Upload /phi-2/int8_batch_size_1_sq_len_256_new_tokens_256/benchmark_report.json with huggingface_hub
Browse files
phi-2/int8_batch_size_1_sq_len_256_new_tokens_256/benchmark_report.json
CHANGED
|
@@ -2,209 +2,180 @@
|
|
| 2 |
"prefill": {
|
| 3 |
"memory": {
|
| 4 |
"unit": "MB",
|
| 5 |
-
"max_ram":
|
| 6 |
-
"max_vram":
|
| 7 |
"max_reserved": 3353.346048,
|
| 8 |
"max_allocated": 3294.665216
|
| 9 |
},
|
| 10 |
"latency": {
|
| 11 |
"unit": "s",
|
| 12 |
-
"mean": 0.
|
| 13 |
-
"stdev": 0.
|
| 14 |
"values": [
|
| 15 |
-
0.
|
| 16 |
-
0.
|
| 17 |
-
0.
|
| 18 |
-
0.
|
| 19 |
-
0.
|
| 20 |
-
0.
|
| 21 |
-
0.
|
| 22 |
-
0.
|
| 23 |
-
0.
|
| 24 |
-
0.
|
| 25 |
-
0.
|
| 26 |
-
0.
|
| 27 |
-
0.
|
| 28 |
-
0.
|
| 29 |
-
0.
|
| 30 |
-
0.
|
| 31 |
-
0.
|
| 32 |
-
0.
|
| 33 |
-
0.
|
| 34 |
-
0.
|
| 35 |
-
0.
|
| 36 |
-
0.
|
| 37 |
-
0.
|
| 38 |
-
0.
|
| 39 |
-
0.
|
| 40 |
-
0.
|
| 41 |
-
0.
|
| 42 |
-
0.
|
| 43 |
-
0.
|
| 44 |
-
0.
|
| 45 |
-
0.
|
| 46 |
-
0.
|
| 47 |
-
0.
|
| 48 |
-
0.
|
| 49 |
-
0.
|
| 50 |
-
0.
|
| 51 |
-
0.
|
| 52 |
-
0.
|
| 53 |
-
0.
|
| 54 |
-
0.
|
| 55 |
-
0.
|
| 56 |
-
0.
|
| 57 |
-
0.
|
| 58 |
-
0.
|
| 59 |
-
0.
|
| 60 |
-
0.
|
| 61 |
-
0.
|
| 62 |
-
0.
|
| 63 |
-
0.
|
| 64 |
-
0.
|
| 65 |
-
0.
|
| 66 |
-
0.
|
| 67 |
-
0.
|
| 68 |
-
0.
|
| 69 |
-
0.
|
| 70 |
-
0.
|
| 71 |
-
0.
|
| 72 |
-
0.
|
| 73 |
-
0.
|
| 74 |
-
0.
|
| 75 |
-
0.
|
| 76 |
-
0.
|
| 77 |
-
0.
|
| 78 |
-
0.
|
| 79 |
-
0.
|
| 80 |
-
0.
|
| 81 |
-
0.
|
| 82 |
-
0.
|
| 83 |
-
0.
|
| 84 |
-
0.
|
| 85 |
-
0.
|
| 86 |
-
0.
|
| 87 |
-
0.
|
| 88 |
-
0.
|
| 89 |
-
0.
|
| 90 |
-
0.
|
| 91 |
-
0.
|
| 92 |
-
0.
|
| 93 |
-
0.
|
| 94 |
-
0.
|
| 95 |
-
0.
|
| 96 |
-
0.
|
| 97 |
-
0.
|
| 98 |
-
0.
|
| 99 |
-
0.
|
| 100 |
-
0.
|
| 101 |
-
0.
|
| 102 |
-
0.
|
| 103 |
-
0.
|
| 104 |
-
0.
|
| 105 |
-
0.
|
| 106 |
-
0.
|
| 107 |
-
0.
|
| 108 |
-
0.
|
| 109 |
-
0.
|
| 110 |
-
0.
|
| 111 |
-
0.
|
| 112 |
-
0.
|
| 113 |
-
0.
|
| 114 |
-
0.
|
| 115 |
-
0.
|
| 116 |
-
0.
|
| 117 |
-
0.
|
| 118 |
-
0.
|
| 119 |
-
0.
|
| 120 |
-
0.
|
| 121 |
-
0.
|
| 122 |
-
0.
|
| 123 |
-
0.
|
| 124 |
-
0.
|
| 125 |
-
0.
|
| 126 |
-
0.
|
| 127 |
-
0.
|
| 128 |
-
0.
|
| 129 |
-
0.
|
| 130 |
-
0.
|
| 131 |
-
0.
|
| 132 |
-
0.
|
| 133 |
-
0.
|
| 134 |
-
0.
|
| 135 |
-
0.
|
| 136 |
-
0.
|
| 137 |
-
0.
|
| 138 |
-
0.
|
| 139 |
-
0.
|
| 140 |
-
0.
|
| 141 |
-
0.
|
| 142 |
-
0.
|
| 143 |
-
0.
|
| 144 |
-
0.
