vec-inf 0.6.1__py3-none-any.whl → 0.7.1__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (22) hide show
  1. vec_inf/README.md +18 -4
  2. vec_inf/cli/_cli.py +191 -34
  3. vec_inf/cli/_helper.py +95 -14
  4. vec_inf/client/_client_vars.py +7 -165
  5. vec_inf/client/_helper.py +386 -40
  6. vec_inf/client/_slurm_script_generator.py +204 -36
  7. vec_inf/client/_slurm_templates.py +248 -0
  8. vec_inf/client/_slurm_vars.py +86 -0
  9. vec_inf/client/_utils.py +189 -70
  10. vec_inf/client/api.py +96 -25
  11. vec_inf/client/config.py +40 -19
  12. vec_inf/client/models.py +44 -4
  13. vec_inf/config/README.md +4 -243
  14. vec_inf/config/environment.yaml +35 -0
  15. vec_inf/config/models.yaml +102 -274
  16. {vec_inf-0.6.1.dist-info → vec_inf-0.7.1.dist-info}/METADATA +43 -73
  17. vec_inf-0.7.1.dist-info/RECORD +27 -0
  18. vec_inf/client/slurm_vars.py +0 -49
  19. vec_inf-0.6.1.dist-info/RECORD +0 -25
  20. {vec_inf-0.6.1.dist-info → vec_inf-0.7.1.dist-info}/WHEEL +0 -0
  21. {vec_inf-0.6.1.dist-info → vec_inf-0.7.1.dist-info}/entry_points.txt +0 -0
  22. {vec_inf-0.6.1.dist-info → vec_inf-0.7.1.dist-info}/licenses/LICENSE +0 -0
vec_inf/config/models.yaml CHANGED
@@ -1,19 +1,4 @@
1
1
  models:
2
- c4ai-command-r-plus:
3
- model_family: c4ai-command-r
4
- model_variant: plus
5
- model_type: LLM
6
- gpus_per_node: 4
7
- num_nodes: 2
8
- vocab_size: 256000
9
- qos: m2
10
- time: 08:00:00
11
- partition: a40
12
- vllm_args:
13
- --pipeline-parallel-size: 2
14
- --tensor-parallel-size: 4
15
- --max-model-len: 8192
16
- --max-num-seqs: 256
17
2
  c4ai-command-r-plus-08-2024:
18
3
  model_family: c4ai-command-r
19
4
  model_variant: plus-08-2024
@@ -21,9 +6,8 @@ models:
21
6
  gpus_per_node: 4
22
7
  num_nodes: 2
23
8
  vocab_size: 256000
24
- qos: m2
25
9
  time: 08:00:00
26
- partition: a40
10
+ resource_type: l40s
27
11
  vllm_args:
28
12
  --pipeline-parallel-size: 2
29
13
  --tensor-parallel-size: 4
@@ -36,14 +20,12 @@ models:
36
20
  gpus_per_node: 2
37
21
  num_nodes: 1
38
22
  vocab_size: 256000
39
- qos: m2
40
23
  time: 08:00:00
41
- partition: a40
24
+ resource_type: l40s
42
25
  vllm_args:
43
26
  --tensor-parallel-size: 2
44
27
  --max-model-len: 32768
45
28
  --max-num-seqs: 256
46
- --compilation-config: 3
47
29
  CodeLlama-7b-hf:
48
30
  model_family: CodeLlama
49
31
  model_variant: 7b-hf
@@ -51,13 +33,11 @@ models:
51
33
  gpus_per_node: 1
52
34
  num_nodes: 1
53
35
  vocab_size: 32000
54
- qos: m2
55
36
  time: 08:00:00
56
- partition: a40
37
+ resource_type: l40s
57
38
  vllm_args:
58
39
  --max-model-len: 16384
59
40
  --max-num-seqs: 256
60
- --compilation-config: 3
61
41
  CodeLlama-7b-Instruct-hf:
62
42
  model_family: CodeLlama
63
43
  model_variant: 7b-Instruct-hf
@@ -65,13 +45,11 @@ models:
65
45
  gpus_per_node: 1
66
46
  num_nodes: 1
67
47
  vocab_size: 32000
68
- qos: m2
69
48
  time: 08:00:00
70
- partition: a40
49
+ resource_type: l40s
71
50
  vllm_args:
72
51
  --max-model-len: 16384
73
52
  --max-num-seqs: 256
74
- --compilation-config: 3
75
53
  CodeLlama-13b-hf:
76
54
  model_family: CodeLlama
77
55
  model_variant: 13b-hf
@@ -79,13 +57,11 @@ models:
79
57
