hf-transformers-bot commited on
Commit
ab34115
·
verified ·
1 Parent(s): 3a4396e

Upload benchmark results for run 18472480065

Browse files
2025-10-13/18472480065/benchmark_results/Llama-2-7b-hf/Llama-2-7b-hf_benchmark_20251013_165257.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": "2025-10-13T16:46:54.507478",
8
+ "commit_id": "b84c0b31c674436b076bba818885226a0e0ddecc",
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
+ 3.6662138671875,
42
+ 3.6625986328125,
43
+ 3.670444580078125,
44
+ 3.672843994140625,
45
+ 3.665005126953125
46
+ ],
47
+ "mean": 3.6674212402343747,
48
+ "median": 3.6662138671875,
49
+ "std": 0.0037175190422169186,
50
+ "min": 3.6625986328125,
51
+ "max": 3.672843994140625,
52
+ "p25": 3.665005126953125,
53
+ "p75": 3.670444580078125,
54
+ "p90": 3.671884228515625,
55
+ "p95": 3.672364111328125,
56
+ "p99": 3.6727480175781246,
57
+ "unit": "seconds"
58
+ },
59
+ "time_to_first_token_seconds": {
60
+ "name": "time_to_first_token_seconds",
61
+ "measurements": [
62
+ 0.03765459060668945,
63
+ 0.03727609634399414,
64
+ 0.037142593383789065,
65
+ 0.03947603225708008,
66
+ 0.03715190505981445
67
+ ],
68
+ "mean": 0.03774024353027343,
69
+ "median": 0.03727609634399414,
70
+ "std": 0.0008875876735707083,
71
+ "min": 0.037142593383789065,
72
+ "max": 0.03947603225708008,
73
+ "p25": 0.03715190505981445,
74
+ "p75": 0.03765459060668945,
75
+ "p90": 0.03874745559692383,
76
+ "p95": 0.03911174392700195,
77
+ "p99": 0.03940317459106445,
78
+ "unit": "seconds"
79
+ },
80
+ "tokens_per_second": {
81
+ "name": "tokens_per_second",
82
+ "measurements": [
83
+ 27.27609561869723,
84
+ 27.303018983330492,
85
+ 27.244656013270063,
86
+ 27.2268574868773,
87
+ 27.285091435365675
88
+ ],
89
+ "mean": 27.267143907508153,
90
+ "median": 27.27609561869723,
91
+ "std": 0.027633201150707498,
92
+ "min": 27.2268574868773,
93
+ "max": 27.303018983330492,
94
+ "p25": 27.244656013270063,
95
+ "p75": 27.285091435365675,
96
+ "p90": 27.295847964144567,
97
+ "p95": 27.29943347373753,
98
+ "p99": 27.3023018814119,
99
+ "unit": "tokens/sec"
100
+ },
101
+ "time_per_output_token_seconds": {
102
+ "name": "time_per_output_token_seconds",
103
+ "measurements": [
104
+ 0.036662138671875,
105
+ 0.036625986328125,
106
+ 0.03670444580078125,
107
+ 0.036728439941406246,
108
+ 0.036650051269531254
109
+ ],
110
+ "mean": 0.036674212402343746,
111
+ "median": 0.036662138671875,
112
+ "std": 3.717519042216942e-05,
113
+ "min": 0.036625986328125,
114
+ "max": 0.036728439941406246,
115
+ "p25": 0.036650051269531254,
116
+ "p75": 0.03670444580078125,
117
+ "p90": 0.03671884228515625,
118
+ "p95": 0.03672364111328125,
119
+ "p99": 0.036727480175781244,
120
+ "unit": "seconds/token"
121
+ }
122
+ },
123
+ "gpu_metrics": {
124
+ "gpu_utilization_mean": 90.42391304347827,
125
+ "gpu_utilization_max": 94,
126
+ "gpu_utilization_min": 85,
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": "2025-10-13T16:47:29.937039",
138
+ "commit_id": "b84c0b31c674436b076bba818885226a0e0ddecc",
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
+ 7.7419560546875,