|
| 145 |
-
0.
|
| 146 |
-
0.
|
| 147 |
-
0.
|
| 148 |
-
0.
|
| 149 |
-
0.
|
| 150 |
-
0.
|
| 151 |
-
0.
|
| 152 |
-
0.
|
| 153 |
-
0.
|
| 154 |
-
0.
|
| 155 |
-
0.
|
| 156 |
-
0.
|
| 157 |
-
0.
|
| 158 |
-
0.
|
| 159 |
-
0.
|
| 160 |
-
0.
|
| 161 |
-
0.
|
| 162 |
-
0.
|
| 163 |
-
0.
|
| 164 |
-
0.
|
| 165 |
-
0.
|
| 166 |
-
0.
|
| 167 |
-
0.
|
| 168 |
-
0.
|
| 169 |
-
0.
|
| 170 |
-
0.
|
| 171 |
-
0.
|
| 172 |
-
0.
|
| 173 |
-
0.
|
| 174 |
-
0.05304729461669922,
|
| 175 |
-
0.05302374267578125,
|
| 176 |
-
0.05337497711181641,
|
| 177 |
-
0.0533309440612793,
|
| 178 |
-
0.053207038879394535,
|
| 179 |
-
0.053735424041748046,
|
| 180 |
-
0.05321334457397461,
|
| 181 |
-
0.05309952163696289,
|
| 182 |
-
0.05318656158447266,
|
| 183 |
-
0.05317529678344726,
|
| 184 |
-
0.05306470489501953,
|
| 185 |
-
0.05321625518798828,
|
| 186 |
-
0.053422080993652345,
|
| 187 |
-
0.05324294281005859,
|
| 188 |
-
0.05323878479003906,
|
| 189 |
-
0.05356748962402344,
|
| 190 |
-
0.05315379333496094,
|
| 191 |
-
0.053207134246826174,
|
| 192 |
-
0.05301760101318359,
|
| 193 |
-
0.05303807830810547,
|
| 194 |
-
0.053100543975830077,
|
| 195 |
-
0.053369857788085937,
|
| 196 |
-
0.05324288177490234,
|
| 197 |
-
0.05326847839355469,
|
| 198 |
-
0.05361459350585938,
|
| 199 |
-
0.05303705596923828,
|
| 200 |
-
0.05309439849853516,
|
| 201 |
-
0.05326969528198242,
|
| 202 |
-
0.05319270324707031
|
| 203 |
]
|
| 204 |
},
|
| 205 |
"throughput": {
|
| 206 |
"unit": "tokens/s",
|
| 207 |
-
"value":
|
| 208 |
},
|
| 209 |
"energy": null,
|
| 210 |
"efficiency": null
|
|
@@ -212,22 +183,22 @@
|
|
| 212 |
"decode": {
|
| 213 |
"memory": {
|
| 214 |
"unit": "MB",
|
| 215 |
-
"max_ram":
|
| 216 |
"max_vram": 4129.292288,
|
| 217 |
"max_reserved": 3636.461568,
|
| 218 |
"max_allocated": 3603.202048
|
| 219 |
},
|
| 220 |
"latency": {
|
| 221 |
"unit": "s",
|
| 222 |
-
"mean":
|
| 223 |
"stdev": 0,
|
| 224 |
"values": [
|
| 225 |
-
|
| 226 |
]
|
| 227 |
},
|
| 228 |
"throughput": {
|
| 229 |
"unit": "tokens/s",
|
| 230 |
-
"value":
|
| 231 |
},
|
| 232 |
"energy": null,
|
| 233 |
"efficiency": null
|
|
@@ -236,269 +207,269 @@
|
|
| 236 |
"memory": null,
|
| 237 |
"latency": {
|
| 238 |
"unit": "s",
|
| 239 |
-
"mean": 0.
|
| 240 |
-
"stdev": 0.
|
| 241 |
"values": [
|
| 242 |
-
0.
|
| 243 |
-
0.
|
| 244 |
-
0.
|
| 245 |
-
0.
|
| 246 |
-
0.
|
| 247 |
-
0.
|
| 248 |
-
0.
|
| 249 |
-
0.
|
| 250 |
-
0.
|
| 251 |
-
0.
|
| 252 |
-
0.
|
| 253 |
-
0.
|
| 254 |
-
0.
|
| 255 |
-
0.
|
| 256 |
-
0.
|
| 257 |
-
0.
|
| 258 |
-
0.
|
| 259 |
-
0.
|
| 260 |
-
0.
|
| 261 |
-
0.
|
| 262 |
-
0.
|
| 263 |
-
0.
|
| 264 |
-
0.
|
| 265 |
-
0.
|
| 266 |
-
0.
|
| 267 |
-
0.
|
| 268 |
-
0.
|
| 269 |
-
0.
|
| 270 |
-
0.
|
| 271 |
-
0.
|
| 272 |
-
0.
|
| 273 |
-
0.
|
| 274 |
-
0.
|
| 275 |
-
0.
|
| 276 |
-
0.
|
| 277 |
-
0.
|
| 278 |
-
0.
|
| 279 |
-
0.