  gpus_per_node: 1
80
58
  num_nodes: 1
81
59
  vocab_size: 32000
82
- qos: m2
83
60
  time: 08:00:00
84
- partition: a40
61
+ resource_type: l40s
85
62
  vllm_args:
86
63
  --max-model-len: 16384
87
64
  --max-num-seqs: 256
88
- --compilation-config: 3
89
65
  CodeLlama-13b-Instruct-hf:
90
66
  model_family: CodeLlama
91
67
  model_variant: 13b-Instruct-hf
@@ -93,13 +69,11 @@ models:
93
69
  gpus_per_node: 1
94
70
  num_nodes: 1
95
71
  vocab_size: 32000
96
- qos: m2
97
72
  time: 08:00:00
98
- partition: a40
73
+ resource_type: l40s
99
74
  vllm_args:
100
75
  --max-model-len: 16384
101
76
  --max-num-seqs: 256
102
- --compilation-config: 3
103
77
  CodeLlama-34b-hf:
104
78
  model_family: CodeLlama
105
79
  model_variant: 34b-hf
@@ -107,14 +81,12 @@ models:
107
81
  gpus_per_node: 2
108
82
  num_nodes: 1
109
83
  vocab_size: 32000
110
- qos: m2
111
84
  time: 08:00:00
112
- partition: a40
85
+ resource_type: l40s
113
86
  vllm_args:
114
87
  --tensor-parallel-size: 2
115
88
  --max-model-len: 16384
116
89
  --max-num-seqs: 256
117
- --compilation-config: 3
118
90
  CodeLlama-34b-Instruct-hf:
119
91
  model_family: CodeLlama
120
92
  model_variant: 34b-Instruct-hf
@@ -122,14 +94,12 @@ models:
122
94
  gpus_per_node: 2
123
95
  num_nodes: 1
124
96
  vocab_size: 32000
125
- qos: m2
126
97
  time: 08:00:00
127
- partition: a40
98
+ resource_type: l40s
128
99
  vllm_args:
129
100
  --tensor-parallel-size: 2
130
101
  --max-model-len: 16384
131
102
  --max-num-seqs: 256
132
- --compilation-config: 3
133
103
  CodeLlama-70b-hf:
134
104
  model_family: CodeLlama
135
105
  model_variant: 70b-hf
@@ -137,14 +107,12 @@ models:
137
107
  gpus_per_node: 4
138
108
  num_nodes: 1
139
109
  vocab_size: 32016
140
- qos: m2
141
110
  time: 08:00:00
142
- partition: a40
111
+ resource_type: l40s
143
112
  vllm_args:
144
113
  --tensor-parallel-size: 4
145
114
  --max-model-len: 4096
146
115
  --max-num-seqs: 256
147
- --compilation-config: 3
148
116
  CodeLlama-70b-Instruct-hf:
149
117
  model_family: CodeLlama
150
118
  model_variant: 70b-Instruct-hf
@@ -152,14 +120,12 @@ models:
152
120
  gpus_per_node: 4
153
121
  num_nodes: 1
154
122
  vocab_size: 32016
155
- qos: m2
156
123
  time: 08:00:00
157
- partition: a40
124
+ resource_type: l40s
158
125
  vllm_args:
159
126
  --tensor-parallel-size: 4
160
127
  --max-model-len: 4096
161
128
  --max-num-seqs: 256
162
- --compilation-config: 3
163
129
  gemma-2-9b:
164
130
  model_family: gemma-2
165
131
  model_variant: 9b
@@ -167,13 +133,11 @@ models:
167
133
  gpus_per_node: 1
168
134
  num_nodes: 1
169
135
  vocab_size: 256000
170
- qos: m2
171
136
  time: 08:00:00
172
- partition: a40
137
+ resource_type: l40s
173
138
  vllm_args:
174
139
  --max-model-len: 4096
175
140
  --max-num-seqs: 256
176
- --compilation-config: 3
177
141
  gemma-2-9b-it:
178
142
  model_family: gemma-2
179
143
  model_variant: 9b-it
@@ -181,13 +145,11 @@ models:
181
145
  gpus_per_node: 1
182
146
  num_nodes: 1
183
147
  vocab_size: 256000
184
- qos: m2
185
148
  time: 08:00:00
186
- partition: a40
149
+ resource_type: l40s
187
150
  vllm_args:
188
151
  --max-model-len: 4096
189
152
  --max-num-seqs: 256
