729
+ },
730
+ "tokens_per_second": {
731
+ "name": "tokens_per_second",
732
+ "measurements": [
733
+ 29.351247148536373,
734
+ 29.367503838890563,
735
+ 29.389540638466244,
736
+ 29.356762935251997,
737
+ 29.355344872984382
738
+ ],
739
+ "mean": 29.36407988682591,
740
+ "median": 29.356762935251997,
741
+ "std": 0.013816589600811157,
742
+ "min": 29.351247148536373,
743
+ "max": 29.389540638466244,
744
+ "p25": 29.355344872984382,
745
+ "p75": 29.367503838890563,
746
+ "p90": 29.38072591863597,
747
+ "p95": 29.38513327855111,
748
+ "p99": 29.388659166483215,
749
+ "unit": "tokens/sec"
750
+ },
751
+ "time_per_output_token_seconds": {
752
+ "name": "time_per_output_token_seconds",
753
+ "measurements": [
754
+ 0.0340701025390625,
755
+ 0.034051242675781246,
756
+ 0.03402571044921875,
757
+ 0.034063701171875,
758
+ 0.0340653466796875
759
+ ],
760
+ "mean": 0.03405522070312501,
761
+ "median": 0.034063701171875,
762
+ "std": 1.6016257862519007e-05,
763
+ "min": 0.03402571044921875,
764
+ "max": 0.0340701025390625,
765
+ "p25": 0.034051242675781246,
766
+ "p75": 0.0340653466796875,
767
+ "p90": 0.0340682001953125,
768
+ "p95": 0.0340691513671875,
769
+ "p99": 0.0340699123046875,
770
+ "unit": "seconds/token"
771
+ }
772
+ },
773
+ "gpu_metrics": {
774
+ "gpu_utilization_mean": 96.93023255813954,
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": "2025-09-25T16:51:36.409631",
788
+ "commit_id": "e691f84412563b6abca098f3e044980725d8daa3",
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
+ 3.685489990234375,
822
+ 3.688985595703125,
823
+ 3.69080712890625,
824
+ 3.685994140625,
825
+ 3.690078369140625
826
+ ],
827
+ "mean": 3.688271044921875,
828
+ "median": 3.688985595703125,
829
+ "std": 0.0021506875169797143,
830
+ "min": 3.685489990234375,
831
+ "max": 3.69080712890625,
832
+ "p25": 3.685994140625,
833
+ "p75": 3.690078369140625,
834
+ "p90": 3.690515625,
835
+ "p95": 3.690661376953125,
836
+ "p99": 3.690777978515625,
837
+ "unit": "seconds"
838
+ },
839
+ "time_to_first_token_seconds": {
840
+ "name": "time_to_first_token_seconds",
841
+ "measurements": [
842
+ 0.03802460861206055,
843
+ 0.03751283264160156,
844
+ 0.037632129669189454,
845
+ 0.037286590576171875,
846
+ 0.03733695983886719
847
+ ],
848
+ "mean": 0.037558624267578125,
849
+ "median": 0.03751283264160156,
850
+ "std": 0.00026373425056074626,
851
+ "min": 0.037286590576171875,
852
+ "max": 0.03802460861206055,
853
+ "p25": 0.03733695983886719,
854
+ "p75": 0.037632129669189454,
855
+ "p90": 0.037867617034912114,
856
+ "p95": 0.03794611282348633,
857
+ "p99": 0.03800890945434571,
858
+ "unit": "seconds"
859
+ },
860
+ "tokens_per_second": {
861
+ "name": "tokens_per_second",
862
+ "measurements": [
863
+ 27.133434160715385,
864
+ 27.107723086931678,
865
+ 27.09434454507365,
866
+ 27.129722995990416,
867
+ 27.099695452616853
868
+ ],
869
+ "mean": 27.1129840482656,
870
+ "median": 27.107723086931678,
871
+ "std": 0.015811989361611947,
872
+ "min": 27.09434454507365,
873
+ "max": 27.133434160715385,
874
+ "p25": 27.099695452616853,
875
+ "p75": 27.129722995990416,
876
+ "p90": 27.131949694825398,
877
+ "p95": 27.13269192777039,
878
+ "p99": 27.133285714126387,
879
+ "unit": "tokens/sec"
880
+ },
881
+ "time_per_output_token_seconds": {
