hf-transformers-bot commited on
Commit
2909bdb
·
verified ·
1 Parent(s): 26bbbb0

Upload benchmark results for run 18228184193

Browse files
2025-10-03/18228184193/benchmark_results/Llama-2-7b-hf/Llama-2-7b-hf_benchmark_20251003_165133.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": "2025-10-03T16:45:32.223789",
8
+ "commit_id": "1bc75db9bdb54c3fe1042c298690cf37132fe2bc",
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
+ 3.656931884765625,
42
+ 3.652819580078125,
43
+ 3.648908447265625,
44
+ 3.6574169921875,
45
+ 3.6517841796875
46
+ ],
47
+ "mean": 3.653572216796875,
48
+ "median": 3.652819580078125,
49
+ "std": 0.003211976078370725,
50
+ "min": 3.648908447265625,
51
+ "max": 3.6574169921875,
52
+ "p25": 3.6517841796875,
53
+ "p75": 3.656931884765625,
54
+ "p90": 3.6572229492187502,
55
+ "p95": 3.657319970703125,
56
+ "p99": 3.657397587890625,
57
+ "unit": "seconds"
58
+ },
59
+ "time_to_first_token_seconds": {
60
+ "name": "time_to_first_token_seconds",
61
+ "measurements": [
62
+ 0.03780694580078125,
63
+ 0.0372665901184082,
64
+ 0.03759430313110351,
65
+ 0.03751571273803711,
66
+ 0.0374453125
67
+ ],
68
+ "mean": 0.037525772857666014,
69
+ "median": 0.03751571273803711,
70
+ "std": 0.00017748543028055832,
71
+ "min": 0.0372665901184082,
72
+ "max": 0.03780694580078125,
73
+ "p25": 0.0374453125,
74
+ "p75": 0.03759430313110351,
75
+ "p90": 0.03772188873291015,
76
+ "p95": 0.0377644172668457,
77
+ "p99": 0.03779844009399414,
78
+ "unit": "seconds"
79
+ },
80
+ "tokens_per_second": {
81
+ "name": "tokens_per_second",
82
+ "measurements": [
83
+ 27.345327490672982,
84
+ 27.37611256394471,
85
+ 27.405456027524284,
86
+ 27.34170049890593,
87
+ 27.38387458827248
88
+ ],
89
+ "mean": 27.370494233864076,
90
+ "median": 27.37611256394471,
91
+ "std": 0.02406395922791954,
92
+ "min": 27.34170049890593,
93
+ "max": 27.405456027524284,
94
+ "p25": 27.345327490672982,
95
+ "p75": 27.38387458827248,
96
+ "p90": 27.39682345182356,
97
+ "p95": 27.401139739673923,
98
+ "p99": 27.404592769954213,
99
+ "unit": "tokens/sec"
100
+ },
101
+ "time_per_output_token_seconds": {
102
+ "name": "time_per_output_token_seconds",
103
+ "measurements": [
104
+ 0.03656931884765625,
105
+ 0.03652819580078125,
106
+ 0.03648908447265625,
107
+ 0.036574169921875004,
108
+ 0.036517841796875
109
+ ],
110
+ "mean": 0.03653572216796876,
111
+ "median": 0.03652819580078125,
112
+ "std": 3.211976078370685e-05,
113
+ "min": 0.03648908447265625,
114
+ "max": 0.036574169921875004,
115
+ "p25": 0.036517841796875,
116
+ "p75": 0.03656931884765625,
117
+ "p90": 0.036572229492187504,
118
+ "p95": 0.036573199707031254,
119
+ "p99": 0.036573975878906255,
120
+ "unit": "seconds/token"
121
+ }
122
+ },
123
+ "gpu_metrics": {
124
+ "gpu_utilization_mean": 91.02173913043478,
125
+ "gpu_utilization_max": 94,
126
+ "gpu_utilization_min": 87,
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": "2025-10-03T16:46:07.522395",
138
+ "commit_id": "1bc75db9bdb54c3fe1042c298690cf37132fe2bc",
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
+ 7.81993505859375,
172
+ 7.823923828125,
173
+ 7.8211259765625,
174
+ 7.88021435546875,
175
+ 7.8867529296875
176