190
- --compilation-config: 3
191
153
  gemma-2-27b:
192
154
  model_family: gemma-2
193
155
  model_variant: 27b
@@ -195,14 +157,12 @@ models:
195
157
  gpus_per_node: 2
196
158
  num_nodes: 1
197
159
  vocab_size: 256000
198
- qos: m2
199
160
  time: 08:00:00
200
- partition: a40
161
+ resource_type: l40s
201
162
  vllm_args:
202
163
  --tensor-parallel-size: 2
203
164
  --max-model-len: 4096
204
165
  --max-num-seqs: 256
205
- --compilation-config: 3
206
166
  gemma-2-27b-it:
207
167
  model_family: gemma-2
208
168
  model_variant: 27b-it
@@ -210,14 +170,12 @@ models:
210
170
  gpus_per_node: 2
211
171
  num_nodes: 1
212
172
  vocab_size: 256000
213
- qos: m2
214
173
  time: 08:00:00
215
- partition: a40
174
+ resource_type: l40s
216
175
  vllm_args:
217
176
  --tensor-parallel-size: 2
218
177
  --max-model-len: 4096
219
178
  --max-num-seqs: 256
220
- --compilation-config: 3
221
179
  Llama-2-7b-hf:
222
180
  model_family: Llama-2
223
181
  model_variant: 7b-hf
@@ -225,13 +183,11 @@ models:
225
183
  gpus_per_node: 1
226
184
  num_nodes: 1
227
185
  vocab_size: 32000
228
- qos: m2
229
186
  time: 08:00:00
230
- partition: a40
187
+ resource_type: l40s
231
188
  vllm_args:
232
189
  --max-model-len: 4096
233
190
  --max-num-seqs: 256
234
- --compilation-config: 3
235
191
  Llama-2-7b-chat-hf:
236
192
  model_family: Llama-2
237
193
  model_variant: 7b-chat-hf
@@ -239,13 +195,11 @@ models:
239
195
  gpus_per_node: 1
240
196
  num_nodes: 1
241
197
  vocab_size: 32000
242
- qos: m2
243
198
  time: 08:00:00
244
- partition: a40
199
+ resource_type: l40s
245
200
  vllm_args:
246
201
  --max-model-len: 4096
247
202
  --max-num-seqs: 256
248
- --compilation-config: 3
249
203
  Llama-2-13b-hf:
250
204
  model_family: Llama-2
251
205
  model_variant: 13b-hf
@@ -253,13 +207,11 @@ models:
253
207
  gpus_per_node: 1
254
208
  num_nodes: 1
255
209
  vocab_size: 32000
256
- qos: m2
257
210
  time: 08:00:00
258
- partition: a40
211
+ resource_type: l40s
259
212
  vllm_args:
260
213
  --max-model-len: 4096
261
214
  --max-num-seqs: 256
262
- --compilation-config: 3
263
215
  Llama-2-13b-chat-hf:
264
216
  model_family: Llama-2
265
217
  model_variant: 13b-chat-hf
@@ -267,13 +219,11 @@ models:
267
219
  gpus_per_node: 1
268
220
  num_nodes: 1
269
221
  vocab_size: 32000
270
- qos: m2
271
222
  time: 08:00:00
272
- partition: a40
223
+ resource_type: l40s
273
224
  vllm_args:
274
225
  --max-model-len: 4096
275
226
  --max-num-seqs: 256
276
- --compilation-config: 3
277
227
  Llama-2-70b-hf:
278
228
  model_family: Llama-2
279
229
  model_variant: 70b-hf
@@ -281,14 +231,12 @@ models:
281
231
  gpus_per_node: 4
282
232
  num_nodes: 1
283
233
  vocab_size: 32000
284
- qos: m2
285
234
  time: 08:00:00
286
- partition: a40
235
+ resource_type: l40s
287
236
  vllm_args:
288
237
  --tensor-parallel-size: 4
289
238
  --max-model-len: 4096
290
239
  --max-num-seqs: 256
291
- --compilation-config: 3
292
240
  Llama-2-70b-chat-hf:
293
241
  model_family: Llama-2
294
242
  model_variant: 70b-chat-hf
@@ -296,14 +244,12 @@ models:
296
244
  gpus_per_node: 4
297
245
  num_nodes: 1
298
246
  vocab_size: 32000
299
- qos: m2
