hf-transformers-bot commited on
Commit
8d2f01b
·
verified ·
1 Parent(s): c3c068f

Upload benchmark results for run 18431995772

Browse files
2025-10-11/18431995772/benchmark_results/Llama-2-7b-hf/Llama-2-7b-hf_benchmark_20251011_164944.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": "2025-10-11T16:43:43.980016",
8
+ "commit_id": "3927ffed31e3c0d2929bf98bd05b7c61fcc48b62",
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
+ 3.65125341796875,
42
+ 3.65687841796875,
43
+ 3.66764013671875,
44
+ 3.659583984375,
45
+ 3.655740966796875
46
+ ],
47
+ "mean": 3.6582193847656255,
48
+ "median": 3.65687841796875,
49
+ "std": 0.0054236176201780005,
50
+ "min": 3.65125341796875,
51
+ "max": 3.66764013671875,
52
+ "p25": 3.655740966796875,
53
+ "p75": 3.659583984375,
54
+ "p90": 3.6644176757812503,
55
+ "p95": 3.66602890625,
56
+ "p99": 3.667317890625,
57
+ "unit": "seconds"
58
+ },
59
+ "time_to_first_token_seconds": {
60
+ "name": "time_to_first_token_seconds",
61
+ "measurements": [
62
+ 0.03764384078979492,
63
+ 0.03725932693481445,
64
+ 0.037032161712646484,
65
+ 0.03724832153320313,
66
+ 0.037058433532714846
67
+ ],
68
+ "mean": 0.03724841690063476,
69
+ "median": 0.03724832153320313,
70
+ "std": 0.00021878714609287935,
71
+ "min": 0.037032161712646484,
72
+ "max": 0.03764384078979492,
73
+ "p25": 0.037058433532714846,
74
+ "p75": 0.03725932693481445,
75
+ "p90": 0.037490035247802735,
76
+ "p95": 0.037566938018798825,
77
+ "p99": 0.037628460235595705,
78
+ "unit": "seconds"
79
+ },
80
+ "tokens_per_second": {
81
+ "name": "tokens_per_second",
82
+ "measurements": [
83
+ 27.387855224694754,
84
+ 27.34572730354705,
85
+ 27.265488508222315,
86
+ 27.32551033859616,
87
+ 27.354235682518564
88
+ ],
89
+ "mean": 27.335763411515767,
90
+ "median": 27.34572730354705,
91
+ "std": 0.040491353343362166,
92
+ "min": 27.265488508222315,
93
+ "max": 27.387855224694754,
94
+ "p25": 27.32551033859616,
95
+ "p75": 27.354235682518564,
96
+ "p90": 27.37440740782428,
97
+ "p95": 27.381131316259516,
98
+ "p99": 27.386510443007705,
99
+ "unit": "tokens/sec"
100
+ },
101
+ "time_per_output_token_seconds": {
102
+ "name": "time_per_output_token_seconds",
103
+ "measurements": [
104
+ 0.0365125341796875,
105
+ 0.0365687841796875,
106
+ 0.0366764013671875,
107
+ 0.036595839843750005,
108
+ 0.03655740966796875
109
+ ],
110
+ "mean": 0.03658219384765625,
111
+ "median": 0.0365687841796875,
112
+ "std": 5.423617620177981e-05,
113
+ "min": 0.0365125341796875,
114
+ "max": 0.0366764013671875,
115
+ "p25": 0.03655740966796875,
116
+ "p75": 0.036595839843750005,
117
+ "p90": 0.0366441767578125,
118
+ "p95": 0.036660289062500004,
119
+ "p99": 0.03667317890625,
120
+ "unit": "seconds/token"
121
+ }
122
+ },
123
+ "gpu_metrics": {
124
+ "gpu_utilization_mean": 90.79347826086956,
125
+ "gpu_utilization_max": 93,
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": "2025-10-11T16:44:19.256732",
138
+ "commit_id": "3927ffed31e3c0d2929bf98bd05b7c61fcc48b62",
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
+ 7.64912109375,
172
+ 7.6802998046875,
173
+ 7.71979638671875,
174