+ ],
177
+ "mean": 7.8463904296875,
178
+ "median": 7.823923828125,
179
+ "std": 0.030384590827216867,
180
+ "min": 7.81993505859375,
181
+ "max": 7.8867529296875,
182
+ "p25": 7.8211259765625,
183
+ "p75": 7.88021435546875,
184
+ "p90": 7.8841375,
185
+ "p95": 7.88544521484375,
186
+ "p99": 7.88649138671875,
187
+ "unit": "seconds"
188
+ },
189
+ "time_to_first_token_seconds": {
190
+ "name": "time_to_first_token_seconds",
191
+ "measurements": [
192
+ 0.03965151977539062,
193
+ 0.038064159393310544,
194
+ 0.03812108612060547,
195
+ 0.038358047485351564,
196
+ 0.03805295944213867
197
+ ],
198
+ "mean": 0.03844955444335938,
199
+ "median": 0.03812108612060547,
200
+ "std": 0.0006110323332552849,
201
+ "min": 0.03805295944213867,
202
+ "max": 0.03965151977539062,
203
+ "p25": 0.038064159393310544,
204
+ "p75": 0.038358047485351564,
205
+ "p90": 0.039134130859375,
206
+ "p95": 0.039392825317382814,
207
+ "p99": 0.039599780883789064,
208
+ "unit": "seconds"
209
+ },
210
+ "tokens_per_second": {
211
+ "name": "tokens_per_second",
212
+ "measurements": [
213
+ 12.787829982053953,
214
+ 12.781310528679438,
215
+ 12.785882787167619,
216
+ 12.690010130321076,
217
+ 12.679489378141625
218
+ ],
219
+ "mean": 12.744904561272744,
220
+ "median": 12.781310528679438,
221
+ "std": 0.04927422438382531,
222
+ "min": 12.679489378141625,
223
+ "max": 12.787829982053953,
224
+ "p25": 12.690010130321076,
225
+ "p75": 12.785882787167619,
226
+ "p90": 12.787051104099419,
227
+ "p95": 12.787440543076686,
228
+ "p99": 12.787752094258499,
229
+ "unit": "tokens/sec"
230
+ },
231
+ "time_per_output_token_seconds": {
232
+ "name": "time_per_output_token_seconds",
233
+ "measurements": [
234
+ 0.07819935058593751,
235
+ 0.07823923828125,
236
+ 0.078211259765625,
237
+ 0.0788021435546875,
238
+ 0.078867529296875
239
+ ],
240
+ "mean": 0.07846390429687501,
241
+ "median": 0.07823923828125,
242
+ "std": 0.0003038459082721682,
243
+ "min": 0.07819935058593751,
244
+ "max": 0.078867529296875,
245
+ "p25": 0.078211259765625,
246
+ "p75": 0.0788021435546875,
247
+ "p90": 0.078841375,
248
+ "p95": 0.0788544521484375,
249
+ "p99": 0.0788649138671875,
250
+ "unit": "seconds/token"
251
+ }
252
+ },
253
+ "gpu_metrics": {
254
+ "gpu_utilization_mean": 43.917647058823526,
255
+ "gpu_utilization_max": 91,
256
+ "gpu_utilization_min": 0,
257
+ "gpu_memory_used_mean": 13629.69411764706,
258
+ "gpu_memory_used_max": 13777,
259
+ "gpu_memory_used_min": 13415,
260
+ "sample_count": 170,
261
+ "gpu_monitoring_status": "success"
262
+ }
263
+ },
264
+ {
265
+ "scenario_name": "eager_sdpa_default",
266
+ "metadata": {
267
+ "timestamp": "2025-10-03T16:47:35.225284",
268
+ "commit_id": "1bc75db9bdb54c3fe1042c298690cf37132fe2bc",
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
+ 3.412320556640625,
302
+ 3.41054248046875,
303
+ 3.41186669921875,
304
+ 3.4123642578125,
305
+ 3.41188916015625
306
+ ],
307
+ "mean": 3.4117966308593752,
308
+ "median": 3.41188916015625,
309
+ "std": 0.0006607666804920782,
310
+ "min": 3.41054248046875,
311
+ "max": 3.4123642578125,
312
+ "p25": 3.41186669921875,
313
+ "p75": 3.412320556640625,
314
+ "p90": 3.41234677734375,
315