|
| 280 |
-
0.
|
| 281 |
-
0.
|
| 282 |
-
0.
|
| 283 |
-
0.
|
| 284 |
-
0.
|
| 285 |
-
0.
|
| 286 |
-
0.
|
| 287 |
-
0.
|
| 288 |
-
0.
|
| 289 |
-
0.
|
| 290 |
-
0.
|
| 291 |
-
0.
|
| 292 |
-
0.
|
| 293 |
-
0.
|
| 294 |
-
0.
|
| 295 |
-
0.
|
| 296 |
-
0.
|
| 297 |
-
0.
|
| 298 |
-
0.
|
| 299 |
-
0.
|
| 300 |
-
0.
|
| 301 |
-
0.
|
| 302 |
-
0.
|
| 303 |
-
0.
|
| 304 |
-
0.
|
| 305 |
-
0.
|
| 306 |
-
0.
|
| 307 |
-
0.
|
| 308 |
-
0.
|
| 309 |
-
0.
|
| 310 |
-
0.
|
| 311 |
-
0.
|
| 312 |
-
0.
|
| 313 |
-
0.
|
| 314 |
-
0.
|
| 315 |
-
0.
|
| 316 |
-
0.
|
| 317 |
-
0.
|
| 318 |
-
0.
|
| 319 |
-
0.
|
| 320 |
-
0.
|
| 321 |
-
0.
|
| 322 |
-
0.
|
| 323 |
-
0.
|
| 324 |
-
0.
|
| 325 |
-
0.
|
| 326 |
-
0.
|
| 327 |
-
0.
|
| 328 |
-
0.
|
| 329 |
-
0.
|
| 330 |
-
0.
|
| 331 |
-
0.
|
| 332 |
-
0.
|
| 333 |
-
0.
|
| 334 |
-
0.
|
| 335 |
-
0.
|
| 336 |
-
0.
|
| 337 |
-
0.
|
| 338 |
-
0.
|
| 339 |
-
0.
|
| 340 |
-
0.
|
| 341 |
-
0.
|
| 342 |
-
0.
|
| 343 |
-
0.
|
| 344 |
-
0.
|
| 345 |
-
0.
|
| 346 |
-
0.
|
| 347 |
-
0.
|
| 348 |
-
0.
|
| 349 |
-
0.
|
| 350 |
-
0.
|
| 351 |
-
0.
|
| 352 |
-
0.
|
| 353 |
-
0.
|
| 354 |
-
0.
|
| 355 |
-
0.
|
| 356 |
-
0.
|
| 357 |
-
0.
|
| 358 |
-
0.
|
| 359 |
-
0.
|
| 360 |
-
0.
|
| 361 |
-
0.
|
| 362 |
-
0.
|
| 363 |
-
0.
|
| 364 |
-
0.
|
| 365 |
-
0.
|
| 366 |
-
0.
|
| 367 |
-
0.
|
| 368 |
-
0.
|
| 369 |
-
0.
|
| 370 |
-
0.
|
| 371 |
-
0.
|
| 372 |
-
0.
|
| 373 |
-
0.
|
| 374 |
-
0.
|
| 375 |
-
0.
|
| 376 |
-
0.
|
| 377 |
-
0.
|
| 378 |
-
0.
|
| 379 |
-
0.
|
| 380 |
-
0.
|
| 381 |
-
0.
|
| 382 |
-
0.
|
| 383 |
-
0.
|
| 384 |
-
0.
|
| 385 |
-
0.
|
| 386 |
-
0.
|
| 387 |
-
0.
|
| 388 |
-
0.
|
| 389 |
-
0.
|
| 390 |
-
0.
|
| 391 |
-
0.
|
| 392 |
-
0.
|
| 393 |
-
0.
|
| 394 |
-
0.
|
| 395 |
-
0.
|
| 396 |
-
0.
|
| 397 |
-
0.
|
| 398 |
-
0.
|
| 399 |
-
0.
|
| 400 |
-
0.
|
| 401 |
-
0.
|
| 402 |
-
0.
|
| 403 |
-
0.
|
| 404 |
-
0.
|
| 405 |
-
0.
|
| 406 |
-
0.
|
| 407 |
-
0.
|
| 408 |
-
0.
|
| 409 |
-
0.
|
| 410 |
-
0.
|
| 411 |
-
0.
|
| 412 |
-
0.
|
| 413 |
-
0.
|
| 414 |
-
0.
|
| 415 |
-
0.
|
| 416 |
-
0.
|
| 417 |
-
0.
|
| 418 |
-
0.
|
| 419 |
-
0.
|
| 420 |
-
0.
|
| 421 |
-
0.
|
| 422 |
-
0.
|
| 423 |
-
0.
|
| 424 |
-
0.
|
| 425 |
-
0.
|
| 426 |
-
0.
|
| 427 |
-
0.
|
| 428 |
-
0.
|
| 429 |
-
0.
|
| 430 |
-
0.
|
| 431 |
-
0.
|
| 432 |
-
0.
|
| 433 |
-
0.
|
| 434 |
-
0.
|
| 435 |
-
0.