+ 7.732490234375,
175
+ 7.75858203125
176
+ ],
177
+ "mean": 7.708057910156251,
178
+ "median": 7.71979638671875,
179
+ "std": 0.03881102384208024,
180
+ "min": 7.64912109375,
181
+ "max": 7.75858203125,
182
+ "p25": 7.6802998046875,
183
+ "p75": 7.732490234375,
184
+ "p90": 7.748145312499999,
185
+ "p95": 7.753363671874999,
186
+ "p99": 7.757538359374999,
187
+ "unit": "seconds"
188
+ },
189
+ "time_to_first_token_seconds": {
190
+ "name": "time_to_first_token_seconds",
191
+ "measurements": [
192
+ 0.038556224822998045,
193
+ 0.03839449691772461,
194
+ 0.03804832077026367,
195
+ 0.038465408325195315,
196
+ 0.039977119445800784
197
+ ],
198
+ "mean": 0.038688314056396486,
199
+ "median": 0.038465408325195315,
200
+ "std": 0.0006669443424472673,
201
+ "min": 0.03804832077026367,
202
+ "max": 0.039977119445800784,
203
+ "p25": 0.03839449691772461,
204
+ "p75": 0.038556224822998045,
205
+ "p90": 0.039408761596679685,
206
+ "p95": 0.039692940521240234,
207
+ "p99": 0.03992028366088867,
208
+ "unit": "seconds"
209
+ },
210
+ "tokens_per_second": {
211
+ "name": "tokens_per_second",
212
+ "measurements": [
213
+ 13.073397423621484,
214
+ 13.020325058009744,
215
+ 12.953709526852476,
216
+ 12.9324443961723,
217
+ 12.888953109887892
218
+ ],
219
+ "mean": 12.973765902908777,
220
+ "median": 12.953709526852476,
221
+ "std": 0.06541732162467455,
222
+ "min": 12.888953109887892,
223
+ "max": 13.073397423621484,
224
+ "p25": 12.9324443961723,
225
+ "p75": 13.020325058009744,
226
+ "p90": 13.052168477376789,
227
+ "p95": 13.062782950499136,
228
+ "p99": 13.071274528997014,
229
+ "unit": "tokens/sec"
230
+ },
231
+ "time_per_output_token_seconds": {
232
+ "name": "time_per_output_token_seconds",
233
+ "measurements": [
234
+ 0.0764912109375,
235
+ 0.076802998046875,
236
+ 0.0771979638671875,
237
+ 0.07732490234375,
238
+ 0.0775858203125
239
+ ],
240
+ "mean": 0.0770805791015625,
241
+ "median": 0.0771979638671875,
242
+ "std": 0.0003881102384208037,
243
+ "min": 0.0764912109375,
244
+ "max": 0.0775858203125,
245
+ "p25": 0.076802998046875,
246
+ "p75": 0.07732490234375,
247
+ "p90": 0.077481453125,
248
+ "p95": 0.07753363671875,
249
+ "p99": 0.07757538359374999,
250
+ "unit": "seconds/token"
251
+ }
252
+ },
253
+ "gpu_metrics": {
254
+ "gpu_utilization_mean": 44.90419161676647,
255
+ "gpu_utilization_max": 94,
256
+ "gpu_utilization_min": 0,
257
+ "gpu_memory_used_mean": 13629.275449101797,
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": "2025-10-11T16:45:45.506000",
268
+ "commit_id": "3927ffed31e3c0d2929bf98bd05b7c61fcc48b62",
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
+ 3.41041455078125,
302
+ 3.410800048828125,
303
+ 3.4115810546875,
304
+ 3.40958642578125,
305
+ 3.412004638671875
306
+ ],
307
+ "mean": 3.4108773437499997,
308
+ "median": 3.410800048828125,
309
+ "std": 0.0008547074310377769,
310
+ "min": 3.40958642578125,
311
+ "max": 3.412004638671875,
312
+ "p25": 3.41041455078125,
313
+ "p75": 3.4115810546875,
314
+ "p90": 3.411835205078125,
315
+ "p95": 3.411919921875,
316
+ "p99": 3.4119876953125,
317
+ "unit": "seconds"
318
+ },
319