+ "p95": 3.412355517578125,
316
+ "p99": 3.412362509765625,
317
+ "unit": "seconds"
318
+ },
319
+ "time_to_first_token_seconds": {
320
+ "name": "time_to_first_token_seconds",
321
+ "measurements": [
322
+ 0.03516684722900391,
323
+ 0.03465574264526367,
324
+ 0.03457427215576172,
325
+ 0.03470528030395508,
326
+ 0.034661376953125
327
+ ],
328
+ "mean": 0.03475270385742187,
329
+ "median": 0.034661376953125,
330
+ "std": 0.00021134979437781713,
331
+ "min": 0.03457427215576172,
332
+ "max": 0.03516684722900391,
333
+ "p25": 0.03465574264526367,
334
+ "p75": 0.03470528030395508,
335
+ "p90": 0.03498222045898438,
336
+ "p95": 0.03507453384399414,
337
+ "p99": 0.035148384552001956,
338
+ "unit": "seconds"
339
+ },
340
+ "tokens_per_second": {
341
+ "name": "tokens_per_second",
342
+ "measurements": [
343
+ 29.305570312083574,
344
+ 29.320848683947737,
345
+ 29.309468632786277,
346
+ 29.30519500403662,
347
+ 29.30927568450683
348
+ ],
349
+ "mean": 29.310071663472208,
350
+ "median": 29.30927568450683,
351
+ "std": 0.00567777016840497,
352
+ "min": 29.30519500403662,
353
+ "max": 29.320848683947737,
354
+ "p25": 29.305570312083574,
355
+ "p75": 29.309468632786277,
356
+ "p90": 29.316296663483154,
357
+ "p95": 29.318572673715444,
358
+ "p99": 29.320393481901277,
359
+ "unit": "tokens/sec"
360
+ },
361
+ "time_per_output_token_seconds": {
362
+ "name": "time_per_output_token_seconds",
363
+ "measurements": [
364
+ 0.03412320556640625,
365
+ 0.0341054248046875,
366
+ 0.0341186669921875,
367
+ 0.034123642578125,
368
+ 0.0341188916015625
369
+ ],
370
+ "mean": 0.03411796630859375,
371
+ "median": 0.0341188916015625,
372
+ "std": 6.60766680492206e-06,
373
+ "min": 0.0341054248046875,
374
+ "max": 0.034123642578125,
375
+ "p25": 0.0341186669921875,
376
+ "p75": 0.03412320556640625,
377
+ "p90": 0.0341234677734375,
378
+ "p95": 0.03412355517578125,
379
+ "p99": 0.03412362509765625,
380
+ "unit": "seconds/token"
381
+ }
382
+ },
383
+ "gpu_metrics": {
384
+ "gpu_utilization_mean": 96.86046511627907,
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": "2025-10-03T16:48:07.718712",
398
+ "commit_id": "1bc75db9bdb54c3fe1042c298690cf37132fe2bc",
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
+ 3.530971435546875,
432
+ 3.530159423828125,
433
+ 3.53622509765625,
434
+ 3.530225830078125,
435
+ 3.53467578125
436
+ ],
437
+ "mean": 3.5324515136718753,
438
+ "median": 3.530971435546875,
439
+ "std": 0.0025133752968326565,
440
+ "min": 3.530159423828125,
441
+ "max": 3.53622509765625,
442
+ "p25": 3.530225830078125,
443
+ "p75": 3.53467578125,
444
+ "p90": 3.53560537109375,
445
+ "p95": 3.535915234375,
446
+ "p99": 3.5361631250000003,
447
+ "unit": "seconds"
448
+ },
449
+ "time_to_first_token_seconds": {
450
+ "name": "time_to_first_token_seconds",
451
+ "measurements": [
452
+ 0.03802233505249023,
453
+ 0.037365150451660153,
454
+ 0.04067356872558594,
455
+ 0.037418495178222655,
456
+ 0.0393026237487793
457
+ ],
458
+ "mean": 0.038556434631347655,
459
+ "median": 0.03802233505249023,
460
+ "std": 0.0012679417595337253,
461
+ "min": 0.037365150451660153,
462
+ "max": 0.04067356872558594,
463
+ "p25": 0.037418495178222655,
464
