hf-transformers-bot commited on
Commit
ea26069
·
verified ·
1 Parent(s): ff485c6

Upload benchmark results for run 18104060704

Browse files
2025-09-29/18104060704/benchmark_results/Llama-2-7b-hf/Llama-2-7b-hf_benchmark_20250929_164813.json ADDED
@@ -0,0 +1,1175 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_name": "Llama-2-7b-hf",
3
+ "benchmark_scenarios": [
4
+ {
5
+ "scenario_name": "eager_eager_attn",
6
+ "metadata": {
7
+ "timestamp": "2025-09-29T16:42:12.003976",
8
+ "commit_id": "8635d8e796c77ee94f64c170168f71138380274b",
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
+ 3.661336669921875,
42
+ 3.648054931640625,
43
+ 3.654614013671875,
44
+ 3.64878955078125,
45
+ 3.653759521484375
46
+ ],
47
+ "mean": 3.6533109375,
48
+ "median": 3.653759521484375,
49
+ "std": 0.004782889324718939,
50
+ "min": 3.648054931640625,
51
+ "max": 3.661336669921875,
52
+ "p25": 3.64878955078125,
53
+ "p75": 3.654614013671875,
54
+ "p90": 3.658647607421875,
55
+ "p95": 3.659992138671875,
56
+ "p99": 3.6610677636718747,
57
+ "unit": "seconds"
58
+ },
59
+ "time_to_first_token_seconds": {
60
+ "name": "time_to_first_token_seconds",
61
+ "measurements": [
62
+ 0.037605022430419924,
63
+ 0.03722675323486328,
64
+ 0.036994495391845704,
65
+ 0.03845404815673828,
66
+ 0.03704207992553711
67
+ ],
68
+ "mean": 0.037464479827880856,
69
+ "median": 0.03722675323486328,
70
+ "std": 0.00053941054517856,
71
+ "min": 0.036994495391845704,
72
+ "max": 0.03845404815673828,
73
+ "p25": 0.03704207992553711,
74
+ "p75": 0.037605022430419924,
75
+ "p90": 0.03811443786621094,
76
+ "p95": 0.03828424301147461,
77
+ "p99": 0.03842008712768555,
78
+ "unit": "seconds"
79
+ },
80
+ "tokens_per_second": {
81
+ "name": "tokens_per_second",
82
+ "measurements": [
83
+ 27.312429589310014,
84
+ 27.41186793342155,
85
+ 27.36267075699403,
86
+ 27.406349039392747,
87
+ 27.369069970804766
88
+ ],
89
+ "mean": 27.372477457984623,
90
+ "median": 27.369069970804766,
91
+ "std": 0.03581179266051129,
92
+ "min": 27.312429589310014,
93
+ "max": 27.41186793342155,
94
+ "p25": 27.36267075699403,
95
+ "p75": 27.406349039392747,
96
+ "p90": 27.40966037581003,
97
+ "p95": 27.41076415461579,
98
+ "p99": 27.4116471776604,
99
+ "unit": "tokens/sec"
100
+ },
101
+ "time_per_output_token_seconds": {
102
+ "name": "time_per_output_token_seconds",
103
+ "measurements": [
104
+ 0.03661336669921875,
105
+ 0.03648054931640625,
106
+ 0.03654614013671875,
107
+ 0.0364878955078125,
108
+ 0.03653759521484375
109
+ ],
110
+ "mean": 0.036533109375,
111
+ "median": 0.03653759521484375,
112
+ "std": 4.782889324718928e-05,
113
+ "min": 0.03648054931640625,
114
+ "max": 0.03661336669921875,
115
+ "p25": 0.0364878955078125,
116
+ "p75": 0.03654614013671875,
117
+ "p90": 0.03658647607421875,
118
+ "p95": 0.03659992138671875,
119
+ "p99": 0.03661067763671875,
120
+ "unit": "seconds/token"
121
+ }
122
+ },
123
+ "gpu_metrics": {
124
+ "gpu_utilization_mean": 90.8913043478261,
125
+ "gpu_utilization_max": 93,
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": "2025-09-29T16:42:47.355972",
138
+ "commit_id": "8635d8e796c77ee94f64c170168f71138380274b",
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
+ 7.8291669921875,