300
247
  time: 08:00:00
301
- partition: a40
248
+ resource_type: l40s
302
249
  vllm_args:
303
250
  --tensor-parallel-size: 4
304
251
  --max-model-len: 4096
305
252
  --max-num-seqs: 256
306
- --compilation-config: 3
307
253
  llava-1.5-7b-hf:
308
254
  model_family: llava-1.5
309
255
  model_variant: 7b-hf
@@ -311,13 +257,11 @@ models:
311
257
  gpus_per_node: 1
312
258
  num_nodes: 1
313
259
  vocab_size: 32000
314
- qos: m2
315
260
  time: 08:00:00
316
- partition: a40
261
+ resource_type: l40s
317
262
  vllm_args:
318
263
  --max-model-len: 4096
319
264
  --max-num-seqs: 256
320
- --compilation-config: 3
321
265
  llava-1.5-13b-hf:
322
266
  model_family: llava-1.5
323
267
  model_variant: 13b-hf
@@ -325,13 +269,11 @@ models:
325
269
  gpus_per_node: 1
326
270
  num_nodes: 1
327
271
  vocab_size: 32000
328
- qos: m2
329
272
  time: 08:00:00
330
- partition: a40
273
+ resource_type: l40s
331
274
  vllm_args:
332
275
  --max-model-len: 4096
333
276
  --max-num-seqs: 256
334
- --compilation-config: 3
335
277
  llava-v1.6-mistral-7b-hf:
336
278
  model_family: llava-v1.6
337
279
  model_variant: mistral-7b-hf
@@ -339,13 +281,11 @@ models:
339
281
  gpus_per_node: 1
340
282
  num_nodes: 1
341
283
  vocab_size: 32064
342
- qos: m2
343
284
  time: 08:00:00
344
- partition: a40
285
+ resource_type: l40s
345
286
  vllm_args:
346
287
  --max-model-len: 32768
347
288
  --max-num-seqs: 256
348
- --compilation-config: 3
349
289
  llava-v1.6-34b-hf:
350
290
  model_family: llava-v1.6
351
291
  model_variant: 34b-hf
@@ -353,14 +293,12 @@ models:
353
293
  gpus_per_node: 2
354
294
  num_nodes: 1
355
295
  vocab_size: 64064
356
- qos: m2
357
296
  time: 08:00:00
358
- partition: a40
297
+ resource_type: l40s
359
298
  vllm_args:
360
299
  --tensor-parallel-size: 2
361
300
  --max-model-len: 4096
362
301
  --max-num-seqs: 256
363
- --compilation-config: 3
364
302
  Meta-Llama-3-8B:
365
303
  model_family: Meta-Llama-3
366
304
  model_variant: 8B
@@ -368,13 +306,11 @@ models:
368
306
  gpus_per_node: 1
369
307
  num_nodes: 1
370
308
  vocab_size: 128256
371
- qos: m2
372
309
  time: 08:00:00
373
- partition: a40
310
+ resource_type: l40s
374
311
  vllm_args:
375
312
  --max-model-len: 8192
376
313
  --max-num-seqs: 256
377
- --compilation-config: 3
378
314
  Meta-Llama-3-8B-Instruct:
379
315
  model_family: Meta-Llama-3
380
316
  model_variant: 8B-Instruct
@@ -382,13 +318,11 @@ models:
382
318
  gpus_per_node: 1
383
319
  num_nodes: 1
384
320
  vocab_size: 128256
385
- qos: m2
386
321
  time: 08:00:00
387
- partition: a40
322
+ resource_type: l40s
388
323
  vllm_args:
389
324
  --max-model-len: 8192
390
325
  --max-num-seqs: 256
391
- --compilation-config: 3
392
326
  Meta-Llama-3-70B:
393
327
  model_family: Meta-Llama-3
394
328
  model_variant: 70B
@@ -396,14 +330,12 @@ models:
396
330
  gpus_per_node: 4
397
331
  num_nodes: 1
398
332
  vocab_size: 128256
399
- qos: m2
400
333
  time: 08:00:00
401
- partition: a40
334
+ resource_type: l40s
402
335
  vllm_args:
403
336
  --tensor-parallel-size: 4
404
337
  --max-model-len: 8192
405
338
  --max-num-seqs: 256
406
- --compilation-config: 3
407
339