+ "p75": 0.0393026237487793,
465
+ "p90": 0.04012519073486328,
466
+ "p95": 0.04039937973022461,
467
+ "p99": 0.04061873092651367,
468
+ "unit": "seconds"
469
+ },
470
+ "tokens_per_second": {
471
+ "name": "tokens_per_second",
472
+ "measurements": [
473
+ 28.320818172949068,
474
+ 28.327332563230087,
475
+ 28.278742794478298,
476
+ 28.326799704422015,
477
+ 28.291137911561464
478
+ ],
479
+ "mean": 28.30896622932819,
480
+ "median": 28.320818172949068,
481
+ "std": 0.020135185874620407,
482
+ "min": 28.278742794478298,
483
+ "max": 28.327332563230087,
484
+ "p25": 28.291137911561464,
485
+ "p75": 28.326799704422015,
486
+ "p90": 28.327119419706857,
487
+ "p95": 28.327225991468474,
488
+ "p99": 28.327311248877763,
489
+ "unit": "tokens/sec"
490
+ },
491
+ "time_per_output_token_seconds": {
492
+ "name": "time_per_output_token_seconds",
493
+ "measurements": [
494
+ 0.03530971435546875,
495
+ 0.03530159423828125,
496
+ 0.0353622509765625,
497
+ 0.03530225830078125,
498
+ 0.0353467578125
499
+ ],
500
+ "mean": 0.035324515136718755,
501
+ "median": 0.03530971435546875,
502
+ "std": 2.5133752968327e-05,
503
+ "min": 0.03530159423828125,
504
+ "max": 0.0353622509765625,
505
+ "p25": 0.03530225830078125,
506
+ "p75": 0.0353467578125,
507
+ "p90": 0.0353560537109375,
508
+ "p95": 0.03535915234375,
509
+ "p99": 0.035361631250000004,
510
+ "unit": "seconds/token"
511
+ }
512
+ },
513
+ "gpu_metrics": {
514
+ "gpu_utilization_mean": 96.2247191011236,
515
+ "gpu_utilization_max": 98,
516
+ "gpu_utilization_min": 91,
517
+ "gpu_memory_used_mean": 13849,
518
+ "gpu_memory_used_max": 13849,
519
+ "gpu_memory_used_min": 13849,
520
+ "sample_count": 89,
521
+ "gpu_monitoring_status": "success"
522
+ }
523
+ },
524
+ {
525
+ "scenario_name": "eager_sdpa_flash_attention",
526
+ "metadata": {
527
+ "timestamp": "2025-10-03T16:48:41.240740",
528
+ "commit_id": "1bc75db9bdb54c3fe1042c298690cf37132fe2bc",
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
+ 3.410536376953125,
562
+ 3.411936767578125,
563
+ 3.407611572265625,
564
+ 3.40797998046875,
565
+ 3.410041748046875
566
+ ],
567
+ "mean": 3.4096212890624997,
568
+ "median": 3.410041748046875,
569
+ "std": 0.0016191644675567429,
570
+ "min": 3.407611572265625,
571
+ "max": 3.411936767578125,
572
+ "p25": 3.40797998046875,
573
+ "p75": 3.410536376953125,
574
+ "p90": 3.411376611328125,
575
+ "p95": 3.411656689453125,
576
+ "p99": 3.411880751953125,
577
+ "unit": "seconds"
578
+ },
579
+ "time_to_first_token_seconds": {
580
+ "name": "time_to_first_token_seconds",
581
+ "measurements": [
582
+ 0.03548748779296875,
583
+ 0.03466976165771484,
584
+ 0.03462643051147461,
585
+ 0.034662944793701175,
586
+ 0.034669662475585936
587
+ ],
588
+ "mean": 0.03482325744628907,
589
+ "median": 0.034669662475585936,
590
+ "std": 0.0003325042298779828,
591
+ "min": 0.03462643051147461,
592
+ "max": 0.03548748779296875,
593
+ "p25": 0.034662944793701175,
594
+ "p75": 0.03466976165771484,
595
+ "p90": 0.03516039733886719,
596
+ "p95": 0.03532394256591797,
597
+ "p99": 0.0354547787475586,
598
+ "unit": "seconds"
599
+ },
600
+ "tokens_per_second": {
601
+ "name": "tokens_per_second",
602
+ "measurements": [