|
| 436 |
-
0.
|
| 437 |
-
0.
|
| 438 |
-
0.
|
| 439 |
-
0.
|
| 440 |
-
0.
|
| 441 |
-
0.
|
| 442 |
-
0.
|
| 443 |
-
0.
|
| 444 |
-
0.
|
| 445 |
-
0.
|
| 446 |
-
0.
|
| 447 |
-
0.
|
| 448 |
-
0.
|
| 449 |
-
0.
|
| 450 |
-
0.
|
| 451 |
-
0.
|
| 452 |
-
0.
|
| 453 |
-
0.
|
| 454 |
-
0.
|
| 455 |
-
0.
|
| 456 |
-
0.
|
| 457 |
-
0.
|
| 458 |
-
0.
|
| 459 |
-
0.
|
| 460 |
-
0.
|
| 461 |
-
0.
|
| 462 |
-
0.
|
| 463 |
-
0.
|
| 464 |
-
0.
|
| 465 |
-
0.
|
| 466 |
-
0.
|
| 467 |
-
0.
|
| 468 |
-
0.
|
| 469 |
-
0.
|
| 470 |
-
0.
|
| 471 |
-
0.
|
| 472 |
-
0.
|
| 473 |
-
0.
|
| 474 |
-
0.
|
| 475 |
-
0.
|
| 476 |
-
0.
|
| 477 |
-
0.
|
| 478 |
-
0.
|
| 479 |
-
0.
|
| 480 |
-
0.
|
| 481 |
-
0.
|
| 482 |
-
0.
|
| 483 |
-
0.
|
| 484 |
-
0.
|
| 485 |
-
0.
|
| 486 |
-
0.
|
| 487 |
-
0.
|
| 488 |
-
0.
|
| 489 |
-
0.
|
| 490 |
-
0.
|
| 491 |
-
0.
|
| 492 |
-
0.
|
| 493 |
-
0.
|
| 494 |
-
0.
|
| 495 |
-
0.
|
| 496 |
-
0.
|
| 497 |
]
|
| 498 |
},
|
| 499 |
"throughput": {
|
| 500 |
"unit": "tokens/s",
|
| 501 |
-
"value":
|
| 502 |
},
|
| 503 |
"energy": null,
|
| 504 |
"efficiency": null
|
|
|
|
| 2 |
"prefill": {
|
| 3 |
"memory": {
|
| 4 |
"unit": "MB",
|
| 5 |
+
"max_ram": 4009.75872,
|
| 6 |
+
"max_vram": 3745.513472,
|
| 7 |
"max_reserved": 3353.346048,
|
| 8 |
"max_allocated": 3294.665216
|
| 9 |
},
|
| 10 |
"latency": {
|
| 11 |
"unit": "s",
|
| 12 |
+
"mean": 0.06290689446791166,
|
| 13 |
+
"stdev": 0.0003786724204415942,
|
| 14 |
"values": [
|
| 15 |
+
0.06582316589355469,
|
| 16 |
+
0.06461644744873046,
|
| 17 |
+
0.06400614166259766,
|
| 18 |
+
0.06320454406738281,
|
| 19 |
+
0.06362112045288086,
|
| 20 |
+
0.0631910400390625,
|
| 21 |
+
0.06360678482055664,
|
| 22 |
+
0.06402867126464844,
|
| 23 |
+
0.06364876937866211,
|
| 24 |
+
0.0632360954284668,
|
| 25 |
+
0.06302412796020508,
|
| 26 |
+
0.06288998413085937,
|
| 27 |
+
0.06279270553588867,
|
| 28 |
+
0.06278246307373046,
|
| 29 |
+
0.06305894470214844,
|
| 30 |
+
0.06304358291625976,
|
| 31 |
+
0.06271283340454102,
|
| 32 |
+
0.0627435531616211,
|
| 33 |
+
0.06281625747680664,
|
| 34 |
+
0.06282035064697265,
|
| 35 |
+
0.06277324676513672,
|
| 36 |
+
0.06275788879394531,
|
| 37 |
+
0.06293708801269532,
|
| 38 |
+
0.0628223991394043,
|
| 39 |
+
0.06289408111572266,
|
| 40 |
+
0.06266777420043945,
|
| 41 |
+
0.0626319694519043,
|
| 42 |
+
0.06269644927978515,
|
| 43 |
+
0.06277030563354492,
|
| 44 |
+
0.06303846359252929,
|
| 45 |
+
0.06280806350708008,
|
| 46 |
+
0.06287667083740234,
|
| 47 |
+
0.06268620681762695,
|
| 48 |
+
0.062740478515625,
|
| 49 |
+
0.06271897506713867,
|
| 50 |
+
0.06263808059692383,
|
| 51 |
+
0.06261555099487305,
|
| 52 |
+
0.06280192184448243,
|
| 53 |
+
0.06302412796020508,
|
| 54 |
+
0.06314393615722656,
|
| 55 |
+
0.06258687973022461,
|
| 56 |
+
0.06277427291870118,
|
| 57 |
+
0.06269443130493164,
|
| 58 |
+
0.06250905609130859,
|
| 59 |
+
0.06270054244995117,
|
| 60 |
+
0.06263097763061523,
|
| 61 |
+
0.06264217758178711,
|
| 62 |
+
0.06268825531005859,
|
| 63 |
+
0.06257766342163086,
|
| 64 |
+
0.06254182434082031,
|
| 65 |
+
0.06245068740844727,
|
| 66 |
+
0.06254079818725586,
|
| 67 |
+
0.06278246307373046,
|
| 68 |
+
0.0626698226928711,
|
| 69 |
+
0.06255615997314454,
|
| 70 |
+
0.06254079818725586,