172
+ 7.8788349609375,
173
+ 7.8329775390625,
174
+ 7.8891591796875,
175
+ 7.920875
176
+ ],
177
+ "mean": 7.870202734375001,
178
+ "median": 7.8788349609375,
179
+ "std": 0.034845912129876526,
180
+ "min": 7.8291669921875,
181
+ "max": 7.920875,
182
+ "p25": 7.8329775390625,
183
+ "p75": 7.8891591796875,
184
+ "p90": 7.908188671875,
185
+ "p95": 7.9145318359375,
186
+ "p99": 7.9196063671875,
187
+ "unit": "seconds"
188
+ },
189
+ "time_to_first_token_seconds": {
190
+ "name": "time_to_first_token_seconds",
191
+ "measurements": [
192
+ 0.03889273452758789,
193
+ 0.03868675231933594,
194
+ 0.037978111267089845,
195
+ 0.03811388778686523,
196
+ 0.038766239166259764
197
+ ],
198
+ "mean": 0.03848754501342773,
199
+ "median": 0.03868675231933594,
200
+ "std": 0.0003689647649317064,
201
+ "min": 0.037978111267089845,
202
+ "max": 0.03889273452758789,
203
+ "p25": 0.03811388778686523,
204
+ "p75": 0.038766239166259764,
205
+ "p90": 0.03884213638305664,
206
+ "p95": 0.038867435455322265,
207
+ "p99": 0.038887674713134764,
208
+ "unit": "seconds"
209
+ },
210
+ "tokens_per_second": {
211
+ "name": "tokens_per_second",
212
+ "measurements": [
213
+ 12.772750932479422,
214
+ 12.69223184592523,
215
+ 12.766537309893605,
216
+ 12.675622043154304,
217
+ 12.62486783341487
218
+ ],
219
+ "mean": 12.706401992973486,
220
+ "median": 12.69223184592523,
221
+ "std": 0.056239586890047544,
222
+ "min": 12.62486783341487,
223
+ "max": 12.772750932479422,
224
+ "p25": 12.675622043154304,
225
+ "p75": 12.766537309893605,
226
+ "p90": 12.770265483445096,
227
+ "p95": 12.771508207962258,
228
+ "p99": 12.772502387575988,
229
+ "unit": "tokens/sec"
230
+ },
231
+ "time_per_output_token_seconds": {
232
+ "name": "time_per_output_token_seconds",
233
+ "measurements": [
234
+ 0.078291669921875,
235
+ 0.078788349609375,
236
+ 0.078329775390625,
237
+ 0.078891591796875,
238
+ 0.07920875
239
+ ],
240
+ "mean": 0.07870202734375,
241
+ "median": 0.078788349609375,
242
+ "std": 0.0003484591212987653,
243
+ "min": 0.078291669921875,
244
+ "max": 0.07920875,
245
+ "p25": 0.078329775390625,
246
+ "p75": 0.078891591796875,
247
+ "p90": 0.07908188671875,
248
+ "p95": 0.079145318359375,
249
+ "p99": 0.079196063671875,
250
+ "unit": "seconds/token"
251
+ }
252
+ },
253
+ "gpu_metrics": {
254
+ "gpu_utilization_mean": 44.57058823529412,
255
+ "gpu_utilization_max": 94,
256
+ "gpu_utilization_min": 0,
257
+ "gpu_memory_used_mean": 13629.470588235294,
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": "2025-09-29T16:44:14.963757",
268
+ "commit_id": "8635d8e796c77ee94f64c170168f71138380274b",
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
+ 3.408482666015625,
302
+ 3.41104443359375,
303
+ 3.410935791015625,
304
+ 3.409746826171875,
305
+ 3.40982861328125
306
+ ],
307
+ "mean": 3.4100076660156247,
308
+ "median": 3.40982861328125,
309
+ "std": 0.000934026266488331,
310
+ "min": 3.408482666015625,
311
+ "max": 3.41104443359375,
312
+ "p25": 3.409746826171875,
313
+ "p75": 3.410935791015625,
314
+ "p90": 3.4110009765625002,
315
+ "p95": 3.411022705078125,
316
+ "p99": 3.4110400878906253,
317
+ "unit": "seconds"
318
+ },
319
+ "time_to_first_token_seconds": {
320