+ "time_to_first_token_seconds": {
320
+ "name": "time_to_first_token_seconds",
321
+ "measurements": [
322
+ 0.035044158935546875,
323
+ 0.034742847442626956,
324
+ 0.034551681518554686,
325
+ 0.03454403305053711,
326
+ 0.034550113677978514
327
+ ],
328
+ "mean": 0.03468656692504883,
329
+ "median": 0.034551681518554686,
330
+ "std": 0.00019399433472931175,
331
+ "min": 0.03454403305053711,
332
+ "max": 0.035044158935546875,
333
+ "p25": 0.034550113677978514,
334
+ "p75": 0.034742847442626956,
335
+ "p90": 0.03492363433837891,
336
+ "p95": 0.03498389663696289,
337
+ "p99": 0.03503210647583008,
338
+ "unit": "seconds"
339
+ },
340
+ "tokens_per_second": {
341
+ "name": "tokens_per_second",
342
+ "measurements": [
343
+ 29.321948552293218,
344
+ 29.31863450463998,
345
+ 29.3119226531641,
346
+ 29.32907030713752,
347
+ 29.308283718783297
348
+ ],
349
+ "mean": 29.31797194720362,
350
+ "median": 29.31863450463998,
351
+ "std": 0.007346859082879283,
352
+ "min": 29.308283718783297,
353
+ "max": 29.32907030713752,
354
+ "p25": 29.3119226531641,
355
+ "p75": 29.321948552293218,
356
+ "p90": 29.326221605199798,
357
+ "p95": 29.32764595616866,
358
+ "p99": 29.328785436943747,
359
+ "unit": "tokens/sec"
360
+ },
361
+ "time_per_output_token_seconds": {
362
+ "name": "time_per_output_token_seconds",
363
+ "measurements": [
364
+ 0.0341041455078125,
365
+ 0.03410800048828125,
366
+ 0.034115810546875,
367
+ 0.0340958642578125,
368
+ 0.03412004638671875
369
+ ],
370
+ "mean": 0.0341087734375,
371
+ "median": 0.03410800048828125,
372
+ "std": 8.547074310378012e-06,
373
+ "min": 0.0340958642578125,
374
+ "max": 0.03412004638671875,
375
+ "p25": 0.0341041455078125,
376
+ "p75": 0.034115810546875,
377
+ "p90": 0.03411835205078125,
378
+ "p95": 0.03411919921875,
379
+ "p99": 0.034119876953124996,
380
+ "unit": "seconds/token"
381
+ }
382
+ },
383
+ "gpu_metrics": {
384
+ "gpu_utilization_mean": 96.93023255813954,
385
+ "gpu_utilization_max": 98,
386
+ "gpu_utilization_min": 95,
387
+ "gpu_memory_used_mean": 13849,
388
+ "gpu_memory_used_max": 13849,
389
+ "gpu_memory_used_min": 13849,
390
+ "sample_count": 86,
391
+ "gpu_monitoring_status": "success"
392
+ }
393
+ },
394
+ {
395
+ "scenario_name": "eager_sdpa_math",
396
+ "metadata": {
397
+ "timestamp": "2025-10-11T16:46:17.971278",
398
+ "commit_id": "3927ffed31e3c0d2929bf98bd05b7c61fcc48b62",
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
+ 3.5505830078125,
432
+ 3.549416259765625,
433
+ 3.547959716796875,
434
+ 3.56267626953125,
435
+ 3.55363916015625
436
+ ],
437
+ "mean": 3.5528548828125004,
438
+ "median": 3.5505830078125,
439
+ "std": 0.005254006525039997,
440
+ "min": 3.547959716796875,
441
+ "max": 3.56267626953125,
442
+ "p25": 3.549416259765625,
443
+ "p75": 3.55363916015625,
444
+ "p90": 3.55906142578125,
445
+ "p95": 3.56086884765625,
446
+ "p99": 3.56231478515625,
447
+ "unit": "seconds"
448
+ },
449
+ "time_to_first_token_seconds": {
450
+ "name": "time_to_first_token_seconds",
451
+ "measurements": [
452
+ 0.04025312042236328,
453
+ 0.040066654205322266,
454
+ 0.039704158782958986,
455
+ 0.03982745742797852,