|
| 71 |
+
0.06242508697509765,
|
| 72 |
+
0.062430206298828124,
|
| 73 |
+
0.06259404754638671,
|
| 74 |
+
0.06230323028564453,
|
| 75 |
+
0.06261555099487305,
|
| 76 |
+
0.06260428619384766,
|
| 77 |
+
0.06278758239746093,
|
| 78 |
+
0.06282547378540039,
|
| 79 |
+
0.06261555099487305,
|
| 80 |
+
0.06267289733886719,
|
| 81 |
+
0.06294015884399413,
|
| 82 |
+
0.06272921752929687,
|
| 83 |
+
0.06281932830810547,
|
| 84 |
+
0.06269132614135742,
|
| 85 |
+
0.0625684471130371,
|
| 86 |
+
0.0628326416015625,
|
| 87 |
+
0.06288691329956055,
|
| 88 |
+
0.06262271881103515,
|
| 89 |
+
0.06297600173950195,
|
| 90 |
+
0.0627701759338379,
|
| 91 |
+
0.06300979232788086,
|
| 92 |
+
0.06312345504760743,
|
| 93 |
+
0.06325452804565429,
|
| 94 |
+
0.06307839965820312,
|
| 95 |
+
0.06279372787475586,
|
| 96 |
+
0.06291875076293946,
|
| 97 |
+
0.0630978546142578,
|
| 98 |
+
0.06292172622680664,
|
| 99 |
+
0.06290329742431641,
|
| 100 |
+
0.0629043197631836,
|
| 101 |
+
0.06277836990356446,
|
| 102 |
+
0.06318399810791016,
|
| 103 |
+
0.06285516738891601,
|
| 104 |
+
0.06253363037109375,
|
| 105 |
+
0.06271897506713867,
|
| 106 |
+
0.0627630729675293,
|
| 107 |
+
0.06268825531005859,
|
| 108 |
+
0.06296985626220702,
|
| 109 |
+
0.06295654296875,
|
| 110 |
+
0.06315212631225586,
|
| 111 |
+
0.0633169937133789,
|
| 112 |
+
0.06302412796020508,
|
| 113 |
+
0.06310310363769531,
|
| 114 |
+
0.06303334426879882,
|
| 115 |
+
0.06287871932983398,
|
| 116 |
+
0.06294220733642578,
|
| 117 |
+
0.06290943908691406,
|
| 118 |
+
0.06291968154907226,
|
| 119 |
+
0.06274873733520508,
|
| 120 |
+
0.06289104080200195,
|
| 121 |
+
0.06299238586425782,
|
| 122 |
+
0.06302719879150391,
|
| 123 |
+
0.06289817428588867,
|
| 124 |
+
0.06293196868896485,
|
| 125 |
+
0.06306611251831054,
|
| 126 |
+
0.06306508636474609,
|
| 127 |
+
0.0631203842163086,
|
| 128 |
+
0.06305587387084961,
|
| 129 |
+
0.06318182373046875,
|
| 130 |
+
0.06322687911987304,
|
| 131 |
+
0.06329139328002929,
|
| 132 |
+
0.06309478378295899,
|
| 133 |
+
0.06318694305419922,
|
| 134 |
+
0.06304892730712891,
|
| 135 |
+
0.0628930549621582,
|
| 136 |
+
0.06279270553588867,
|
| 137 |
+
0.06274457550048829,
|
| 138 |
+
0.06258073425292969,
|
| 139 |
+
0.06292899322509765,
|
| 140 |
+
0.06334873580932618,
|
| 141 |
+
0.06314700698852539,
|
| 142 |
+
0.0629596176147461,
|
| 143 |
+
0.0630835189819336,
|
| 144 |
+
0.0628223991394043,
|
| 145 |
+
0.06295552062988281,
|
| 146 |
+
0.06281113433837891,
|
| 147 |
+
0.0629648323059082,
|
| 148 |
+
0.06293708801269532,
|
| 149 |
+
0.06321049499511719,
|
| 150 |
+
0.06270054244995117,
|
| 151 |
+
0.06285823822021484,
|
| 152 |
+
0.06253363037109375,
|
| 153 |
+
0.06264934539794922,
|
| 154 |
+
0.0627507209777832,
|
| 155 |
+
0.0629043197631836,
|
| 156 |
+
0.06319615936279296,
|
| 157 |
+
0.06292377471923828,
|
| 158 |
+
0.06292480087280274,
|
| 159 |
+
0.06288281631469726,
|
| 160 |
+
0.062339073181152345,
|
| 161 |
+
0.06257254409790039,
|
| 162 |
+
0.06256639862060547,
|
| 163 |
+
0.06253772735595703,
|
| 164 |
+
0.06281216049194335,
|
| 165 |
+
0.06274457550048829,
|
| 166 |
+
0.06308966445922852,
|
| 167 |
+
0.06297087860107421,
|
| 168 |
+
0.0629749755859375,
|
| 169 |
+
0.06289817428588867,
|