+ "name": "time_to_first_token_seconds",
321
+ "measurements": [
322
+ 0.035171585083007814,
323
+ 0.03461676788330078,
324
+ 0.03459721755981445,
325
+ 0.034598655700683593,
326
+ 0.034590625762939455
327
+ ],
328
+ "mean": 0.03471497039794922,
329
+ "median": 0.034598655700683593,
330
+ "std": 0.00022847193807508856,
331
+ "min": 0.034590625762939455,
332
+ "max": 0.035171585083007814,
333
+ "p25": 0.03459721755981445,
334
+ "p75": 0.03461676788330078,
335
+ "p90": 0.034949658203125,
336
+ "p95": 0.035060621643066406,
337
+ "p99": 0.03514939239501953,
338
+ "unit": "seconds"
339
+ },
340
+ "tokens_per_second": {
341
+ "name": "tokens_per_second",
342
+ "measurements": [
343
+ 29.33856786101713,
344
+ 29.31653396688348,
345
+ 29.317467735217747,
346
+ 29.327690616921863,
347
+ 29.326987171877484
348
+ ],
349
+ "mean": 29.32544947038354,
350
+ "median": 29.326987171877484,
351
+ "std": 0.008033352169830449,
352
+ "min": 29.31653396688348,
353
+ "max": 29.33856786101713,
354
+ "p25": 29.317467735217747,
355
+ "p75": 29.327690616921863,
356
+ "p90": 29.334216963379024,
357
+ "p95": 29.336392412198077,
358
+ "p99": 29.338132771253317,
359
+ "unit": "tokens/sec"
360
+ },
361
+ "time_per_output_token_seconds": {
362
+ "name": "time_per_output_token_seconds",
363
+ "measurements": [
364
+ 0.03408482666015625,
365
+ 0.034110444335937504,
366
+ 0.03410935791015625,
367
+ 0.03409746826171875,
368
+ 0.034098286132812496
369
+ ],
370
+ "mean": 0.03410007666015625,
371
+ "median": 0.034098286132812496,
372
+ "std": 9.340262664885055e-06,
373
+ "min": 0.03408482666015625,
374
+ "max": 0.034110444335937504,
375
+ "p25": 0.03409746826171875,
376
+ "p75": 0.03410935791015625,
377
+ "p90": 0.034110009765625005,
378
+ "p95": 0.034110227050781254,
379
+ "p99": 0.03411040087890625,
380
+ "unit": "seconds/token"
381
+ }
382
+ },
383
+ "gpu_metrics": {
384
+ "gpu_utilization_mean": 96.8953488372093,
385
+ "gpu_utilization_max": 98,
386
+ "gpu_utilization_min": 95,
387
+ "gpu_memory_used_mean": 13849,
388
+ "gpu_memory_used_max": 13849,
389
+ "gpu_memory_used_min": 13849,
390
+ "sample_count": 86,
391
+ "gpu_monitoring_status": "success"
392
+ }
393
+ },
394
+ {
395
+ "scenario_name": "eager_sdpa_math",
396
+ "metadata": {
397
+ "timestamp": "2025-09-29T16:44:47.454880",
398
+ "commit_id": "8635d8e796c77ee94f64c170168f71138380274b",
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
+ 3.53882763671875,
432
+ 3.55037255859375,
433
+ 3.5365849609375,
434
+ 3.537248291015625,
435
+ 3.544261474609375
436
+ ],
437
+ "mean": 3.541458984375,
438
+ "median": 3.53882763671875,
439
+ "std": 0.0052099096036488605,
440
+ "min": 3.5365849609375,
441
+ "max": 3.55037255859375,
442
+ "p25": 3.537248291015625,
443
+ "p75": 3.544261474609375,
444
+ "p90": 3.547928125,
445
+ "p95": 3.549150341796875,
446
+ "p99": 3.550128115234375,
447
+ "unit": "seconds"
448
+ },
449
+ "time_to_first_token_seconds": {
450
+ "name": "time_to_first_token_seconds",
451
+ "measurements": [
452
+ 0.040119007110595704,
453
+ 0.0398131217956543,
454
+ 0.03785382461547852,
455
+ 0.03930243301391602,
456
+ 0.03807964706420899
457
+ ],
458
+ "mean": 0.039033606719970707,
459
+ "median": 0.03930243301391602,
460