456
+ 0.040304702758789064
457
+ ],
458
+ "mean": 0.040031218719482425,
459
+ "median": 0.040066654205322266,
460
+ "std": 0.00023399847228033804,
461
+ "min": 0.039704158782958986,
462
+ "max": 0.040304702758789064,
463
+ "p25": 0.03982745742797852,
464
+ "p75": 0.04025312042236328,
465
+ "p90": 0.04028406982421875,
466
+ "p95": 0.040294386291503906,
467
+ "p99": 0.04030263946533203,
468
+ "unit": "seconds"
469
+ },
470
+ "tokens_per_second": {
471
+ "name": "tokens_per_second",
472
+ "measurements": [
473
+ 28.16438871587165,
474
+ 28.17364678624738,
475
+ 28.185212906047525,
476
+ 28.068786618425268,
477
+ 28.140167161936358
478
+ ],
479
+ "mean": 28.146440437705632,
480
+ "median": 28.16438871587165,
481
+ "std": 0.04155743021200376,
482
+ "min": 28.068786618425268,
483
+ "max": 28.185212906047525,
484
+ "p25": 28.140167161936358,
485
+ "p75": 28.17364678624738,
486
+ "p90": 28.18058645812747,
487
+ "p95": 28.182899682087495,
488
+ "p99": 28.18475026125552,
489
+ "unit": "tokens/sec"
490
+ },
491
+ "time_per_output_token_seconds": {
492
+ "name": "time_per_output_token_seconds",
493
+ "measurements": [
494
+ 0.035505830078125,
495
+ 0.03549416259765625,
496
+ 0.03547959716796875,
497
+ 0.0356267626953125,
498
+ 0.0355363916015625
499
+ ],
500
+ "mean": 0.035528548828125,
501
+ "median": 0.035505830078125,
502
+ "std": 5.254006525040032e-05,
503
+ "min": 0.03547959716796875,
504
+ "max": 0.0356267626953125,
505
+ "p25": 0.03549416259765625,
506
+ "p75": 0.0355363916015625,
507
+ "p90": 0.0355906142578125,
508
+ "p95": 0.035608688476562504,
509
+ "p99": 0.0356231478515625,
510
+ "unit": "seconds/token"
511
+ }
512
+ },
513
+ "gpu_metrics": {
514
+ "gpu_utilization_mean": 95.71111111111111,
515
+ "gpu_utilization_max": 98,
516
+ "gpu_utilization_min": 86,
517
+ "gpu_memory_used_mean": 13849,
518
+ "gpu_memory_used_max": 13849,
519
+ "gpu_memory_used_min": 13849,
520
+ "sample_count": 90,
521
+ "gpu_monitoring_status": "success"
522
+ }
523
+ },
524
+ {
525
+ "scenario_name": "eager_sdpa_flash_attention",
526
+ "metadata": {
527
+ "timestamp": "2025-10-11T16:46:51.639222",
528
+ "commit_id": "3927ffed31e3c0d2929bf98bd05b7c61fcc48b62",
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
+ 3.410203369140625,
562
+ 3.409483154296875,
563
+ 3.407915283203125,
564
+ 3.41034912109375,
565
+ 3.410473388671875
566
+ ],
567
+ "mean": 3.4096848632812504,
568
+ "median": 3.410203369140625,
569
+ "std": 0.0009491005454227452,
570
+ "min": 3.407915283203125,
571
+ "max": 3.410473388671875,
572
+ "p25": 3.409483154296875,
573
+ "p75": 3.41034912109375,
574
+ "p90": 3.410423681640625,
575
+ "p95": 3.41044853515625,
576
+ "p99": 3.41046841796875,
577
+ "unit": "seconds"
578
+ },
579
+ "time_to_first_token_seconds": {
580
+ "name": "time_to_first_token_seconds",
581
+ "measurements": [
582
+ 0.03523936080932617,
583
+ 0.03462428665161133,
584
+ 0.034614334106445314,
585
+ 0.03462566375732422,
586
+ 0.03462652969360352
587
+ ],
588
+ "mean": 0.034746035003662114,
589
+ "median": 0.03462566375732422,
590
+ "std": 0.00024670180143891273,
591
+ "min": 0.034614334106445314,