  Meta-Llama-3-70B-Instruct:
408
340
  model_family: Meta-Llama-3
409
341
  model_variant: 70B-Instruct
@@ -411,14 +343,12 @@ models:
411
343
  gpus_per_node: 4
412
344
  num_nodes: 1
413
345
  vocab_size: 128256
414
- qos: m2
415
346
  time: 08:00:00
416
- partition: a40
347
+ resource_type: l40s
417
348
  vllm_args:
418
349
  --tensor-parallel-size: 4
419
350
  --max-model-len: 8192
420
351
  --max-num-seqs: 256
421
- --compilation-config: 3
422
352
  Meta-Llama-3.1-8B:
423
353
  model_family: Meta-Llama-3.1
424
354
  model_variant: 8B
@@ -426,13 +356,11 @@ models:
426
356
  gpus_per_node: 1
427
357
  num_nodes: 1
428
358
  vocab_size: 128256
429
- qos: m2
430
359
  time: 08:00:00
431
- partition: a40
360
+ resource_type: l40s
432
361
  vllm_args:
433
362
  --max-model-len: 131072
434
363
  --max-num-seqs: 256
435
- --compilation-config: 3
436
364
  Meta-Llama-3.1-8B-Instruct:
437
365
  model_family: Meta-Llama-3.1
438
366
  model_variant: 8B-Instruct
@@ -440,13 +368,11 @@ models:
440
368
  gpus_per_node: 1
441
369
  num_nodes: 1
442
370
  vocab_size: 128256
443
- qos: m2
444
371
  time: 08:00:00
445
- partition: a40
372
+ resource_type: l40s
446
373
  vllm_args:
447
374
  --max-model-len: 131072
448
375
  --max-num-seqs: 256
449
- --compilation-config: 3
450
376
  Meta-Llama-3.1-70B:
451
377
  model_family: Meta-Llama-3.1
452
378
  model_variant: 70B
@@ -454,14 +380,12 @@ models:
454
380
  gpus_per_node: 4
455
381
  num_nodes: 1
456
382
  vocab_size: 128256
457
- qos: m2
458
383
  time: 08:00:00
459
- partition: a40
384
+ resource_type: l40s
460
385
  vllm_args:
461
386
  --tensor-parallel-size: 4
462
387
  --max-model-len: 65536
463
388
  --max-num-seqs: 256
464
- --compilation-config: 3
465
389
  Meta-Llama-3.1-70B-Instruct:
466
390
  model_family: Meta-Llama-3.1
467
391
  model_variant: 70B-Instruct
@@ -469,14 +393,12 @@ models:
469
393
  gpus_per_node: 4
470
394
  num_nodes: 1
471
395
  vocab_size: 128256
472
- qos: m2
473
396
  time: 08:00:00
474
- partition: a40
397
+ resource_type: l40s
475
398
  vllm_args:
476
399
  --tensor-parallel-size: 4
477
400
  --max-model-len: 65536
478
401
  --max-num-seqs: 256
479
- --compilation-config: 3
480
402
  Meta-Llama-3.1-405B-Instruct:
481
403
  model_family: Meta-Llama-3.1
482
404
  model_variant: 405B-Instruct
@@ -486,7 +408,7 @@ models:
486
408
  vocab_size: 128256
487
409
  qos: m4
488
410
  time: 02:00:00
489
- partition: a40
411
+ resource_type: l40s
490
412
  vllm_args:
491
413
  --pipeline-parallel-size: 8
492
414
  --tensor-parallel-size: 4
@@ -499,13 +421,11 @@ models:
499
421
  gpus_per_node: 1
500
422
  num_nodes: 1
501
423
  vocab_size: 32000
502
- qos: m2
503
424
  time: 08:00:00
504
- partition: a40
425
+ resource_type: l40s
505
426
  vllm_args:
506
427
  --max-model-len: 32768
507
428
  --max-num-seqs: 256
508
- --compilation-config: 3
509
429
  Mistral-7B-Instruct-v0.2:
510
430
  model_family: Mistral
511
431
  model_variant: 7B-Instruct-v0.2
@@ -513,13 +433,11 @@ models:
513
433
  gpus_per_node: 1
514
434
  num_nodes: 1
515
435
  vocab_size: 32000
516
- qos: m2
517
436
  time: 08:00:00
518
- partition: a40
437
+ resource_type: l40s
519
438
  vllm_args:
520
439
  --max-model-len: 32768
521
440
  --max-num-seqs: 256
522
- --compilation-config: 3
523
441
  Mistral-7B-v0.3:
524
442
  model_family: Mistral
525
443
  model_variant: 7B-v0.3
@@ -527,13 +445,11 @@ models:
527
445
  gpus_per_node: 1
528
446
  num_nodes: 1
529
447
  vocab_size: 32768
530
- qos: m2
531
448
  time: 08:00:00
532
- partition: a40
449
+ resource_type: l40s
533
450
  vllm_args:
534
451
  --max-model-len: 32768
535
452
  --max-num-seqs: 256
536
- --compilation-config: 3
537
453
  Mistral-7B-Instruct-v0.3:
538
454
  model_family: Mistral
539
455
  model_variant: 7B-Instruct-v0.3
@@ -541,13 +457,11 @@ models:
541
457
  gpus_per_node: 1
542
458
  num_nodes: 1
543
459
  vocab_size: 32768
544
- qos: m2
545
460
  time: 08:00:00
546
- partition: a40
461
+ resource_type: l40s
547
462
  vllm_args:
548
463
  --max-model-len: 32768
549
464
  --max-num-seqs: 256
550
- --compilation-config: 3
551
465
  Mistral-Large-Instruct-2407:
552
466
  model_family: Mistral
553
467
  model_variant: Large-Instruct-2407
@@ -555,9 +469,8 @@ models:
555
469
  gpus_per_node: 4
556
470
  num_nodes: 2
557
471
  vocab_size: 32768
558
- qos: m2
559
472
  time: 08:00:00
560
- partition: a40
473
+ resource_type: l40s
561
474
  vllm_args:
562
475
  --pipeline-parallel-size: 2
563
476
  --tensor-parallel-size: 4
@@ -570,9 +483,8 @@ models:
570
483
  gpus_per_node: 4
571
484
  num_nodes: 2
572
485
  vocab_size: 32768
573
- qos: m2
574
486
  time: 08:00:00
575
- partition: a40
487
+ resource_type: l40s
576
488
  vllm_args:
577
489
  --pipeline-parallel-size: 2
578
490
  --tensor-parallel-size: 4
@@ -585,14 +497,12 @@ models:
585
497
  gpus_per_node: 4
586
498
  num_nodes: 1
587
499
  vocab_size: 32000
588
- qos: m2
589
500
  time: 08:00:00
590
- partition: a40
501
+ resource_type: l40s
591
502
  vllm_args:
592
503
  --tensor-parallel-size: 4
593
504
  --max-model-len: 32768
594
505
  --max-num-seqs: 256
595
- --compilation-config: 3
596
506
  Mixtral-8x22B-v0.1:
597
507
  model_family: Mixtral
598
508
  model_variant: 8x22B-v0.1
@@ -600,9 +510,8 @@ models:
600
510
  gpus_per_node: 4
601
511
  num_nodes: 2
602
512
  vocab_size: 32768
603
- qos: m2
604
513
  time: 08:00:00
605
- partition: a40
514
+ resource_type: l40s
606
515
  vllm_args:
607
516
  --pipeline-parallel-size: 2
608
517
  --tensor-parallel-size: 4
@@ -615,9 +524,8 @@ models:
615
524
  gpus_per_node: 4
616
525
  num_nodes: 2
617
526
  vocab_size: 32768
618
- qos: m2
619
527
  time: 08:00:00
620
- partition: a40
528
+ resource_type: l40s
621
529
  vllm_args:
622
530
  --pipeline-parallel-size: 2
623
531
  --tensor-parallel-size: 4
@@ -630,14 +538,12 @@ models:
630
538
  gpus_per_node: 2
631
539
  num_nodes: 1
632
540
  vocab_size: 32064
633
- qos: m2
634
541
  time: 08:00:00
635
- partition: a40
542
+ resource_type: l40s
636
543
  vllm_args:
637
544
  --tensor-parallel-size: 2
638
545
  --max-model-len: 131072
639
546
  --max-num-seqs: 256
640
- --compilation-config: 3
641
547
  Phi-3-vision-128k-instruct:
642
548
  model_family: Phi-3-vision
643
549
  model_variant: 128k-instruct
@@ -645,14 +551,12 @@ models:
645
551
  gpus_per_node: 2
646
552
  num_nodes: 1