172
+ 7.75443115234375,
173
+ 7.825423828125,
174
+ 7.9065908203125,
175
+ 7.917876953125
176
+ ],
177
+ "mean": 7.829255761718751,
178
+ "median": 7.825423828125,
179
+ "std": 0.07357862676741425,
180
+ "min": 7.7419560546875,
181
+ "max": 7.917876953125,
182
+ "p25": 7.75443115234375,
183
+ "p75": 7.9065908203125,
184
+ "p90": 7.9133625,
185
+ "p95": 7.9156197265625,
186
+ "p99": 7.9174255078125,
187
+ "unit": "seconds"
188
+ },
189
+ "time_to_first_token_seconds": {
190
+ "name": "time_to_first_token_seconds",
191
+ "measurements": [
192
+ 0.038714111328125,
193
+ 0.03806723022460937,
194
+ 0.038072254180908205,
195
+ 0.038692672729492186,
196
+ 0.038360671997070314
197
+ ],
198
+ "mean": 0.038381388092041016,
199
+ "median": 0.038360671997070314,
200
+ "std": 0.0002836515378472705,
201
+ "min": 0.03806723022460937,
202
+ "max": 0.038714111328125,
203
+ "p25": 0.038072254180908205,
204
+ "p75": 0.038692672729492186,
205
+ "p90": 0.03870553588867188,
206
+ "p95": 0.03870982360839844,
207
+ "p99": 0.03871325378417969,
208
+ "unit": "seconds"
209
+ },
210
+ "tokens_per_second": {
211
+ "name": "tokens_per_second",
212
+ "measurements": [
213
+ 12.916632346350415,
214
+ 12.895852453313141,
215
+ 12.778860569902239,
216
+ 12.64767613155016,
217
+ 12.629648148362845
218
+ ],
219
+ "mean": 12.77373392989576,
220
+ "median": 12.778860569902239,
221
+ "std": 0.12000871184071299,
222
+ "min": 12.629648148362845,
223
+ "max": 12.916632346350415,
224
+ "p25": 12.64767613155016,
225
+ "p75": 12.895852453313141,
226
+ "p90": 12.908320389135506,
227
+ "p95": 12.91247636774296,
228
+ "p99": 12.915801150628925,
229
+ "unit": "tokens/sec"
230
+ },
231
+ "time_per_output_token_seconds": {
232
+ "name": "time_per_output_token_seconds",
233
+ "measurements": [
234
+ 0.077419560546875,
235
+ 0.0775443115234375,
236
+ 0.07825423828125,
237
+ 0.07906590820312501,
238
+ 0.07917876953125
239
+ ],
240
+ "mean": 0.0782925576171875,
241
+ "median": 0.07825423828125,
242
+ "std": 0.0007357862676741423,
243
+ "min": 0.077419560546875,
244
+ "max": 0.07917876953125,
245
+ "p25": 0.0775443115234375,
246
+ "p75": 0.07906590820312501,
247
+ "p90": 0.079133625,
248
+ "p95": 0.079156197265625,
249
+ "p99": 0.079174255078125,
250
+ "unit": "seconds/token"
251
+ }
252
+ },
253
+ "gpu_metrics": {
254
+ "gpu_utilization_mean": 45.11309523809524,
255
+ "gpu_utilization_max": 92,
256
+ "gpu_utilization_min": 0,
257
+ "gpu_memory_used_mean": 13630.047619047618,
258
+ "gpu_memory_used_max": 13777,
259
+ "gpu_memory_used_min": 13415,
260
+ "sample_count": 168,
261
+ "gpu_monitoring_status": "success"
262
+ }
263
+ },
264
+ {
265
+ "scenario_name": "eager_sdpa_default",
266
+ "metadata": {
267
+ "timestamp": "2025-10-13T16:48:57.989311",
268
+ "commit_id": "b84c0b31c674436b076bba818885226a0e0ddecc",
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
+ 3.416404052734375,
302
+ 3.415478759765625,
303
+ 3.4159775390625,
304
+ 3.416638671875,
305
+ 3.415705810546875
306
+ ],
307
+ "mean": 3.416040966796875,
308
+ "median": 3.4159775390625,
309
+ "std": 0.0004293046223407779,