| 170 |
+
0.06276403045654297,
|
| 171 |
+
0.06274457550048829,
|
| 172 |
+
0.06286950302124024,
|
| 173 |
+
0.06293708801269532
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 174 |
]
|
| 175 |
},
|
| 176 |
"throughput": {
|
| 177 |
"unit": "tokens/s",
|
| 178 |
+
"value": 4069.5062467371317
|
| 179 |
},
|
| 180 |
"energy": null,
|
| 181 |
"efficiency": null
|
|
|
|
| 183 |
"decode": {
|
| 184 |
"memory": {
|
| 185 |
"unit": "MB",
|
| 186 |
+
"max_ram": 4009.75872,
|
| 187 |
"max_vram": 4129.292288,
|
| 188 |
"max_reserved": 3636.461568,
|
| 189 |
"max_allocated": 3603.202048
|
| 190 |
},
|
| 191 |
"latency": {
|
| 192 |
"unit": "s",
|
| 193 |
+
"mean": 16.163796966552734,
|
| 194 |
"stdev": 0,
|
| 195 |
"values": [
|
| 196 |
+
16.163796966552734
|
| 197 |
]
|
| 198 |
},
|
| 199 |
"throughput": {
|
| 200 |
"unit": "tokens/s",
|
| 201 |
+
"value": 15.775996229577984
|
| 202 |
},
|
| 203 |
"energy": null,
|
| 204 |
"efficiency": null
|
|
|
|
| 207 |
"memory": null,
|
| 208 |
"latency": {
|
| 209 |
"unit": "s",
|
| 210 |
+
"mean": 0.06338743908452053,
|
| 211 |
+
"stdev": 0.00021097461187123576,
|
| 212 |
"values": [
|
| 213 |
+
0.06388838577270507,
|
| 214 |
+
0.0632248306274414,
|
| 215 |
+
0.06287871932983398,
|
| 216 |
+
0.06329036712646484,
|
| 217 |
+
0.06310297775268554,
|
| 218 |
+
0.06338764953613281,
|
| 219 |
+
0.0628930549621582,
|
| 220 |
+
0.06305382537841797,
|
| 221 |
+
0.0630015983581543,
|
| 222 |
+
0.06299033737182617,
|
| 223 |
+
0.06292684936523438,
|
| 224 |
+
0.06295142364501953,
|
| 225 |
+
0.06317772674560547,
|
| 226 |
+
0.06312652969360352,
|
| 227 |
+
0.06322175979614258,
|
| 228 |
+
0.06295859146118164,
|
| 229 |
+
0.06317670440673828,
|
| 230 |
+
0.06290329742431641,
|
| 231 |
+
0.06291968154907226,
|
| 232 |
+
0.06306304168701173,
|
| 233 |
+
0.06320127868652343,
|
| 234 |
+
0.06324838256835938,
|
| 235 |
+
0.06341734313964843,
|
| 236 |
+
0.0631982078552246,
|
| 237 |
+
0.06310297775268554,
|
| 238 |
+
0.06286643218994141,
|
| 239 |
+
0.06345523071289062,
|
| 240 |
+
0.0628674545288086,
|
| 241 |
+
0.06296063995361328,
|
| 242 |
+
0.0627845115661621,
|
| 243 |
+
0.06320844650268555,
|
| 244 |
+
0.06307942581176758,
|
| 245 |
+
0.06330572891235352,
|
| 246 |
+
0.06288998413085937,
|
| 247 |
+
0.06309580612182618,
|
| 248 |
+
0.06322892761230468,
|
| 249 |
+
0.06318489456176758,
|
| 250 |
+
0.06309580612182618,
|
| 251 |
+
0.0634071044921875,
|
| 252 |
+
0.06306611251831054,
|
| 253 |
+
0.06311423873901367,
|
| 254 |
+
0.06337228775024414,
|
| 255 |
+
0.06350745773315429,
|
| 256 |
+
0.06321254348754883,
|
| 257 |
+
0.06332825469970703,
|
| 258 |
+
0.06314700698852539,
|
| 259 |
+
0.06315008163452149,
|
| 260 |
+
0.06307839965820312,
|
| 261 |
+
0.06320230484008789,
|
| 262 |
+
0.06327705764770508,
|
| 263 |
+
0.06306304168701173,
|
| 264 |
+
0.06323199844360351,
|
| 265 |
+
0.06322687911987304,
|
| 266 |
+
0.06313369750976562,
|
| 267 |
+
0.06313062286376953,
|
| 268 |
+
0.06310809707641601,
|
| 269 |
+
0.06317055892944336,
|
| 270 |
+
0.0632176628112793,
|
| 271 |
+
0.06330060958862305,
|
| 272 |
+
0.06348799896240234,
|
| 273 |
+
0.0635525131225586,
|
| 274 |
+
0.0630978546142578,
|
| 275 |
+
0.06342348861694336,
|
| 276 |
+
0.06308044815063477,