603
+ 29.320901156708118,
604
+ 29.30886672644359,
605
+ 29.346067730810297,
606
+ 29.342895372949204,
607
+ 29.32515417363312
608
+ ],
609
+ "mean": 29.328777032108867,
610
+ "median": 29.32515417363312,
611
+ "std": 0.013927459195530038,
612
+ "min": 29.30886672644359,
613
+ "max": 29.346067730810297,
614
+ "p25": 29.320901156708118,
615
+ "p75": 29.342895372949204,
616
+ "p90": 29.34479878766586,
617
+ "p95": 29.345433259238078,
618
+ "p99": 29.345940836495853,
619
+ "unit": "tokens/sec"
620
+ },
621
+ "time_per_output_token_seconds": {
622
+ "name": "time_per_output_token_seconds",
623
+ "measurements": [
624
+ 0.03410536376953125,
625
+ 0.03411936767578125,
626
+ 0.03407611572265625,
627
+ 0.0340797998046875,
628
+ 0.03410041748046875
629
+ ],
630
+ "mean": 0.034096212890625,
631
+ "median": 0.03410041748046875,
632
+ "std": 1.6191644675566156e-05,
633
+ "min": 0.03407611572265625,
634
+ "max": 0.03411936767578125,
635
+ "p25": 0.0340797998046875,
636
+ "p75": 0.03410536376953125,
637
+ "p90": 0.03411376611328125,
638
+ "p95": 0.03411656689453125,
639
+ "p99": 0.034118807519531245,
640
+ "unit": "seconds/token"
641
+ }
642
+ },
643
+ "gpu_metrics": {
644
+ "gpu_utilization_mean": 96.88372093023256,
645
+ "gpu_utilization_max": 98,
646
+ "gpu_utilization_min": 93,
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": "2025-10-03T16:49:13.777826",
658
+ "commit_id": "1bc75db9bdb54c3fe1042c298690cf37132fe2bc",
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
+ 3.41007568359375,
692
+ 3.409101806640625,
693
+ 3.412410400390625,
694
+ 3.413089111328125,
695
+ 3.408107177734375
696
+ ],
697
+ "mean": 3.4105568359375,
698
+ "median": 3.41007568359375,
699
+ "std": 0.0019077506640361947,
700
+ "min": 3.408107177734375,
701
+ "max": 3.413089111328125,
702
+ "p25": 3.409101806640625,
703
+ "p75": 3.412410400390625,
704
+ "p90": 3.412817626953125,
705
+ "p95": 3.412953369140625,
706
+ "p99": 3.4130619628906254,
707
+ "unit": "seconds"
708
+ },
709
+ "time_to_first_token_seconds": {
710
+ "name": "time_to_first_token_seconds",
711
+ "measurements": [
712
+ 0.035283329010009766,
713
+ 0.03476144027709961,
714
+ 0.034805248260498044,
715
+ 0.03475081634521485,
716
+ 0.03475177764892578
717
+ ],
718
+ "mean": 0.03487052230834961,
719
+ "median": 0.03476144027709961,
720
+ "std": 0.00020736382305107812,
721
+ "min": 0.03475081634521485,
722
+ "max": 0.035283329010009766,
723
+ "p25": 0.03475177764892578,
724
+ "p75": 0.034805248260498044,
725
+ "p90": 0.03509209671020508,
726
+ "p95": 0.03518771286010742,
727
+ "p99": 0.0352642057800293,
728
+ "unit": "seconds"
729
+ },
730
+ "tokens_per_second": {
731
+ "name": "tokens_per_second",
732
+ "measurements": [
733
+ 29.324862342824538,
734
+ 29.33323956627196,
735
+ 29.304798739493002,
736
+ 29.29897132427559,
737
+ 29.341800238359145
738
+ ],
739
+ "mean": 29.32073444224485,
740
+ "median": 29.324862342824538,
741
+ "std": 0.01639978891325966,
742
+ "min": 29.29897132427559,
743
+ "max": 29.341800238359145,
744
+ "p25": 29.304798739493002,
745
+ "p75": 29.33323956627196,
746
+ "p90": 29.33837596952427,
747
+ "p95": 29.340088103941707,
748
+ "p99": 29.341457811475657,