882
+ "name": "time_per_output_token_seconds",
883
+ "measurements": [
884
+ 0.03685489990234375,
885
+ 0.036889855957031245,
886
+ 0.036908071289062504,
887
+ 0.03685994140625,
888
+ 0.03690078369140625
889
+ ],
890
+ "mean": 0.036882710449218745,
891
+ "median": 0.036889855957031245,
892
+ "std": 2.1506875169797787e-05,
893
+ "min": 0.03685489990234375,
894
+ "max": 0.036908071289062504,
895
+ "p25": 0.03685994140625,
896
+ "p75": 0.03690078369140625,
897
+ "p90": 0.03690515625,
898
+ "p95": 0.036906613769531256,
899
+ "p99": 0.036907779785156254,
900
+ "unit": "seconds/token"
901
+ }
902
+ },
903
+ "gpu_metrics": {
904
+ "gpu_utilization_mean": 90.65591397849462,
905
+ "gpu_utilization_max": 94,
906
+ "gpu_utilization_min": 89,
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": "2025-09-25T16:52:11.058121",
918
+ "commit_id": "e691f84412563b6abca098f3e044980725d8daa3",
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
+ 3.917157958984375,
952
+ 3.92244580078125,
953
+ 3.89287158203125,
954
+ 3.90989697265625,
955
+ 3.894402587890625
956
+ ],
957
+ "mean": 3.9073549804687504,
958
+ "median": 3.90989697265625,
959
+ "std": 0.011898119797350568,
960
+ "min": 3.89287158203125,
961
+ "max": 3.92244580078125,
962
+ "p25": 3.894402587890625,
963
+ "p75": 3.917157958984375,
964
+ "p90": 3.9203306640625,
965
+ "p95": 3.921388232421875,
966
+ "p99": 3.922234287109375,
967
+ "unit": "seconds"
968
+ },
969
+ "time_to_first_token_seconds": {
970
+ "name": "time_to_first_token_seconds",
971
+ "measurements": [
972
+ 0.04168207931518555,
973
+ 0.039371070861816404,
974
+ 0.04108143997192383,
975
+ 0.04178953552246094,
976
+ 0.0389222412109375
977
+ ],
978
+ "mean": 0.04056927337646484,
979
+ "median": 0.04108143997192383,
980
+ "std": 0.0011948301640064708,
981
+ "min": 0.0389222412109375,
982
+ "max": 0.04178953552246094,
983
+ "p25": 0.039371070861816404,
984
+ "p75": 0.04168207931518555,
985
+ "p90": 0.041746553039550785,
986
+ "p95": 0.04176804428100586,
987
+ "p99": 0.041785237274169926,
988
+ "unit": "seconds"
989
+ },
990
+ "tokens_per_second": {
991
+ "name": "tokens_per_second",
992
+ "measurements": [
993
+ 25.528712665426337,
994
+ 25.494297455960407,
995
+ 25.687978114043336,
996
+ 25.576121493570565,
997
+ 25.677879403362937
998
+ ],
999
+ "mean": 25.59299782647272,
1000
+ "median": 25.576121493570565,
1001
+ "std": 0.07795549089656018,
1002
+ "min": 25.494297455960407,
1003
+ "max": 25.687978114043336,
1004
+ "p25": 25.528712665426337,
1005
+ "p75": 25.677879403362937,
1006
+ "p90": 25.683938629771177,
1007
+ "p95": 25.685958371907255,
1008
+ "p99": 25.68757416561612,
1009
+ "unit": "tokens/sec"
1010
+ },
1011
+ "time_per_output_token_seconds": {
1012
+ "name": "time_per_output_token_seconds",
1013
+ "measurements": [
1014
+ 0.03917157958984375,
1015
+ 0.0392244580078125,
1016
+ 0.0389287158203125,
1017
+ 0.0390989697265625,
1018
+ 0.03894402587890625
1019
+ ],
1020
+ "mean": 0.039073549804687496,
1021
+ "median": 0.0390989697265625,
1022
+ "std": 0.000118981197973505,
1023
+ "min": 0.0389287158203125,
1024
+ "max": 0.0392244580078125,
1025
+ "p25": 0.03894402587890625,
1026
+ "p75": 0.03917157958984375,
1027
+ "p90": 0.039203306640625,
1028
+ "p95": 0.03921388232421875,
1029