310
+ "min": 3.415478759765625,
311
+ "max": 3.416638671875,
312
+ "p25": 3.415705810546875,
313
+ "p75": 3.416404052734375,
314
+ "p90": 3.41654482421875,
315
+ "p95": 3.416591748046875,
316
+ "p99": 3.416629287109375,
317
+ "unit": "seconds"
318
+ },
319
+ "time_to_first_token_seconds": {
320
+ "name": "time_to_first_token_seconds",
321
+ "measurements": [
322
+ 0.03509587097167969,
323
+ 0.03463907241821289,
324
+ 0.0345824966430664,
325
+ 0.03463164901733398,
326
+ 0.03459139251708984
327
+ ],
328
+ "mean": 0.03470809631347656,
329
+ "median": 0.03463164901733398,
330
+ "std": 0.00019512699764391797,
331
+ "min": 0.0345824966430664,
332
+ "max": 0.03509587097167969,
333
+ "p25": 0.03459139251708984,
334
+ "p75": 0.03463907241821289,
335
+ "p90": 0.03491315155029297,
336
+ "p95": 0.03500451126098633,
337
+ "p99": 0.035077599029541016,
338
+ "unit": "seconds"
339
+ },
340
+ "tokens_per_second": {
341
+ "name": "tokens_per_second",
342
+ "measurements": [
343
+ 29.270542493345705,
344
+ 29.278472224158158,
345
+ 29.274197168007305,
346
+ 29.268532497503315,
347
+ 29.27652600854094
348
+ ],
349
+ "mean": 29.273654078311086,
350
+ "median": 29.274197168007305,
351
+ "std": 0.003678858791371557,
352
+ "min": 29.268532497503315,
353
+ "max": 29.278472224158158,
354
+ "p25": 29.270542493345705,
355
+ "p75": 29.27652600854094,
356
+ "p90": 29.27769373791127,
357
+ "p95": 29.278082981034714,
358
+ "p99": 29.27839437553347,
359
+ "unit": "tokens/sec"
360
+ },
361
+ "time_per_output_token_seconds": {
362
+ "name": "time_per_output_token_seconds",
363
+ "measurements": [
364
+ 0.03416404052734375,
365
+ 0.03415478759765625,
366
+ 0.034159775390625,
367
+ 0.03416638671875,
368
+ 0.03415705810546875
369
+ ],
370
+ "mean": 0.03416040966796875,
371
+ "median": 0.034159775390625,
372
+ "std": 4.293046223407007e-06,
373
+ "min": 0.03415478759765625,
374
+ "max": 0.03416638671875,
375
+ "p25": 0.03415705810546875,
376
+ "p75": 0.03416404052734375,
377
+ "p90": 0.0341654482421875,
378
+ "p95": 0.034165917480468745,
379
+ "p99": 0.03416629287109375,
380
+ "unit": "seconds/token"
381
+ }
382
+ },
383
+ "gpu_metrics": {
384
+ "gpu_utilization_mean": 96.62790697674419,
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": "2025-10-13T16:49:30.449008",
398
+ "commit_id": "b84c0b31c674436b076bba818885226a0e0ddecc",
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
+ 3.57020068359375,
432
+ 3.595344970703125,
433
+ 3.57142431640625,
434
+ 3.559490478515625,
435
+ 3.570995361328125
436
+ ],
437
+ "mean": 3.573491162109375,
438
+ "median": 3.570995361328125,
439
+ "std": 0.011789285073375361,
440
+ "min": 3.559490478515625,
441
+ "max": 3.595344970703125,
442
+ "p25": 3.57020068359375,
443
+ "p75": 3.57142431640625,
444
+ "p90": 3.585776708984375,
445
+ "p95": 3.59056083984375,
446
+ "p99": 3.59438814453125,
447
+ "unit": "seconds"
448
+ },
449
+ "time_to_first_token_seconds": {
450
+ "name": "time_to_first_token_seconds",
451
+ "measurements": [
452
+ 0.04043539047241211,
453
+ 0.040699134826660155,
454
+ 0.03881363296508789,
455