647
553
  vocab_size: 32064
648
- qos: m2
649
554
  time: 08:00:00
650
- partition: a40
555
+ resource_type: l40s
651
556
  vllm_args:
652
557
  --tensor-parallel-size: 2
653
558
  --max-model-len: 65536
654
559
  --max-num-seqs: 256
655
- --compilation-config: 3
656
560
  Llama3-OpenBioLLM-70B:
657
561
  model_family: Llama3-OpenBioLLM
658
562
  model_variant: 70B
@@ -660,14 +564,12 @@ models:
660
564
  gpus_per_node: 4
661
565
  num_nodes: 1
662
566
  vocab_size: 128256
663
- qos: m2
664
567
  time: 08:00:00
665
- partition: a40
568
+ resource_type: l40s
666
569
  vllm_args:
667
570
  --tensor-parallel-size: 4
668
571
  --max-model-len: 8192
669
572
  --max-num-seqs: 256
670
- --compilation-config: 3
671
573
  Llama-3.1-Nemotron-70B-Instruct-HF:
672
574
  model_family: Llama-3.1-Nemotron
673
575
  model_variant: 70B-Instruct-HF
@@ -675,14 +577,12 @@ models:
675
577
  gpus_per_node: 4
676
578
  num_nodes: 1
677
579
  vocab_size: 128256
678
- qos: m2
679
580
  time: 08:00:00
680
- partition: a40
581
+ resource_type: l40s
681
582
  vllm_args:
682
583
  --tensor-parallel-size: 4
683
584
  --max-model-len: 65536
684
585
  --max-num-seqs: 256
685
- --compilation-config: 3
686
586
  Llama-3.2-1B:
687
587
  model_family: Llama-3.2
688
588
  model_variant: 1B
@@ -690,13 +590,11 @@ models:
690
590
  gpus_per_node: 1
691
591
  num_nodes: 1
692
592
  vocab_size: 128256
693
- qos: m2
694
593
  time: 08:00:00
695
- partition: a40
594
+ resource_type: l40s
696
595
  vllm_args:
697
596
  --max-model-len: 131072
698
597
  --max-num-seqs: 256
699
- --compilation-config: 3
700
598
  Llama-3.2-1B-Instruct:
701
599
  model_family: Llama-3.2
702
600
  model_variant: 1B-Instruct
@@ -704,13 +602,11 @@ models:
704
602
  gpus_per_node: 1
705
603
  num_nodes: 1
706
604
  vocab_size: 128256
707
- qos: m2
708
605
  time: 08:00:00
709
- partition: a40
606
+ resource_type: l40s
710
607
  vllm_args:
711
608
  --max-model-len: 131072
712
609
  --max-num-seqs: 256
713
- --compilation-config: 3
714
610
  Llama-3.2-3B:
715
611
  model_family: Llama-3.2
716
612
  model_variant: 3B
@@ -718,13 +614,11 @@ models:
718
614
  gpus_per_node: 1
719
615
  num_nodes: 1
720
616
  vocab_size: 128256
721
- qos: m2
722
617
  time: 08:00:00
723
- partition: a40
618
+ resource_type: l40s
724
619
  vllm_args:
725
620
  --max-model-len: 131072
726
621
  --max-num-seqs: 256
727
- --compilation-config: 3
728
622
  Llama-3.2-3B-Instruct:
729
623
  model_family: Llama-3.2
730
624
  model_variant: 3B-Instruct
@@ -732,13 +626,11 @@ models:
732
626
  gpus_per_node: 1
733
627
  num_nodes: 1
734
628
  vocab_size: 128256
735
- qos: m2
736
629
  time: 08:00:00
737
- partition: a40
630
+ resource_type: l40s
738
631
  vllm_args:
739
632
  --max-model-len: 131072
740
633
  --max-num-seqs: 256
741
- --compilation-config: 3
742
634
  Llama-3.2-11B-Vision:
743
635
  model_family: Llama-3.2
744
636
  model_variant: 11B-Vision
@@ -746,14 +638,12 @@ models:
746
638
  gpus_per_node: 2
747
639
  num_nodes: 1
748
640
  vocab_size: 128256
749
- qos: m2
750
641
  time: 08:00:00
751
- partition: a40
642
+ resource_type: l40s
752
643
  vllm_args:
753
644
  --tensor-parallel-size: 2