592
+ "max": 0.03523936080932617,
593
+ "p25": 0.03462428665161133,
594
+ "p75": 0.03462652969360352,
595
+ "p90": 0.03499422836303711,
596
+ "p95": 0.035116794586181645,
597
+ "p99": 0.03521484756469727,
598
+ "unit": "seconds"
599
+ },
600
+ "tokens_per_second": {
601
+ "name": "tokens_per_second",
602
+ "measurements": [
603
+ 29.323764355203284,
604
+ 29.32995866953407,
605
+ 29.343452430545533,
606
+ 29.32251111227243,
607
+ 29.32144268656573
608
+ ],
609
+ "mean": 29.328225850824207,
610
+ "median": 29.323764355203284,
611
+ "std": 0.008166099288390617,
612
+ "min": 29.32144268656573,
613
+ "max": 29.343452430545533,
614
+ "p25": 29.32251111227243,
615
+ "p75": 29.32995866953407,
616
+ "p90": 29.33805492614095,
617
+ "p95": 29.34075367834324,
618
+ "p99": 29.342912680105073,
619
+ "unit": "tokens/sec"
620
+ },
621
+ "time_per_output_token_seconds": {
622
+ "name": "time_per_output_token_seconds",
623
+ "measurements": [
624
+ 0.03410203369140625,
625
+ 0.03409483154296875,
626
+ 0.03407915283203125,
627
+ 0.0341034912109375,
628
+ 0.03410473388671875
629
+ ],
630
+ "mean": 0.0340968486328125,
631
+ "median": 0.03410203369140625,
632
+ "std": 9.491005454227164e-06,
633
+ "min": 0.03407915283203125,
634
+ "max": 0.03410473388671875,
635
+ "p25": 0.03409483154296875,
636
+ "p75": 0.0341034912109375,
637
+ "p90": 0.03410423681640625,
638
+ "p95": 0.034104485351562505,
639
+ "p99": 0.034104684179687504,
640
+ "unit": "seconds/token"
641
+ }
642
+ },
643
+ "gpu_metrics": {
644
+ "gpu_utilization_mean": 97.02325581395348,
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": "2025-10-11T16:47:24.087418",
658
+ "commit_id": "3927ffed31e3c0d2929bf98bd05b7c61fcc48b62",
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
+ 3.406935791015625,
692
+ 3.4061513671875,
693
+ 3.405931396484375,
694
+ 3.4077802734375,
695
+ 3.40796533203125
696
+ ],
697
+ "mean": 3.40695283203125,
698
+ "median": 3.406935791015625,
699
+ "std": 0.0008241095071867344,
700
+ "min": 3.405931396484375,
701
+ "max": 3.40796533203125,
702
+ "p25": 3.4061513671875,
703
+ "p75": 3.4077802734375,
704
+ "p90": 3.40789130859375,
705
+ "p95": 3.4079283203125,
706
+ "p99": 3.4079579296875,
707
+ "unit": "seconds"
708
+ },
709
+ "time_to_first_token_seconds": {
710
+ "name": "time_to_first_token_seconds",
711
+ "measurements": [
712
+ 0.03532467269897461,
713
+ 0.034947486877441404,
714
+ 0.03472803115844727,
715
+ 0.034694881439208985,
716
+ 0.03468537521362305
717
+ ],
718
+ "mean": 0.03487608947753906,
719
+ "median": 0.03472803115844727,
720
+ "std": 0.00024390737780356577,
721
+ "min": 0.03468537521362305,
722
+ "max": 0.03532467269897461,
723
+ "p25": 0.034694881439208985,
724
+ "p75": 0.034947486877441404,
725
+ "p90": 0.035173798370361324,
726
+ "p95": 0.035249235534667966,
727
+ "p99": 0.03530958526611328,
728
+ "unit": "seconds"
729
+ },
730
+ "tokens_per_second": {
731
+ "name": "tokens_per_second",
732
+ "measurements": [
733
+ 29.35188865716471,
734
+ 29.35864828654729,
735
+ 29.360544402984942,
736
+ 29.344614962257495,
737
+ 29.34302149734516
738
+ ],
739
+ "mean": 29.351743561259916,