749
+ "unit": "tokens/sec"
750
+ },
751
+ "time_per_output_token_seconds": {
752
+ "name": "time_per_output_token_seconds",
753
+ "measurements": [
754
+ 0.0341007568359375,
755
+ 0.03409101806640625,
756
+ 0.03412410400390625,
757
+ 0.03413089111328125,
758
+ 0.03408107177734375
759
+ ],
760
+ "mean": 0.034105568359375005,
761
+ "median": 0.0341007568359375,
762
+ "std": 1.907750664036188e-05,
763
+ "min": 0.03408107177734375,
764
+ "max": 0.03413089111328125,
765
+ "p25": 0.03409101806640625,
766
+ "p75": 0.03412410400390625,
767
+ "p90": 0.03412817626953125,
768
+ "p95": 0.03412953369140625,
769
+ "p99": 0.03413061962890625,
770
+ "unit": "seconds/token"
771
+ }
772
+ },
773
+ "gpu_metrics": {
774
+ "gpu_utilization_mean": 96.88372093023256,
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": "2025-10-03T16:49:46.283344",
788
+ "commit_id": "1bc75db9bdb54c3fe1042c298690cf37132fe2bc",
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
+ 3.690556884765625,
822
+ 3.697396240234375,
823
+ 3.695514404296875,
824
+ 3.694315673828125,
825
+ 3.69690625
826
+ ],
827
+ "mean": 3.6949378906250003,
828
+ "median": 3.695514404296875,
829
+ "std": 0.0024425664996930393,
830
+ "min": 3.690556884765625,
831
+ "max": 3.697396240234375,
832
+ "p25": 3.694315673828125,
833
+ "p75": 3.69690625,
834
+ "p90": 3.697200244140625,
835
+ "p95": 3.6972982421875,
836
+ "p99": 3.697376640625,
837
+ "unit": "seconds"
838
+ },
839
+ "time_to_first_token_seconds": {
840
+ "name": "time_to_first_token_seconds",
841
+ "measurements": [
842
+ 0.03800764846801758,
843
+ 0.03742707061767578,
844
+ 0.03740182495117188,
845
+ 0.037477375030517575,
846
+ 0.0373744010925293
847
+ ],
848
+ "mean": 0.03753766403198242,
849
+ "median": 0.03742707061767578,
850
+ "std": 0.00023742707527796563,
851
+ "min": 0.0373744010925293,
852
+ "max": 0.03800764846801758,
853
+ "p25": 0.03740182495117188,
854
+ "p75": 0.037477375030517575,
855
+ "p90": 0.037795539093017576,
856
+ "p95": 0.03790159378051758,
857
+ "p99": 0.03798643753051758,
858
+ "unit": "seconds"
859
+ },
860
+ "tokens_per_second": {
861
+ "name": "tokens_per_second",
862
+ "measurements": [
863
+ 27.096181720648552,
864
+ 27.04605984931198,
865
+ 27.059832288497454,
866
+ 27.06861265495971,
867
+ 27.049644550764576
868
+ ],
869
+ "mean": 27.064066212836458,
870
+ "median": 27.059832288497454,
871
+ "std": 0.017900924878946206,
872
+ "min": 27.04605984931198,
873
+ "max": 27.096181720648552,
874
+ "p25": 27.049644550764576,
875
+ "p75": 27.06861265495971,
876
+ "p90": 27.085154094373017,
877
+ "p95": 27.090667907510785,
878
+ "p99": 27.095078958021,
879
+ "unit": "tokens/sec"
880
+ },
881
+ "time_per_output_token_seconds": {
882
+ "name": "time_per_output_token_seconds",
883
+ "measurements": [
884
+ 0.03690556884765625,
885
+ 0.03697396240234375,
886
+ 0.03695514404296875,
887
+ 0.036943156738281246,
888
+ 0.036969062500000004
889
+ ],
890
+ "mean": 0.036949378906250004,
891
+ "median": 0.03695514404296875,
892
+ "std": 2.4425664996931033e-05,
893
+ "min": 0.03690556884765625,
894
+ "max": 0.03697396240234375,
895
+ "p25": 0.036943156738281246,
896
+ "p75": 0.036969062500000004,
897