+ 0.03881804656982422,
456
+ 0.03845420837402344
457
+ ],
458
+ "mean": 0.03944408264160156,
459
+ "median": 0.03881804656982422,
460
+ "std": 0.0009302780472571182,
461
+ "min": 0.03845420837402344,
462
+ "max": 0.040699134826660155,
463
+ "p25": 0.03881363296508789,
464
+ "p75": 0.04043539047241211,
465
+ "p90": 0.040593637084960936,
466
+ "p95": 0.04064638595581055,
467
+ "p99": 0.040688585052490234,
468
+ "unit": "seconds"
469
+ },
470
+ "tokens_per_second": {
471
+ "name": "tokens_per_second",
472
+ "measurements": [
473
+ 28.009629951485078,
474
+ 27.813742718669765,
475
+ 28.000033359414743,
476
+ 28.093908553367417,
477
+ 28.00339677921284
478
+ ],
479
+ "mean": 27.984142272429967,
480
+ "median": 28.00339677921284,
481
+ "std": 0.09204078314871166,
482
+ "min": 27.813742718669765,
483
+ "max": 28.093908553367417,
484
+ "p25": 28.000033359414743,
485
+ "p75": 28.009629951485078,
486
+ "p90": 28.060197112614482,
487
+ "p95": 28.077052832990947,
488
+ "p99": 28.09053740929212,
489
+ "unit": "tokens/sec"
490
+ },
491
+ "time_per_output_token_seconds": {
492
+ "name": "time_per_output_token_seconds",
493
+ "measurements": [
494
+ 0.0357020068359375,
495
+ 0.03595344970703125,
496
+ 0.0357142431640625,
497
+ 0.03559490478515625,
498
+ 0.03570995361328125
499
+ ],
500
+ "mean": 0.035734911621093754,
501
+ "median": 0.03570995361328125,
502
+ "std": 0.00011789285073375525,
503
+ "min": 0.03559490478515625,
504
+ "max": 0.03595344970703125,
505
+ "p25": 0.0357020068359375,
506
+ "p75": 0.0357142431640625,
507
+ "p90": 0.03585776708984375,
508
+ "p95": 0.035905608398437504,
509
+ "p99": 0.0359438814453125,
510
+ "unit": "seconds/token"
511
+ }
512
+ },
513
+ "gpu_metrics": {
514
+ "gpu_utilization_mean": 95.13333333333334,
515
+ "gpu_utilization_max": 97,
516
+ "gpu_utilization_min": 89,
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": "2025-10-13T16:50:04.211382",
528
+ "commit_id": "b84c0b31c674436b076bba818885226a0e0ddecc",
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
+ 3.41370263671875,
562
+ 3.414087158203125,
563
+ 3.411837158203125,
564
+ 3.41313330078125,
565
+ 3.411944580078125
566
+ ],
567
+ "mean": 3.4129409667968753,
568
+ "median": 3.41313330078125,
569
+ "std": 0.0009101722997469729,
570
+ "min": 3.411837158203125,
571
+ "max": 3.414087158203125,
572
+ "p25": 3.411944580078125,
573
+ "p75": 3.41370263671875,
574
+ "p90": 3.413933349609375,
575
+ "p95": 3.41401025390625,
576
+ "p99": 3.41407177734375,
577
+ "unit": "seconds"
578
+ },
579
+ "time_to_first_token_seconds": {
580
+ "name": "time_to_first_token_seconds",
581
+ "measurements": [
582
+ 0.035256736755371096,
583
+ 0.03462873458862305,
584
+ 0.03471011352539063,
585
+ 0.03469382476806641,
586
+ 0.03469033432006836
587
+ ],
588
+ "mean": 0.034795948791503904,
589
+ "median": 0.03469382476806641,
590
+ "std": 0.00023205066814743804,
591
+ "min": 0.03462873458862305,
592
+ "max": 0.035256736755371096,
593
+ "p25": 0.03469033432006836,
594
+ "p75": 0.03471011352539063,
595
+ "p90": 0.03503808746337891,
596
+ "p95": 0.035147412109375005,
597