+ "std": 0.0009121326851291216,
461
+ "min": 0.03785382461547852,
462
+ "max": 0.040119007110595704,
463
+ "p25": 0.03807964706420899,
464
+ "p75": 0.0398131217956543,
465
+ "p90": 0.039996652984619144,
466
+ "p95": 0.04005783004760742,
467
+ "p99": 0.04010677169799805,
468
+ "unit": "seconds"
469
+ },
470
+ "tokens_per_second": {
471
+ "name": "tokens_per_second",
472
+ "measurements": [
473
+ 28.257945926047245,
474
+ 28.166058167036002,
475
+ 28.275865306369276,
476
+ 28.270562814036364,
477
+ 28.21462262770027
478
+ ],
479
+ "mean": 28.237010968237833,
480
+ "median": 28.257945926047245,
481
+ "std": 0.0414950507820275,
482
+ "min": 28.166058167036002,
483
+ "max": 28.275865306369276,
484
+ "p25": 28.21462262770027,
485
+ "p75": 28.270562814036364,
486
+ "p90": 28.27374430943611,
487
+ "p95": 28.274804807902694,
488
+ "p99": 28.27565320667596,
489
+ "unit": "tokens/sec"
490
+ },
491
+ "time_per_output_token_seconds": {
492
+ "name": "time_per_output_token_seconds",
493
+ "measurements": [
494
+ 0.0353882763671875,
495
+ 0.0355037255859375,
496
+ 0.035365849609375,
497
+ 0.035372482910156246,
498
+ 0.035442614746093754
499
+ ],
500
+ "mean": 0.03541458984375,
501
+ "median": 0.0353882763671875,
502
+ "std": 5.209909603649076e-05,
503
+ "min": 0.035365849609375,
504
+ "max": 0.0355037255859375,
505
+ "p25": 0.035372482910156246,
506
+ "p75": 0.035442614746093754,
507
+ "p90": 0.03547928125,
508
+ "p95": 0.03549150341796875,
509
+ "p99": 0.03550128115234375,
510
+ "unit": "seconds/token"
511
+ }
512
+ },
513
+ "gpu_metrics": {
514
+ "gpu_utilization_mean": 96.12359550561797,
515
+ "gpu_utilization_max": 98,
516
+ "gpu_utilization_min": 90,
517
+ "gpu_memory_used_mean": 13849,
518
+ "gpu_memory_used_max": 13849,
519
+ "gpu_memory_used_min": 13849,
520
+ "sample_count": 89,
521
+ "gpu_monitoring_status": "success"
522
+ }
523
+ },
524
+ {
525
+ "scenario_name": "eager_sdpa_flash_attention",
526
+ "metadata": {
527
+ "timestamp": "2025-09-29T16:45:21.024613",
528
+ "commit_id": "8635d8e796c77ee94f64c170168f71138380274b",
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
+ 3.411535400390625,
562
+ 3.40937451171875,
563
+ 3.408629638671875,
564
+ 3.41137646484375,
565
+ 3.4079775390625
566
+ ],
567
+ "mean": 3.4097787109374997,
568
+ "median": 3.40937451171875,
569
+ "std": 0.0014399124683824101,
570
+ "min": 3.4079775390625,
571
+ "max": 3.411535400390625,
572
+ "p25": 3.408629638671875,
573
+ "p75": 3.41137646484375,
574
+ "p90": 3.411471826171875,
575
+ "p95": 3.41150361328125,
576
+ "p99": 3.4115290429687497,
577
+ "unit": "seconds"
578
+ },
579
+ "time_to_first_token_seconds": {
580
+ "name": "time_to_first_token_seconds",
581
+ "measurements": [
582
+ 0.03525049591064453,
583
+ 0.03467919921875,
584
+ 0.03463555145263672,
585
+ 0.0346530876159668,
586
+ 0.034616798400878904
587
+ ],
588
+ "mean": 0.03476702651977539,
589
+ "median": 0.0346530876159668,
590
+ "std": 0.00024260772514500932,
591
+ "min": 0.034616798400878904,
592
+ "max": 0.03525049591064453,
593
+ "p25": 0.03463555145263672,
594
+ "p75": 0.03467919921875,
595
+ "p90": 0.03502197723388672,
596
+ "p95": 0.035136236572265625,
597
+ "p99": 0.03522764404296875,
598
+ "unit": "seconds"
599
+ },