+ "p90": 0.03697200244140625,
898
+ "p95": 0.036972982421875,
899
+ "p99": 0.03697376640625,
900
+ "unit": "seconds/token"
901
+ }
902
+ },
903
+ "gpu_metrics": {
904
+ "gpu_utilization_mean": 90.50537634408602,
905
+ "gpu_utilization_max": 93,
906
+ "gpu_utilization_min": 88,
907
+ "gpu_memory_used_mean": 13845,
908
+ "gpu_memory_used_max": 13845,
909
+ "gpu_memory_used_min": 13845,
910
+ "sample_count": 93,
911
+ "gpu_monitoring_status": "success"
912
+ }
913
+ },
914
+ {
915
+ "scenario_name": "compiled_compile_max-autotune_sdpa_math",
916
+ "metadata": {
917
+ "timestamp": "2025-10-03T16:50:21.045345",
918
+ "commit_id": "1bc75db9bdb54c3fe1042c298690cf37132fe2bc",
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
+ 3.858753173828125,
952
+ 3.85711181640625,
953
+ 3.843275634765625,
954
+ 3.85389111328125,
955
+ 3.8894443359375
956
+ ],
957
+ "mean": 3.8604952148437497,
958
+ "median": 3.85711181640625,
959
+ "std": 0.015444492866406303,
960
+ "min": 3.843275634765625,
961
+ "max": 3.8894443359375,
962
+ "p25": 3.85389111328125,
963
+ "p75": 3.858753173828125,
964
+ "p90": 3.87716787109375,
965
+ "p95": 3.883306103515625,
966
+ "p99": 3.888216689453125,
967
+ "unit": "seconds"
968
+ },
969
+ "time_to_first_token_seconds": {
970
+ "name": "time_to_first_token_seconds",
971
+ "measurements": [
972
+ 0.04055635070800781,
973
+ 0.04046361541748047,
974
+ 0.039589054107666014,
975
+ 0.041204959869384765,
976
+ 0.04120367813110352
977
+ ],
978
+ "mean": 0.040603531646728516,
979
+ "median": 0.04055635070800781,
980
+ "std": 0.0005954593150103537,
981
+ "min": 0.039589054107666014,
982
+ "max": 0.041204959869384765,
983
+ "p25": 0.04046361541748047,
984
+ "p75": 0.04120367813110352,
985
+ "p90": 0.04120444717407227,
986
+ "p95": 0.04120470352172852,
987
+ "p99": 0.04120490859985351,
988
+ "unit": "seconds"
989
+ },
990
+ "tokens_per_second": {
991
+ "name": "tokens_per_second",
992
+ "measurements": [
993
+ 25.915106640727096,
994
+ 25.926134569044475,
995
+ 26.01947128002395,
996
+ 25.947801082231088,
997
+ 25.710613486873903
998
+ ],
999
+ "mean": 25.903825411780105,
1000
+ "median": 25.926134569044475,
1001
+ "std": 0.10321205257512005,
1002
+ "min": 25.710613486873903,
1003
+ "max": 26.01947128002395,
1004
+ "p25": 25.915106640727096,
1005
+ "p75": 25.947801082231088,
1006
+ "p90": 25.990803200906804,
1007
+ "p95": 26.005137240465377,
1008
+ "p99": 26.016604472112235,
1009
+ "unit": "tokens/sec"
1010
+ },
1011
+ "time_per_output_token_seconds": {
1012
+ "name": "time_per_output_token_seconds",
1013
+ "measurements": [
1014
+ 0.03858753173828125,
1015
+ 0.0385711181640625,
1016
+ 0.03843275634765625,
1017
+ 0.0385389111328125,
1018
+ 0.038894443359375
1019
+ ],
1020
+ "mean": 0.038604952148437496,
1021
+ "median": 0.0385711181640625,
1022
+ "std": 0.000154444928664064,
1023
+ "min": 0.03843275634765625,
1024
+ "max": 0.038894443359375,
1025
+ "p25": 0.0385389111328125,
1026
+ "p75": 0.03858753173828125,
1027
+ "p90": 0.0387716787109375,
1028
+ "p95": 0.038833061035156255,
1029
+ "p99": 0.03888216689453125,
1030
+ "unit": "seconds/token"
1031
+ }
1032
+ },
1033
+ "gpu_metrics": {
1034
+ "gpu_utilization_mean": 89.84536082474227,