+ "p99": 0.035234871826171875,
598
+ "unit": "seconds"
599
+ },
600
+ "tokens_per_second": {
601
+ "name": "tokens_per_second",
602
+ "measurements": [
603
+ 29.29370558652993,
604
+ 29.29040629783781,
605
+ 29.30972240558688,
606
+ 29.298591993787785,
607
+ 29.308799616466878
608
+ ],
609
+ "mean": 29.300245180041856,
610
+ "median": 29.298591993787785,
611
+ "std": 0.007814078291109585,
612
+ "min": 29.29040629783781,
613
+ "max": 29.30972240558688,
614
+ "p25": 29.29370558652993,
615
+ "p75": 29.308799616466878,
616
+ "p90": 29.30935328993888,
617
+ "p95": 29.30953784776288,
618
+ "p99": 29.30968549402208,
619
+ "unit": "tokens/sec"
620
+ },
621
+ "time_per_output_token_seconds": {
622
+ "name": "time_per_output_token_seconds",
623
+ "measurements": [
624
+ 0.0341370263671875,
625
+ 0.03414087158203125,
626
+ 0.03411837158203125,
627
+ 0.0341313330078125,
628
+ 0.03411944580078125
629
+ ],
630
+ "mean": 0.03412940966796875,
631
+ "median": 0.0341313330078125,
632
+ "std": 9.101722997469443e-06,
633
+ "min": 0.03411837158203125,
634
+ "max": 0.03414087158203125,
635
+ "p25": 0.03411944580078125,
636
+ "p75": 0.0341370263671875,
637
+ "p90": 0.03413933349609375,
638
+ "p95": 0.0341401025390625,
639
+ "p99": 0.0341407177734375,
640
+ "unit": "seconds/token"
641
+ }
642
+ },
643
+ "gpu_metrics": {
644
+ "gpu_utilization_mean": 96.75581395348837,
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": "2025-10-13T16:50:36.734924",
658
+ "commit_id": "b84c0b31c674436b076bba818885226a0e0ddecc",
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
+ 3.419146240234375,
692
+ 3.410887939453125,
693
+ 3.41175634765625,
694
+ 3.41238916015625,
695
+ 3.4109189453125
696
+ ],
697
+ "mean": 3.4130197265625,
698
+ "median": 3.41175634765625,
699
+ "std": 0.0031140233835079117,
700
+ "min": 3.410887939453125,
701
+ "max": 3.419146240234375,
702
+ "p25": 3.4109189453125,
703
+ "p75": 3.41238916015625,
704
+ "p90": 3.416443408203125,
705
+ "p95": 3.41779482421875,
706
+ "p99": 3.41887595703125,
707
+ "unit": "seconds"
708
+ },
709
+ "time_to_first_token_seconds": {
710
+ "name": "time_to_first_token_seconds",
711
+ "measurements": [
712
+ 0.03597644805908203,
713
+ 0.03479667282104492,
714
+ 0.03471139144897461,
715
+ 0.034791038513183596,
716
+ 0.03475120162963867
717
+ ],
718
+ "mean": 0.03500535049438477,
719
+ "median": 0.034791038513183596,
720
+ "std": 0.00048652030638567247,
721
+ "min": 0.03471139144897461,
722
+ "max": 0.03597644805908203,
723
+ "p25": 0.03475120162963867,
724
+ "p75": 0.03479667282104492,
725
+ "p90": 0.03550453796386719,
726
+ "p95": 0.03574049301147461,
727
+ "p99": 0.03592925704956055,
728
+ "unit": "seconds"
729
+ },
730
+ "tokens_per_second": {
731
+ "name": "tokens_per_second",
732
+ "measurements": [
733
+ 29.247067242477822,
734
+ 29.31787903182572,
735
+ 29.310416632974476,
736
+ 29.304981145650192,
737
+ 29.31761252709517
738
+ ],
739
+ "mean": 29.299591316004676,
740
+ "median": 29.310416632974476,
741
+ "std": 0.02669914099771034,
742
+ "min": 29.247067242477822,
743
+ "max": 29.31787903182572,
744