600
+ "tokens_per_second": {
601
+ "name": "tokens_per_second",
602
+ "measurements": [
603
+ 29.312314915023272,
604
+ 29.330893293264964,
605
+ 29.337302846126637,
606
+ 29.313680571628222,
607
+ 29.34291639360657
608
+ ],
609
+ "mean": 29.327421603929935,
610
+ "median": 29.330893293264964,
611
+ "std": 0.012383925869863354,
612
+ "min": 29.312314915023272,
613
+ "max": 29.34291639360657,
614
+ "p25": 29.313680571628222,
615
+ "p75": 29.337302846126637,
616
+ "p90": 29.340670974614596,
617
+ "p95": 29.34179368411058,
618
+ "p99": 29.34269185170737,
619
+ "unit": "tokens/sec"
620
+ },
621
+ "time_per_output_token_seconds": {
622
+ "name": "time_per_output_token_seconds",
623
+ "measurements": [
624
+ 0.03411535400390625,
625
+ 0.034093745117187504,
626
+ 0.03408629638671875,
627
+ 0.034113764648437496,
628
+ 0.034079775390625
629
+ ],
630
+ "mean": 0.034097787109375,
631
+ "median": 0.034093745117187504,
632
+ "std": 1.4399124683823815e-05,
633
+ "min": 0.034079775390625,
634
+ "max": 0.03411535400390625,
635
+ "p25": 0.03408629638671875,
636
+ "p75": 0.034113764648437496,
637
+ "p90": 0.034114718261718746,
638
+ "p95": 0.0341150361328125,
639
+ "p99": 0.0341152904296875,
640
+ "unit": "seconds/token"
641
+ }
642
+ },
643
+ "gpu_metrics": {
644
+ "gpu_utilization_mean": 96.95348837209302,
645
+ "gpu_utilization_max": 98,
646
+ "gpu_utilization_min": 94,
647
+ "gpu_memory_used_mean": 13849,
648
+ "gpu_memory_used_max": 13849,
649
+ "gpu_memory_used_min": 13849,
650
+ "sample_count": 86,
651
+ "gpu_monitoring_status": "success"
652
+ }
653
+ },
654
+ {
655
+ "scenario_name": "eager_sdpa_efficient_attention",
656
+ "metadata": {
657
+ "timestamp": "2025-09-29T16:45:53.543085",
658
+ "commit_id": "8635d8e796c77ee94f64c170168f71138380274b",
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
+ 3.408671875,
692
+ 3.407046630859375,
693
+ 3.410819091796875,
694
+ 3.412315673828125,
695
+ 3.40947900390625
696
+ ],
697
+ "mean": 3.409666455078125,
698
+ "median": 3.40947900390625,
699
+ "std": 0.0018019288444081195,
700
+ "min": 3.407046630859375,
701
+ "max": 3.412315673828125,
702
+ "p25": 3.408671875,
703
+ "p75": 3.410819091796875,
704
+ "p90": 3.411717041015625,
705
+ "p95": 3.4120163574218747,
706
+ "p99": 3.412255810546875,
707
+ "unit": "seconds"
708
+ },
709
+ "time_to_first_token_seconds": {
710
+ "name": "time_to_first_token_seconds",
711
+ "measurements": [
712
+ 0.03530451202392578,
713
+ 0.03477344131469726,
714
+ 0.03475820922851563,
715
+ 0.03476643371582031,
716
+ 0.03474630355834961
717
+ ],
718
+ "mean": 0.03486977996826172,
719
+ "median": 0.03476643371582031,
720
+ "std": 0.0002175536706463349,
721
+ "min": 0.03474630355834961,
722
+ "max": 0.03530451202392578,
723
+ "p25": 0.03475820922851563,
724
+ "p75": 0.03477344131469726,
725
+ "p90": 0.03509208374023438,
726
+ "p95": 0.03519829788208008,
727
+ "p99": 0.03528326919555664,
728
+ "unit": "seconds"
729
+ },
730
+ "tokens_per_second": {
731
+ "name": "tokens_per_second",
732
+ "measurements": [
733
+ 29.33693933212624,
734
+ 29.35093376599208,
735
+ 29.318470815559547,
736
+ 29.30561224654062,
737
+ 29.329994373166606
738
+ ],
739
+ "mean": 29.328390106677016,
740