740
+ "median": 29.35188865716471,
741
+ "std": 0.007099893264762082,
742
+ "min": 29.34302149734516,
743
+ "max": 29.360544402984942,
744
+ "p25": 29.344614962257495,
745
+ "p75": 29.35864828654729,
746
+ "p90": 29.35978595640988,
747
+ "p95": 29.36016517969741,
748
+ "p99": 29.360468558327437,
749
+ "unit": "tokens/sec"
750
+ },
751
+ "time_per_output_token_seconds": {
752
+ "name": "time_per_output_token_seconds",
753
+ "measurements": [
754
+ 0.03406935791015625,
755
+ 0.034061513671875,
756
+ 0.03405931396484375,
757
+ 0.034077802734375,
758
+ 0.0340796533203125
759
+ ],
760
+ "mean": 0.0340695283203125,
761
+ "median": 0.03406935791015625,
762
+ "std": 8.241095071867914e-06,
763
+ "min": 0.03405931396484375,
764
+ "max": 0.0340796533203125,
765
+ "p25": 0.034061513671875,
766
+ "p75": 0.034077802734375,
767
+ "p90": 0.0340789130859375,
768
+ "p95": 0.034079283203125,
769
+ "p99": 0.034079579296875,
770
+ "unit": "seconds/token"
771
+ }
772
+ },
773
+ "gpu_metrics": {
774
+ "gpu_utilization_mean": 96.93023255813954,
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": "2025-10-11T16:47:56.501400",
788
+ "commit_id": "3927ffed31e3c0d2929bf98bd05b7c61fcc48b62",
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
+ 3.690744384765625,
822
+ 3.692240234375,
823
+ 3.69592919921875,
824
+ 3.6918798828125,
825
+ 3.6963828125
826
+ ],
827
+ "mean": 3.6934353027343754,
828
+ "median": 3.692240234375,
829
+ "std": 0.002280170718376948,
830
+ "min": 3.690744384765625,
831
+ "max": 3.6963828125,
832
+ "p25": 3.6918798828125,
833
+ "p75": 3.69592919921875,
834
+ "p90": 3.6962013671875,
835
+ "p95": 3.69629208984375,
836
+ "p99": 3.69636466796875,
837
+ "unit": "seconds"
838
+ },
839
+ "time_to_first_token_seconds": {
840
+ "name": "time_to_first_token_seconds",
841
+ "measurements": [
842
+ 0.03809132766723633,
843
+ 0.037510528564453124,
844
+ 0.03746271896362305,
845
+ 0.03749347305297852,
846
+ 0.037463134765625
847
+ ],
848
+ "mean": 0.037604236602783205,
849
+ "median": 0.03749347305297852,
850
+ "std": 0.000244231216729545,
851
+ "min": 0.03746271896362305,
852
+ "max": 0.03809132766723633,
853
+ "p25": 0.037463134765625,
854
+ "p75": 0.037510528564453124,
855
+ "p90": 0.03785900802612305,
856
+ "p95": 0.03797516784667969,
857
+ "p99": 0.038068095703125,
858
+ "unit": "seconds"
859
+ },
860
+ "tokens_per_second": {
861
+ "name": "tokens_per_second",
862
+ "measurements": [
863
+ 27.094805159840497,
864
+ 27.08382815099446,
865
+ 27.056795357751476,
866
+ 27.08647170931772,
867
+ 27.053474997727932
868
+ ],
869
+ "mean": 27.075075075126414,
870
+ "median": 27.08382815099446,
871
+ "std": 0.016712194719995167,
872
+ "min": 27.053474997727932,
873
+ "max": 27.094805159840497,
874
+ "p25": 27.056795357751476,
875
+ "p75": 27.08647170931772,
876
+ "p90": 27.091471779631387,
877
+ "p95": 27.09313846973594,
878
+ "p99": 27.094471821819585,
879
+ "unit": "tokens/sec"
880
+ },
881
+ "time_per_output_token_seconds": {
882
+ "name": "time_per_output_token_seconds",
883
+ "measurements": [
884
+ 0.03690744384765625,
885
+ 0.03692240234375,
886