+ "p25": 29.304981145650192,
745
+ "p75": 29.31761252709517,
746
+ "p90": 29.3177724299335,
747
+ "p95": 29.31782573087961,
748
+ "p99": 29.317868371636497,
749
+ "unit": "tokens/sec"
750
+ },
751
+ "time_per_output_token_seconds": {
752
+ "name": "time_per_output_token_seconds",
753
+ "measurements": [
754
+ 0.03419146240234375,
755
+ 0.03410887939453125,
756
+ 0.0341175634765625,
757
+ 0.034123891601562505,
758
+ 0.034109189453125
759
+ ],
760
+ "mean": 0.034130197265625,
761
+ "median": 0.0341175634765625,
762
+ "std": 3.1140233835077254e-05,
763
+ "min": 0.03410887939453125,
764
+ "max": 0.03419146240234375,
765
+ "p25": 0.034109189453125,
766
+ "p75": 0.034123891601562505,
767
+ "p90": 0.03416443408203125,
768
+ "p95": 0.0341779482421875,
769
+ "p99": 0.0341887595703125,
770
+ "unit": "seconds/token"
771
+ }
772
+ },
773
+ "gpu_metrics": {
774
+ "gpu_utilization_mean": 96.73255813953489,
775
+ "gpu_utilization_max": 98,
776
+ "gpu_utilization_min": 93,
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": "2025-10-13T16:51:09.209405",
788
+ "commit_id": "b84c0b31c674436b076bba818885226a0e0ddecc",
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
+ 3.699820556640625,
822
+ 3.69959375,
823
+ 3.7055244140625,
824
+ 3.70462646484375,
825
+ 3.7092783203125
826
+ ],
827
+ "mean": 3.7037687011718754,
828
+ "median": 3.70462646484375,
829
+ "std": 0.003665849202381753,
830
+ "min": 3.69959375,
831
+ "max": 3.7092783203125,
832
+ "p25": 3.699820556640625,
833
+ "p75": 3.7055244140625,
834
+ "p90": 3.7077767578125,
835
+ "p95": 3.7085275390625,
836
+ "p99": 3.7091281640625,
837
+ "unit": "seconds"
838
+ },
839
+ "time_to_first_token_seconds": {
840
+ "name": "time_to_first_token_seconds",
841
+ "measurements": [
842
+ 0.03870800018310547,
843
+ 0.03751974487304687,
844
+ 0.03748543930053711,
845
+ 0.037633312225341796,
846
+ 0.03745497512817383
847
+ ],
848
+ "mean": 0.03776029434204102,
849
+ "median": 0.03751974487304687,
850
+ "std": 0.00047768189883625487,
851
+ "min": 0.03745497512817383,
852
+ "max": 0.03870800018310547,
853
+ "p25": 0.03748543930053711,
854
+ "p75": 0.037633312225341796,
855
+ "p90": 0.038278124999999996,
856
+ "p95": 0.03849306259155273,
857
+ "p99": 0.03866501266479492,
858
+ "unit": "seconds"
859
+ },
860
+ "tokens_per_second": {
861
+ "name": "tokens_per_second",
862
+ "measurements": [
863
+ 27.02833785290342,
864
+ 27.02999484740723,
865
+ 26.98673354316573,
866
+ 26.993274746855672,
867
+ 26.959422120574438
868
+ ],
869
+ "mean": 26.9995526221813,
870
+ "median": 26.993274746855672,
871
+ "std": 0.026718967978055952,
872
+ "min": 26.959422120574438,
873
+ "max": 27.02999484740723,
874
+ "p25": 26.98673354316573,
875
+ "p75": 27.02833785290342,
876
+ "p90": 27.029332049605706,
877
+ "p95": 27.02966344850647,
878
+ "p99": 27.029928567627078,
879
+ "unit": "tokens/sec"
880
+ },
881
+ "time_per_output_token_seconds": {
882
+ "name": "time_per_output_token_seconds",
883
+ "measurements": [
884
+ 0.03699820556640625,
885
+ 0.0369959375,
886
+ 0.037055244140625,
887
+ 0.0370462646484375,
888
+ 0.037092783203125