+ "median": 29.329994373166606,
741
+ "std": 0.015499049975604157,
742
+ "min": 29.30561224654062,
743
+ "max": 29.35093376599208,
744
+ "p25": 29.318470815559547,
745
+ "p75": 29.33693933212624,
746
+ "p90": 29.345335992445744,
747
+ "p95": 29.348134879218915,
748
+ "p99": 29.350373988637447,
749
+ "unit": "tokens/sec"
750
+ },
751
+ "time_per_output_token_seconds": {
752
+ "name": "time_per_output_token_seconds",
753
+ "measurements": [
754
+ 0.03408671875,
755
+ 0.034070466308593755,
756
+ 0.03410819091796875,
757
+ 0.03412315673828125,
758
+ 0.0340947900390625
759
+ ],
760
+ "mean": 0.03409666455078125,
761
+ "median": 0.0340947900390625,
762
+ "std": 1.8019288444080966e-05,
763
+ "min": 0.034070466308593755,
764
+ "max": 0.03412315673828125,
765
+ "p25": 0.03408671875,
766
+ "p75": 0.03410819091796875,
767
+ "p90": 0.03411717041015625,
768
+ "p95": 0.03412016357421875,
769
+ "p99": 0.03412255810546875,
770
+ "unit": "seconds/token"
771
+ }
772
+ },
773
+ "gpu_metrics": {
774
+ "gpu_utilization_mean": 96.8953488372093,
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": "2025-09-29T16:46:26.212438",
788
+ "commit_id": "8635d8e796c77ee94f64c170168f71138380274b",
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
+ 3.69020556640625,
822
+ 3.697838623046875,
823
+ 3.6951767578125,
824
+ 3.696549072265625,
825
+ 3.697183349609375
826
+ ],
827
+ "mean": 3.6953906738281246,
828
+ "median": 3.696549072265625,
829
+ "std": 0.0027378505369134074,
830
+ "min": 3.69020556640625,
831
+ "max": 3.697838623046875,
832
+ "p25": 3.6951767578125,
833
+ "p75": 3.697183349609375,
834
+ "p90": 3.697576513671875,
835
+ "p95": 3.697707568359375,
836
+ "p99": 3.697812412109375,
837
+ "unit": "seconds"
838
+ },
839
+ "time_to_first_token_seconds": {
840
+ "name": "time_to_first_token_seconds",
841
+ "measurements": [
842
+ 0.03806732940673828,
843
+ 0.037494144439697265,
844
+ 0.03746252822875976,
845
+ 0.03753551864624023,
846
+ 0.037426910400390624
847
+ ],
848
+ "mean": 0.03759728622436523,
849
+ "median": 0.037494144439697265,
850
+ "std": 0.00023773165271874627,
851
+ "min": 0.037426910400390624,
852
+ "max": 0.03806732940673828,
853
+ "p25": 0.03746252822875976,
854
+ "p75": 0.03753551864624023,
855
+ "p90": 0.03785460510253906,
856
+ "p95": 0.03796096725463867,
857
+ "p99": 0.038046056976318354,
858
+ "unit": "seconds"
859
+ },
860
+ "tokens_per_second": {
861
+ "name": "tokens_per_second",
862
+ "measurements": [
863
+ 27.098761356372396,
864
+ 27.042824253267142,
865
+ 27.062304878535443,
866
+ 27.052258213012095,
867
+ 27.047617211238787
868
+ ],
869
+ "mean": 27.060753182485172,
870
+ "median": 27.052258213012095,
871
+ "std": 0.0200658605233535,
872
+ "min": 27.042824253267142,
873
+ "max": 27.098761356372396,
874
+ "p25": 27.047617211238787,
875
+ "p75": 27.062304878535443,
876
+ "p90": 27.084178765237613,
877
+ "p95": 27.091470060805005,
878
+ "p99": 27.09730309725892,
879
+ "unit": "tokens/sec"
880
+ },
881
+ "time_per_output_token_seconds": {
882
+ "name": "time_per_output_token_seconds",
883
+ "measurements": [
884
+ 0.0369020556640625,
885
+ 0.03697838623046875,
886
+ 0.036951767578124996,
887
+ 0.03696549072265625,
888
+ 0.03697183349609375
889
+ ],
890