|
| 277 |
+
0.06338560104370117,
|
| 278 |
+
0.06349209594726563,
|
| 279 |
+
0.06344089508056641,
|
| 280 |
+
0.0635146255493164,
|
| 281 |
+
0.06335283279418945,
|
| 282 |
+
0.06334873580932618,
|
| 283 |
+
0.06360883331298828,
|
| 284 |
+
0.06339379119873047,
|
| 285 |
+
0.06337740707397461,
|
| 286 |
+
0.06321971130371094,
|
| 287 |
+
0.06344704055786132,
|
| 288 |
+
0.06339891052246094,
|
| 289 |
+
0.06325964736938476,
|
| 290 |
+
0.06331903839111327,
|
| 291 |
+
0.06317567825317383,
|
| 292 |
+
0.06334873580932618,
|
| 293 |
+
0.06328319931030274,
|
| 294 |
+
0.0632883186340332,
|
| 295 |
+
0.06353715133666993,
|
| 296 |
+
0.06358937454223633,
|
| 297 |
+
0.06351359939575195,
|
| 298 |
+
0.06332108688354492,
|
| 299 |
+
0.0638658561706543,
|
| 300 |
+
0.06340607833862305,
|
| 301 |
+
0.06314803314208985,
|
| 302 |
+
0.06335283279418945,
|
| 303 |
+
0.06311936187744141,
|
| 304 |
+
0.06340198516845703,
|
| 305 |
+
0.06334771347045899,
|
| 306 |
+
0.0632985610961914,
|
| 307 |
+
0.06330368041992188,
|
| 308 |
+
0.06326374435424804,
|
| 309 |
+
0.0633589744567871,
|
| 310 |
+
0.06338355255126953,
|
| 311 |
+
0.0634777603149414,
|
| 312 |
+
0.06348492813110351,
|
| 313 |
+
0.06337228775024414,
|
| 314 |
+
0.06328217697143555,
|
| 315 |
+
0.06345318222045898,
|
| 316 |
+
0.06348697662353515,
|
| 317 |
+
0.06339686584472656,
|
| 318 |
+
0.06344294357299805,
|
| 319 |
+
0.06350233459472657,
|
| 320 |
+
0.06358732986450195,
|
| 321 |
+
0.06361088180541992,
|
| 322 |
+
0.06358937454223633,
|
| 323 |
+
0.06327500915527344,
|
| 324 |
+
0.06337945556640626,
|
| 325 |
+
0.06331401443481445,
|
| 326 |
+
0.06342748641967773,
|
| 327 |
+
0.06362214279174805,
|
| 328 |
+
0.06343987274169922,
|
| 329 |
+
0.06356275177001953,
|
| 330 |
+
0.06367334365844726,
|
| 331 |
+
0.06362931060791016,
|
| 332 |
+
0.06342963027954102,
|
| 333 |
+
0.06338150405883788,
|
| 334 |
+
0.06324838256835938,
|
| 335 |
+
0.06337535858154297,
|
| 336 |
+
0.06334771347045899,
|
| 337 |
+
0.06348083114624023,
|
| 338 |
+
0.06344089508056641,
|
| 339 |
+
0.06379622268676757,
|
| 340 |
+
0.06397542572021485,
|
| 341 |
+
0.06342348861694336,
|
| 342 |
+
0.06363750457763671,
|
| 343 |
+
0.06341734313964843,
|
| 344 |
+
0.06331596755981446,
|
| 345 |
+
0.0633733139038086,
|
| 346 |
+
0.06331289672851563,
|
| 347 |
+
0.06333951950073242,
|
| 348 |
+
0.06339481735229492,
|
| 349 |
+
0.06362623977661133,
|
| 350 |
+
0.06348185729980468,
|
| 351 |
+
0.06323199844360351,
|
| 352 |
+
0.06339583969116211,
|
| 353 |
+
0.06341324615478515,
|
| 354 |
+
0.06356582260131836,
|
| 355 |
+
0.06359347152709961,
|
| 356 |
+
0.06343167877197266,
|
| 357 |
+
0.0632811508178711,
|
| 358 |
+
0.06361497497558594,
|
| 359 |
+
0.06347264099121094,
|
| 360 |
+
0.06362521743774414,
|
| 361 |
+
0.0633589744567871,
|
| 362 |
+
0.06353715133666993,
|
| 363 |
+
0.0635146255493164,
|
| 364 |
+
0.06344089508056641,
|
| 365 |
+
0.06329651260375976,
|
| 366 |
+
0.06339993667602539,
|
| 367 |
+
0.06389247894287109,
|
| 368 |
+
0.06362214279174805,
|
| 369 |
+
0.06343065643310547,
|
| 370 |
+
0.0634439697265625,
|
| 371 |
+
0.0634224624633789,
|
| 372 |
+
0.06349619293212891,
|
| 373 |
+
0.06354022216796874,
|
| 374 |
+
0.06349209594726563,
|