889
+ ],
890
+ "mean": 0.03703768701171875,
891
+ "median": 0.0370462646484375,
892
+ "std": 3.665849202381863e-05,
893
+ "min": 0.0369959375,
894
+ "max": 0.037092783203125,
895
+ "p25": 0.03699820556640625,
896
+ "p75": 0.037055244140625,
897
+ "p90": 0.037077767578125,
898
+ "p95": 0.037085275390625,
899
+ "p99": 0.037091281640625,
900
+ "unit": "seconds/token"
901
+ }
902
+ },
903
+ "gpu_metrics": {
904
+ "gpu_utilization_mean": 90.15053763440861,
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": "2025-10-13T16:51:44.029228",
918
+ "commit_id": "b84c0b31c674436b076bba818885226a0e0ddecc",
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
+ 3.987075439453125,
952
+ 3.924992431640625,
953
+ 3.915767578125,
954
+ 3.972696044921875,
955
+ 3.966458251953125
956
+ ],
957
+ "mean": 3.9533979492187497,
958
+ "median": 3.966458251953125,
959
+ "std": 0.027928765140268583,
960
+ "min": 3.915767578125,
961
+ "max": 3.987075439453125,
962
+ "p25": 3.924992431640625,
963
+ "p75": 3.972696044921875,
964
+ "p90": 3.981323681640625,
965
+ "p95": 3.984199560546875,
966
+ "p99": 3.986500263671875,
967
+ "unit": "seconds"
968
+ },
969
+ "time_to_first_token_seconds": {
970
+ "name": "time_to_first_token_seconds",
971
+ "measurements": [
972
+ 0.04114191818237305,
973
+ 0.041314655303955075,
974
+ 0.04044793701171875,
975
+ 0.04078921508789062,
976
+ 0.04038492965698242
977
+ ],
978
+ "mean": 0.040815731048583985,
979
+ "median": 0.04078921508789062,
980
+ "std": 0.00036793718148023176,
981
+ "min": 0.04038492965698242,
982
+ "max": 0.041314655303955075,
983
+ "p25": 0.04044793701171875,
984
+ "p75": 0.04114191818237305,
985
+ "p90": 0.04124556045532227,
986
+ "p95": 0.04128010787963867,
987
+ "p99": 0.041307745819091794,
988
+ "unit": "seconds"
989
+ },
990
+ "tokens_per_second": {
991
+ "name": "tokens_per_second",
992
+ "measurements": [
993
+ 25.08104035616547,
994
+ 25.4777561337107,
995
+ 25.537777205837976,
996
+ 25.171822578227616,
997
+ 25.211408679458298
998
+ ],
999
+ "mean": 25.295960990680012,
1000
+ "median": 25.211408679458298,
1001
+ "std": 0.1790390314217203,
1002
+ "min": 25.08104035616547,
1003
+ "max": 25.537777205837976,
1004
+ "p25": 25.171822578227616,
1005
+ "p75": 25.4777561337107,
1006
+ "p90": 25.513768776987064,
1007
+ "p95": 25.52577299141252,
1008
+ "p99": 25.535376362952885,
1009
+ "unit": "tokens/sec"
1010
+ },
1011
+ "time_per_output_token_seconds": {
1012
+ "name": "time_per_output_token_seconds",
1013
+ "measurements": [
1014
+ 0.03987075439453125,
1015
+ 0.039249924316406254,
1016
+ 0.03915767578125,
1017
+ 0.03972696044921875,
1018
+ 0.03966458251953125
1019
+ ],
1020
+ "mean": 0.039533979492187496,
1021
+ "median": 0.03966458251953125,
1022
+ "std": 0.00027928765140268583,
1023
+ "min": 0.03915767578125,
1024
+ "max": 0.03987075439453125,
1025
+ "p25": 0.039249924316406254,
1026
+ "p75": 0.03972696044921875,
1027
+ "p90": 0.03981323681640625,
1028
+ "p95": 0.03984199560546875,
1029
+ "p99": 0.03986500263671875,
1030
+ "unit": "seconds/token"
1031
+ }
1032
+ },
1033
+ "gpu_metrics": {
1034