+ "mean": 0.03695390673828125,
891
+ "median": 0.03696549072265625,
892
+ "std": 2.7378505369133722e-05,
893
+ "min": 0.0369020556640625,
894
+ "max": 0.03697838623046875,
895
+ "p25": 0.036951767578124996,
896
+ "p75": 0.03697183349609375,
897
+ "p90": 0.03697576513671875,
898
+ "p95": 0.03697707568359375,
899
+ "p99": 0.03697812412109375,
900
+ "unit": "seconds/token"
901
+ }
902
+ },
903
+ "gpu_metrics": {
904
+ "gpu_utilization_mean": 90.38709677419355,
905
+ "gpu_utilization_max": 94,
906
+ "gpu_utilization_min": 86,
907
+ "gpu_memory_used_mean": 13845,
908
+ "gpu_memory_used_max": 13845,
909
+ "gpu_memory_used_min": 13845,
910
+ "sample_count": 93,
911
+ "gpu_monitoring_status": "success"
912
+ }
913
+ },
914
+ {
915
+ "scenario_name": "compiled_compile_max-autotune_sdpa_math",
916
+ "metadata": {
917
+ "timestamp": "2025-09-29T16:47:00.988692",
918
+ "commit_id": "8635d8e796c77ee94f64c170168f71138380274b",
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
+ 3.88004736328125,
952
+ 3.866556396484375,
953
+ 3.893905517578125,
954
+ 3.91741943359375,
955
+ 3.92266064453125
956
+ ],
957
+ "mean": 3.8961178710937503,
958
+ "median": 3.893905517578125,
959
+ "std": 0.021425741047657123,
960
+ "min": 3.866556396484375,
961
+ "max": 3.92266064453125,
962
+ "p25": 3.88004736328125,
963
+ "p75": 3.91741943359375,
964
+ "p90": 3.92056416015625,
965
+ "p95": 3.92161240234375,
966
+ "p99": 3.92245099609375,
967
+ "unit": "seconds"
968
+ },
969
+ "time_to_first_token_seconds": {
970
+ "name": "time_to_first_token_seconds",
971
+ "measurements": [
972
+ 0.040627552032470704,
973
+ 0.040252479553222656,
974
+ 0.04049971389770508,
975
+ 0.04100873565673828,
976
+ 0.04105068969726562
977
+ ],
978
+ "mean": 0.040687834167480466,
979
+ "median": 0.040627552032470704,
980
+ "std": 0.00030436777528536607,
981
+ "min": 0.040252479553222656,
982
+ "max": 0.04105068969726562,
983
+ "p25": 0.04049971389770508,
984
+ "p75": 0.04100873565673828,
985
+ "p90": 0.041033908081054685,
986
+ "p95": 0.04104229888916015,
987
+ "p99": 0.041049011535644525,
988
+ "unit": "seconds"
989
+ },
990
+ "tokens_per_second": {
991
+ "name": "tokens_per_second",
992
+ "measurements": [
993
+ 25.772881265921644,
994
+ 25.862806524928466,
995
+ 25.681157272197133,
996
+ 25.527008709471357,
997
+ 25.492901135716213
998
+ ],
999
+ "mean": 25.66735098164697,
1000
+ "median": 25.681157272197133,
1001
+ "std": 0.14117990371295233,
1002
+ "min": 25.492901135716213,
1003
+ "max": 25.862806524928466,
1004
+ "p25": 25.527008709471357,
1005
+ "p75": 25.772881265921644,
1006
+ "p90": 25.826836421325737,
1007
+ "p95": 25.8448214731271,
1008
+ "p99": 25.859209514568192,
1009
+ "unit": "tokens/sec"
1010
+ },
1011
+ "time_per_output_token_seconds": {
1012
+ "name": "time_per_output_token_seconds",
1013
+ "measurements": [
1014
+ 0.0388004736328125,
1015
+ 0.038665563964843754,
1016
+ 0.03893905517578125,
1017
+ 0.0391741943359375,
1018
+ 0.0392266064453125
1019
+ ],
1020
+ "mean": 0.0389611787109375,
1021
+ "median": 0.03893905517578125,
1022
+ "std": 0.0002142574104765712,
1023
+ "min": 0.038665563964843754,
1024
+ "max": 0.0392266064453125,
1025
+ "p25": 0.0388004736328125,
1026
+ "p75": 0.0391741943359375,