| 375 |
+
0.06348185729980468,
|
| 376 |
+
0.06337535858154297,
|
| 377 |
+
0.06347264099121094,
|
| 378 |
+
0.06325555038452148,
|
| 379 |
+
0.06328217697143555,
|
| 380 |
+
0.06341120147705079,
|
| 381 |
+
0.06329446411132812,
|
| 382 |
+
0.06352793502807617,
|
| 383 |
+
0.063351806640625,
|
| 384 |
+
0.06351974487304687,
|
| 385 |
+
0.063604736328125,
|
| 386 |
+
0.06341529464721679,
|
| 387 |
+
0.06369894409179687,
|
| 388 |
+
0.06332108688354492,
|
| 389 |
+
0.06360575866699218,
|
| 390 |
+
0.06363033676147462,
|
| 391 |
+
0.06362828826904297,
|
| 392 |
+
0.06354534530639648,
|
| 393 |
+
0.06341120147705079,
|
| 394 |
+
0.0633251838684082,
|
| 395 |
+
0.06339891052246094,
|
| 396 |
+
0.06390476989746094,
|
| 397 |
+
0.0636753921508789,
|
| 398 |
+
0.06348287963867187,
|
| 399 |
+
0.06330879974365235,
|
| 400 |
+
0.06356787109375,
|
| 401 |
+
0.06356275177001953,
|
| 402 |
+
0.06340300750732422,
|
| 403 |
+
0.06380543899536133,
|
| 404 |
+
0.06352384185791016,
|
| 405 |
+
0.06304153442382812,
|
| 406 |
+
0.06367641448974609,
|
| 407 |
+
0.06364774322509766,
|
| 408 |
+
0.06353919982910156,
|
| 409 |
+
0.0635412483215332,
|
| 410 |
+
0.0634224624633789,
|
| 411 |
+
0.06327500915527344,
|
| 412 |
+
0.06339276885986328,
|
| 413 |
+
0.06340095901489258,
|
| 414 |
+
0.06345727920532226,
|
| 415 |
+
0.06366003036499024,
|
| 416 |
+
0.06357196807861328,
|
| 417 |
+
0.06364262390136718,
|
| 418 |
+
0.06334156799316407,
|
| 419 |
+
0.06356275177001953,
|
| 420 |
+
0.06349824142456055,
|
| 421 |
+
0.06358937454223633,
|
| 422 |
+
0.0636835823059082,
|
| 423 |
+
0.06356172943115235,
|
| 424 |
+
0.06360166549682617,
|
| 425 |
+
0.0639733772277832,
|
| 426 |
+
0.06333030319213867,
|
| 427 |
+
0.06327705764770508,
|
| 428 |
+
0.06332723236083984,
|
| 429 |
+
0.06349209594726563,
|
| 430 |
+
0.06334771347045899,
|
| 431 |
+
0.06351359939575195,
|
| 432 |
+
0.06350950241088867,
|
| 433 |
+
0.06330572891235352,
|
| 434 |
+
0.0635863037109375,
|
| 435 |
+
0.0635750389099121,
|
| 436 |
+
0.06344908905029296,
|
| 437 |
+
0.06360883331298828,
|
| 438 |
+
0.06336614227294922,
|
| 439 |
+
0.06358425521850586,
|
| 440 |
+
0.06331494522094727,
|
| 441 |
+
0.06327910232543946,
|
| 442 |
+
0.06339174270629883,
|
| 443 |
+
0.0633804817199707,
|
| 444 |
+
0.06396108627319336,
|
| 445 |
+
0.06338457489013671,
|
| 446 |
+
0.06327603149414063,
|
| 447 |
+
0.06362009429931641,
|
| 448 |
+
0.06354431915283203,
|
| 449 |
+
0.06345420837402344,
|
| 450 |
+
0.06330777740478516,
|
| 451 |
+
0.06341324615478515,
|
| 452 |
+
0.063246337890625,
|
| 453 |
+
0.06367641448974609,
|
| 454 |
+
0.06353203201293946,
|
| 455 |
+
0.0636119041442871,
|
| 456 |
+
0.06343884658813477,
|
| 457 |
+
0.06353203201293946,
|
| 458 |
+
0.06345318222045898,
|
| 459 |
+
0.0634071044921875,
|
| 460 |
+
0.06350745773315429,
|
| 461 |
+
0.06348287963867187,
|
| 462 |
+
0.06365388870239258,
|
| 463 |
+
0.06368255996704102,
|
| 464 |
+
0.06342348861694336,
|
| 465 |
+
0.063425537109375,
|
| 466 |
+
0.06323916625976563,
|
| 467 |
+
0.06340198516845703
|
| 468 |
]
|
| 469 |
},
|
| 470 |
"throughput": {
|
| 471 |
"unit": "tokens/s",
|
| 472 |
+
"value": 15.775996229577983
|
| 473 |
},
|
| 474 |
"energy": null,
|
| 475 |
"efficiency": null
|