+ "p99": 0.039222342871093746,
1030
+ "unit": "seconds/token"
1031
+ }
1032
+ },
1033
+ "gpu_metrics": {
1034
+ "gpu_utilization_mean": 88.5204081632653,
1035
+ "gpu_utilization_max": 93,
1036
+ "gpu_utilization_min": 84,
1037
+ "gpu_memory_used_mean": 13845,
1038
+ "gpu_memory_used_max": 13845,
1039
+ "gpu_memory_used_min": 13845,
1040
+ "sample_count": 98,
1041
+ "gpu_monitoring_status": "success"
1042
+ }
1043
+ },
1044
+ {
1045
+ "scenario_name": "compiled_compile_max-autotune_sdpa_efficient_attention",
1046
+ "metadata": {
1047
+ "timestamp": "2025-09-25T16:52:52.019088",
1048
+ "commit_id": "e691f84412563b6abca098f3e044980725d8daa3",
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
+ 3.692386962890625,
1082
+ 3.69098828125,
1083
+ 3.691014404296875,
1084
+ 3.69522314453125,
1085
+ 3.700642822265625
1086
+ ],
1087
+ "mean": 3.694051123046875,
1088
+ "median": 3.692386962890625,
1089
+ "std": 0.0036385772663293145,
1090
+ "min": 3.69098828125,
1091
+ "max": 3.700642822265625,
1092
+ "p25": 3.691014404296875,
1093
+ "p75": 3.69522314453125,
1094
+ "p90": 3.698474951171875,
1095
+ "p95": 3.69955888671875,
1096
+ "p99": 3.70042603515625,
1097
+ "unit": "seconds"
1098
+ },
1099
+ "time_to_first_token_seconds": {
1100
+ "name": "time_to_first_token_seconds",
1101
+ "measurements": [
1102
+ 0.0385986557006836,
1103
+ 0.037429759979248044,
1104
+ 0.037345375061035156,
1105
+ 0.03738188934326172,
1106
+ 0.03735270309448242
1107
+ ],
1108
+ "mean": 0.03762167663574219,
1109
+ "median": 0.03738188934326172,
1110
+ "std": 0.0004893888777020171,
1111
+ "min": 0.037345375061035156,
1112
+ "max": 0.0385986557006836,
1113
+ "p25": 0.03735270309448242,
1114
+ "p75": 0.037429759979248044,
1115
+ "p90": 0.03813109741210938,
1116
+ "p95": 0.038364876556396484,
1117
+ "p99": 0.038551899871826174,
1118
+ "unit": "seconds"
1119
+ },
1120
+ "tokens_per_second": {
1121
+ "name": "tokens_per_second",
1122
+ "measurements": [
1123
+ 27.082751890585683,
1124
+ 27.093014764634724,
1125
+ 27.092823014612332,
1126
+ 27.061965161155456,
1127
+ 27.022332281929746
1128
+ ],
1129
+ "mean": 27.070577422583586,
1130
+ "median": 27.082751890585683,
1131
+ "std": 0.026639380615091378,
1132
+ "min": 27.022332281929746,
1133
+ "max": 27.093014764634724,
1134
+ "p25": 27.061965161155456,
1135
+ "p75": 27.092823014612332,
1136
+ "p90": 27.092938064625766,
1137
+ "p95": 27.092976414630247,
1138
+ "p99": 27.09300709463383,
1139
+ "unit": "tokens/sec"
1140
+ },
1141
+ "time_per_output_token_seconds": {
1142
+ "name": "time_per_output_token_seconds",
1143
+ "measurements": [
1144
+ 0.03692386962890625,
1145
+ 0.0369098828125,
1146
+ 0.03691014404296875,
1147
+ 0.0369522314453125,
1148
+ 0.03700642822265625
1149
+ ],
1150
+ "mean": 0.03694051123046875,
1151
+ "median": 0.03692386962890625,
1152
+ "std": 3.638577266329127e-05,
1153
+ "min": 0.0369098828125,
1154
+ "max": 0.03700642822265625,
1155
+ "p25": 0.03691014404296875,
1156
+ "p75": 0.0369522314453125,
1157
+ "p90": 0.03698474951171875,
1158
+ "p95": 0.0369955888671875,
1159
+ "p99": 0.0370042603515625,
1160
+ "unit": "seconds/token"
1161
+ }
1162
+ },
1163
+ "gpu_metrics": {
1164
+ "gpu_utilization_mean": 90.51612903225806,
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
+ }