1027
+ "p90": 0.0392056416015625,
1028
+ "p95": 0.0392161240234375,
1029
+ "p99": 0.0392245099609375,
1030
+ "unit": "seconds/token"
1031
+ }
1032
+ },
1033
+ "gpu_metrics": {
1034
+ "gpu_utilization_mean": 89.16326530612245,
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": 98,
1041
+ "gpu_monitoring_status": "success"
1042
+ }
1043
+ },
1044
+ {
1045
+ "scenario_name": "compiled_compile_max-autotune_sdpa_efficient_attention",
1046
+ "metadata": {
1047
+ "timestamp": "2025-09-29T16:47:42.145591",
1048
+ "commit_id": "8635d8e796c77ee94f64c170168f71138380274b",
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
+ 3.6924560546875,
1082
+ 3.694727294921875,
1083
+ 3.701572021484375,
1084
+ 3.6854599609375,
1085
+ 3.690012451171875
1086
+ ],
1087
+ "mean": 3.692845556640625,
1088
+ "median": 3.6924560546875,
1089
+ "std": 0.005336990722999065,
1090
+ "min": 3.6854599609375,
1091
+ "max": 3.701572021484375,
1092
+ "p25": 3.690012451171875,
1093
+ "p75": 3.694727294921875,
1094
+ "p90": 3.6988341308593746,
1095
+ "p95": 3.7002030761718747,
1096
+ "p99": 3.7012982324218746,
1097
+ "unit": "seconds"
1098
+ },
1099
+ "time_to_first_token_seconds": {
1100
+ "name": "time_to_first_token_seconds",
1101
+ "measurements": [
1102
+ 0.03803593444824219,
1103
+ 0.03828390502929688,
1104
+ 0.03752304077148438,
1105
+ 0.037406047821044924,
1106
+ 0.037754302978515626
1107
+ ],
1108
+ "mean": 0.0378006462097168,
1109
+ "median": 0.037754302978515626,
1110
+ "std": 0.0003236726755990165,
1111
+ "min": 0.037406047821044924,
1112
+ "max": 0.03828390502929688,
1113
+ "p25": 0.03752304077148438,
1114
+ "p75": 0.03803593444824219,
1115
+ "p90": 0.038184716796875004,
1116
+ "p95": 0.03823431091308594,
1117
+ "p99": 0.03827398620605469,
1118
+ "unit": "seconds"
1119
+ },
1120
+ "tokens_per_second": {
1121
+ "name": "tokens_per_second",
1122
+ "measurements": [
1123
+ 27.082245128700173,
1124
+ 27.065597002908035,
1125
+ 27.01554891262086,
1126
+ 27.133655245181988,
1127
+ 27.10017955853834
1128
+ ],
1129
+ "mean": 27.079445169589878,
1130
+ "median": 27.082245128700173,
1131
+ "std": 0.03911774616944684,
1132
+ "min": 27.01554891262086,
1133
+ "max": 27.133655245181988,
1134
+ "p25": 27.065597002908035,
1135
+ "p75": 27.10017955853834,
1136
+ "p90": 27.120264970524527,
1137
+ "p95": 27.126960107853257,
1138
+ "p99": 27.13231621771624,
1139
+ "unit": "tokens/sec"
1140
+ },
1141
+ "time_per_output_token_seconds": {
1142
+ "name": "time_per_output_token_seconds",
1143
+ "measurements": [
1144
+ 0.036924560546875,
1145
+ 0.03694727294921875,
1146
+ 0.03701572021484375,
1147
+ 0.036854599609375004,
1148
+ 0.03690012451171875
1149
+ ],
1150
+ "mean": 0.036928455566406254,
1151
+ "median": 0.036924560546875,
1152
+ "std": 5.33699072299899e-05,
1153
+ "min": 0.036854599609375004,
1154
+ "max": 0.03701572021484375,
1155
+ "p25": 0.03690012451171875,
1156
+ "p75": 0.03694727294921875,
1157
+ "p90": 0.03698834130859375,
1158
+ "p95": 0.03700203076171875,
1159
+ "p99": 0.03701298232421875,
1160
+ "unit": "seconds/token"
1161
+ }
1162
+ },
1163
+ "gpu_metrics": {
1164
+ "gpu_utilization_mean": 90.48387096774194,
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
+ }