+ 0.0369592919921875,
887
+ 0.036918798828125,
888
+ 0.036963828125
889
+ ],
890
+ "mean": 0.03693435302734375,
891
+ "median": 0.03692240234375,
892
+ "std": 2.2801707183769393e-05,
893
+ "min": 0.03690744384765625,
894
+ "max": 0.036963828125,
895
+ "p25": 0.036918798828125,
896
+ "p75": 0.0369592919921875,
897
+ "p90": 0.036962013671875,
898
+ "p95": 0.0369629208984375,
899
+ "p99": 0.036963646679687504,
900
+ "unit": "seconds/token"
901
+ }
902
+ },
903
+ "gpu_metrics": {
904
+ "gpu_utilization_mean": 90.56989247311827,
905
+ "gpu_utilization_max": 93,
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": "2025-10-11T16:48:31.259857",
918
+ "commit_id": "3927ffed31e3c0d2929bf98bd05b7c61fcc48b62",
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
+ 3.854359375,
952
+ 3.96859423828125,
953
+ 3.94088623046875,
954
+ 3.93324462890625,
955
+ 3.910897705078125
956
+ ],
957
+ "mean": 3.921596435546875,
958
+ "median": 3.93324462890625,
959
+ "std": 0.03834541746585029,
960
+ "min": 3.854359375,
961
+ "max": 3.96859423828125,
962
+ "p25": 3.910897705078125,
963
+ "p75": 3.94088623046875,
964
+ "p90": 3.95751103515625,
965
+ "p95": 3.96305263671875,
966
+ "p99": 3.96748591796875,
967
+ "unit": "seconds"
968
+ },
969
+ "time_to_first_token_seconds": {
970
+ "name": "time_to_first_token_seconds",
971
+ "measurements": [
972
+ 0.040360511779785155,
973
+ 0.040336097717285156,
974
+ 0.04253401565551758,
975
+ 0.03981612777709961,
976
+ 0.041331424713134765
977
+ ],
978
+ "mean": 0.04087563552856445,
979
+ "median": 0.040360511779785155,
980
+ "std": 0.0009630138761217227,
981
+ "min": 0.03981612777709961,
982
+ "max": 0.04253401565551758,
983
+ "p25": 0.040336097717285156,
984
+ "p75": 0.041331424713134765,
985
+ "p90": 0.042052979278564454,
986
+ "p95": 0.042293497467041016,
987
+ "p99": 0.042485912017822265,
988
+ "unit": "seconds"
989
+ },
990
+ "tokens_per_second": {
991
+ "name": "tokens_per_second",
992
+ "measurements": [
993
+ 25.94464871351027,
994
+ 25.197839334491597,
995
+ 25.375003020096184,
996
+ 25.424302181735346,
997
+ 25.56957699766846
998
+ ],
999
+ "mean": 25.50227404950037,
1000
+ "median": 25.424302181735346,
1001
+ "std": 0.25107092742124787,
1002
+ "min": 25.197839334491597,
1003
+ "max": 25.94464871351027,
1004
+ "p25": 25.375003020096184,
1005
+ "p75": 25.56957699766846,
1006
+ "p90": 25.794620027173544,
1007
+ "p95": 25.869634370341906,
1008
+ "p99": 25.929645844876596,
1009
+ "unit": "tokens/sec"
1010
+ },
1011
+ "time_per_output_token_seconds": {
1012
+ "name": "time_per_output_token_seconds",
1013
+ "measurements": [
1014
+ 0.03854359375,
1015
+ 0.0396859423828125,
1016
+ 0.0394088623046875,
1017
+ 0.039332446289062496,
1018
+ 0.03910897705078125
1019
+ ],
1020
+ "mean": 0.03921596435546875,
1021
+ "median": 0.039332446289062496,
1022
+ "std": 0.0003834541746585027,
1023
+ "min": 0.03854359375,
1024
+ "max": 0.0396859423828125,
1025
+ "p25": 0.03910897705078125,
1026
+ "p75": 0.0394088623046875,
1027
+ "p90": 0.0395751103515625,
1028
+ "p95": 0.0396305263671875,
1029
+ "p99": 0.0396748591796875,
1030
+ "unit": "seconds/token"
1031