1035
+ "gpu_utilization_max": 93,
1036
+ "gpu_utilization_min": 83,
1037
+ "gpu_memory_used_mean": 13845,
1038
+ "gpu_memory_used_max": 13845,
1039
+ "gpu_memory_used_min": 13845,
1040
+ "sample_count": 97,
1041
+ "gpu_monitoring_status": "success"
1042
+ }
1043
+ },
1044
+ {
1045
+ "scenario_name": "compiled_compile_max-autotune_sdpa_efficient_attention",
1046
+ "metadata": {
1047
+ "timestamp": "2025-10-03T16:51:01.875608",
1048
+ "commit_id": "1bc75db9bdb54c3fe1042c298690cf37132fe2bc",
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
+ 3.692674560546875,
1082
+ 3.689333740234375,
1083
+ 3.689283447265625,
1084
+ 3.6909326171875,
1085
+ 3.687671630859375
1086
+ ],
1087
+ "mean": 3.6899791992187496,
1088
+ "median": 3.689333740234375,
1089
+ "std": 0.0016970290702438256,
1090
+ "min": 3.687671630859375,
1091
+ "max": 3.692674560546875,
1092
+ "p25": 3.689283447265625,
1093
+ "p75": 3.6909326171875,
1094
+ "p90": 3.691977783203125,
1095
+ "p95": 3.692326171875,
1096
+ "p99": 3.6926048828124998,
1097
+ "unit": "seconds"
1098
+ },
1099
+ "time_to_first_token_seconds": {
1100
+ "name": "time_to_first_token_seconds",
1101
+ "measurements": [
1102
+ 0.037977535247802736,
1103
+ 0.03750332641601563,
1104
+ 0.03746377563476563,
1105
+ 0.03748422241210937,
1106
+ 0.037422046661376954
1107
+ ],
1108
+ "mean": 0.03757018127441406,
1109
+ "median": 0.03748422241210937,
1110
+ "std": 0.0002054564583868702,
1111
+ "min": 0.037422046661376954,
1112
+ "max": 0.037977535247802736,
1113
+ "p25": 0.03746377563476563,
1114
+ "p75": 0.03750332641601563,
1115
+ "p90": 0.037787851715087896,
1116
+ "p95": 0.037882693481445316,
1117
+ "p99": 0.037958566894531254,
1118
+ "unit": "seconds"
1119
+ },
1120
+ "tokens_per_second": {
1121
+ "name": "tokens_per_second",
1122
+ "measurements": [
1123
+ 27.08064259667396,
1124
+ 27.10516506258044,
1125
+ 27.105534565016058,
1126
+ 27.09342336252138,
1127
+ 27.117381917406785
1128
+ ],
1129
+ "mean": 27.100429500839727,
1130
+ "median": 27.10516506258044,
1131
+ "std": 0.012461775180997474,
1132
+ "min": 27.08064259667396,
1133
+ "max": 27.117381917406785,
1134
+ "p25": 27.09342336252138,
1135
+ "p75": 27.105534565016058,
1136
+ "p90": 27.112642976450495,
1137
+ "p95": 27.115012446928638,
1138
+ "p99": 27.116908023311154,
1139
+ "unit": "tokens/sec"
1140
+ },
1141
+ "time_per_output_token_seconds": {
1142
+ "name": "time_per_output_token_seconds",
1143
+ "measurements": [
1144
+ 0.03692674560546875,
1145
+ 0.036893337402343754,
1146
+ 0.036892834472656254,
1147
+ 0.036909326171875,
1148
+ 0.03687671630859375
1149
+ ],
1150
+ "mean": 0.0368997919921875,
1151
+ "median": 0.036893337402343754,
1152
+ "std": 1.697029070243687e-05,
1153
+ "min": 0.03687671630859375,
1154
+ "max": 0.03692674560546875,
1155
+ "p25": 0.036892834472656254,
1156
+ "p75": 0.036909326171875,
1157
+ "p90": 0.03691977783203125,
1158
+ "p95": 0.036923261718749996,
1159
+ "p99": 0.036926048828124995,
1160
+ "unit": "seconds/token"
1161
+ }
1162
+ },
1163
+ "gpu_metrics": {
1164
+ "gpu_utilization_mean": 90.61290322580645,
1165
+ "gpu_utilization_max": 93,
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
+ }