+ "gpu_utilization_mean": 88.17,
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": 100,
1041
+ "gpu_monitoring_status": "success"
1042
+ }
1043
+ },
1044
+ {
1045
+ "scenario_name": "compiled_compile_max-autotune_sdpa_efficient_attention",
1046
+ "metadata": {
1047
+ "timestamp": "2025-10-13T16:52:25.650183",
1048
+ "commit_id": "b84c0b31c674436b076bba818885226a0e0ddecc",
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
+ 3.709071044921875,
1082
+ 3.69849560546875,
1083
+ 3.699703857421875,
1084
+ 3.711072509765625,
1085
+ 3.70815478515625
1086
+ ],
1087
+ "mean": 3.7052995605468753,
1088
+ "median": 3.70815478515625,
1089
+ "std": 0.005163507389583061,
1090
+ "min": 3.69849560546875,
1091
+ "max": 3.711072509765625,
1092
+ "p25": 3.699703857421875,
1093
+ "p75": 3.709071044921875,
1094
+ "p90": 3.7102719238281248,
1095
+ "p95": 3.710672216796875,
1096
+ "p99": 3.710992451171875,
1097
+ "unit": "seconds"
1098
+ },
1099
+ "time_to_first_token_seconds": {
1100
+ "name": "time_to_first_token_seconds",
1101
+ "measurements": [
1102
+ 0.03813119888305664,
1103
+ 0.03771257781982422,
1104
+ 0.03818524932861328,
1105
+ 0.03750118255615234,
1106
+ 0.03745856094360352
1107
+ ],
1108
+ "mean": 0.037797753906249995,
1109
+ "median": 0.03771257781982422,
1110
+ "std": 0.00030711550016778263,
1111
+ "min": 0.03745856094360352,
1112
+ "max": 0.03818524932861328,
1113
+ "p25": 0.03750118255615234,
1114
+ "p75": 0.03813119888305664,
1115
+ "p90": 0.03816362915039062,
1116
+ "p95": 0.03817443923950195,
1117
+ "p99": 0.038183087310791015,
1118
+ "unit": "seconds"
1119
+ },
1120
+ "tokens_per_second": {
1121
+ "name": "tokens_per_second",
1122
+ "measurements": [
1123
+ 26.96092870394353,
1124
+ 27.038020500047594,
1125
+ 27.029190403819143,
1126
+ 26.946388068907755,
1127
+ 26.967590565609658
1128
+ ],
1129
+ "mean": 26.988423648465535,
1130
+ "median": 26.967590565609658,
1131
+ "std": 0.03762646752811152,
1132
+ "min": 26.946388068907755,
1133
+ "max": 27.038020500047594,
1134
+ "p25": 26.96092870394353,
1135
+ "p75": 27.029190403819143,
1136
+ "p90": 27.034488461556215,
1137
+ "p95": 27.036254480801905,
1138
+ "p99": 27.037667296198457,
1139
+ "unit": "tokens/sec"
1140
+ },
1141
+ "time_per_output_token_seconds": {
1142
+ "name": "time_per_output_token_seconds",
1143
+ "measurements": [
1144
+ 0.03709071044921875,
1145
+ 0.036984956054687504,
1146
+ 0.036997038574218745,
1147
+ 0.03711072509765625,
1148
+ 0.0370815478515625
1149
+ ],
1150
+ "mean": 0.03705299560546875,
1151
+ "median": 0.0370815478515625,
1152
+ "std": 5.163507389583044e-05,
1153
+ "min": 0.036984956054687504,
1154
+ "max": 0.03711072509765625,
1155
+ "p25": 0.036997038574218745,
1156
+ "p75": 0.03709071044921875,
1157
+ "p90": 0.03710271923828125,
1158
+ "p95": 0.03710672216796875,
1159
+ "p99": 0.037109924511718746,
1160
+ "unit": "seconds/token"
1161
+ }
1162
+ },
1163
+ "gpu_metrics": {
1164
+ "gpu_utilization_mean": 90.13978494623656,
1165
+ "gpu_utilization_max": 93,
1166
+ "gpu_utilization_min": 86,
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
+ }