+ }
1032
+ },
1033
+ "gpu_metrics": {
1034
+ "gpu_utilization_mean": 88.61616161616162,
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": 99,
1041
+ "gpu_monitoring_status": "success"
1042
+ }
1043
+ },
1044
+ {
1045
+ "scenario_name": "compiled_compile_max-autotune_sdpa_efficient_attention",
1046
+ "metadata": {
1047
+ "timestamp": "2025-10-11T16:49:12.683573",
1048
+ "commit_id": "3927ffed31e3c0d2929bf98bd05b7c61fcc48b62",
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
+ 3.6903876953125,
1082
+ 3.692961181640625,
1083
+ 3.692920654296875,
1084
+ 3.700833740234375,
1085
+ 3.695314453125
1086
+ ],
1087
+ "mean": 3.694483544921875,
1088
+ "median": 3.692961181640625,
1089
+ "std": 0.0035369927915489454,
1090
+ "min": 3.6903876953125,
1091
+ "max": 3.700833740234375,
1092
+ "p25": 3.692920654296875,
1093
+ "p75": 3.695314453125,
1094
+ "p90": 3.698626025390625,
1095
+ "p95": 3.6997298828125,
1096
+ "p99": 3.70061296875,
1097
+ "unit": "seconds"
1098
+ },
1099
+ "time_to_first_token_seconds": {
1100
+ "name": "time_to_first_token_seconds",
1101
+ "measurements": [
1102
+ 0.038072128295898434,
1103
+ 0.0374334716796875,
1104
+ 0.037422176361083984,
1105
+ 0.037451168060302735,
1106
+ 0.037408416748046874
1107
+ ],
1108
+ "mean": 0.0375574722290039,
1109
+ "median": 0.0374334716796875,
1110
+ "std": 0.00025770917251231114,
1111
+ "min": 0.037408416748046874,
1112
+ "max": 0.038072128295898434,
1113
+ "p25": 0.037422176361083984,
1114
+ "p75": 0.037451168060302735,
1115
+ "p90": 0.03782374420166015,
1116
+ "p95": 0.0379479362487793,
1117
+ "p99": 0.03804728988647461,
1118
+ "unit": "seconds"
1119
+ },
1120
+ "tokens_per_second": {
1121
+ "name": "tokens_per_second",
1122
+ "measurements": [
1123
+ 27.097423971746704,
1124
+ 27.07854079191113,
1125
+ 27.078837960855083,
1126
+ 27.02093825854143,
1127
+ 27.06129647923019
1128
+ ],
1129
+ "mean": 27.06740749245691,
1130
+ "median": 27.07854079191113,
1131
+ "std": 0.025893355042721774,
1132
+ "min": 27.02093825854143,
1133
+ "max": 27.097423971746704,
1134
+ "p25": 27.06129647923019,
1135
+ "p75": 27.078837960855083,
1136
+ "p90": 27.089989567390056,
1137
+ "p95": 27.093706769568378,
1138
+ "p99": 27.096680531311037,
1139
+ "unit": "tokens/sec"
1140
+ },
1141
+ "time_per_output_token_seconds": {
1142
+ "name": "time_per_output_token_seconds",
1143
+ "measurements": [
1144
+ 0.036903876953125,
1145
+ 0.03692961181640625,
1146
+ 0.036929206542968745,
1147
+ 0.03700833740234375,
1148
+ 0.03695314453125
1149
+ ],
1150
+ "mean": 0.03694483544921875,
1151
+ "median": 0.03692961181640625,
1152
+ "std": 3.5369927915490843e-05,
1153
+ "min": 0.036903876953125,
1154
+ "max": 0.03700833740234375,
1155
+ "p25": 0.036929206542968745,
1156
+ "p75": 0.03695314453125,
1157
+ "p90": 0.03698626025390625,
1158
+ "p95": 0.036997298828125004,
1159
+ "p99": 0.0370061296875,
1160
+ "unit": "seconds/token"
1161
+ }
1162
+ },
1163
+ "gpu_metrics": {
1164
+ "gpu_utilization_mean": 90.49462365591398,
1165
+ "gpu_utilization_max": 93,
1166
+ "gpu_utilization_min": 87,
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
+ }