754
645
  --max-model-len: 4096
755
646
  --max-num-seqs: 64
756
- --compilation-config: 3
757
647
  --enforce-eager: true
758
648
  Llama-3.2-11B-Vision-Instruct:
759
649
  model_family: Llama-3.2
@@ -762,14 +652,12 @@ models:
762
652
  gpus_per_node: 2
763
653
  num_nodes: 1
764
654
  vocab_size: 128256
765
- qos: m2
766
655
  time: 08:00:00
767
- partition: a40
656
+ resource_type: l40s
768
657
  vllm_args:
769
658
  --tensor-parallel-size: 2
770
659
  --max-model-len: 4096
771
660
  --max-num-seqs: 64
772
- --compilation-config: 3
773
661
  --enforce-eager: true
774
662
  Llama-3.2-90B-Vision:
775
663
  model_family: Llama-3.2
@@ -778,14 +666,12 @@ models:
778
666
  gpus_per_node: 4
779
667
  num_nodes: 2
780
668
  vocab_size: 128256
781
- qos: m2
782
669
  time: 08:00:00
783
- partition: a40
670
+ resource_type: l40s
784
671
  vllm_args:
785
672
  --tensor-parallel-size: 8
786
673
  --max-model-len: 4096
787
674
  --max-num-seqs: 32
788
- --compilation-config: 3
789
675
  --enforce-eager: true
790
676
  Llama-3.2-90B-Vision-Instruct:
791
677
  model_family: Llama-3.2
@@ -794,14 +680,12 @@ models:
794
680
  gpus_per_node: 4
795
681
  num_nodes: 2
796
682
  vocab_size: 128256
797
- qos: m2
798
683
  time: 08:00:00
799
- partition: a40
684
+ resource_type: l40s
800
685
  vllm_args:
801
686
  --tensor-parallel-size: 8
802
687
  --max-model-len: 4096
803
688
  --max-num-seqs: 32
804
- --compilation-config: 3
805
689
  --enforce-eager: true
806
690
  Qwen2.5-0.5B-Instruct:
807
691
  model_family: Qwen2.5
@@ -810,13 +694,11 @@ models:
810
694
  gpus_per_node: 1
811
695
  num_nodes: 1
812
696
  vocab_size: 152064
813
- qos: m2
814
697
  time: 08:00:00
815
- partition: a40
698
+ resource_type: l40s
816
699
  vllm_args:
817
700
  --max-model-len: 32768
818
701
  --max-num-seqs: 256
819
- --compilation-config: 3
820
702
  Qwen2.5-1.5B-Instruct:
821
703
  model_family: Qwen2.5
822
704
  model_variant: 1.5B-Instruct
@@ -824,13 +706,11 @@ models:
824
706
  gpus_per_node: 1
825
707
  num_nodes: 1
826
708
  vocab_size: 152064
827
- qos: m2
828
709
  time: 08:00:00
829
- partition: a40
710
+ resource_type: l40s
830
711
  vllm_args:
831
712
  --max-model-len: 32768
832
713
  --max-num-seqs: 256
833
- --compilation-config: 3
834
714
  Qwen2.5-3B-Instruct:
835
715
  model_family: Qwen2.5
836
716
  model_variant: 3B-Instruct
@@ -838,13 +718,11 @@ models:
838
718
  gpus_per_node: 1
839
719
  num_nodes: 1
840
720
  vocab_size: 152064
841
- qos: m2
842
721
  time: 08:00:00
843
- partition: a40
722
+ resource_type: l40s
844
723
  vllm_args:
845
724
  --max-model-len: 32768
846
725
  --max-num-seqs: 256
847
- --compilation-config: 3
848
726
  Qwen2.5-7B-Instruct:
849
727
  model_family: Qwen2.5
850
728
  model_variant: 7B-Instruct
@@ -852,13 +730,11 @@ models:
852
730
  gpus_per_node: 1
853
731
  num_nodes: 1
854
732
  vocab_size: 152064
855
- qos: m2
856
733
  time: 08:00:00
857
- partition: a40
734
+ resource_type: l40s
858
735
  vllm_args:
859
736
  --max-model-len: 32768
860
737
  --max-num-seqs: 256
861
- --compilation-config: 3
862
738
  Qwen2.5-14B-Instruct:
863
739
  model_family: Qwen2.5
864
740
  model_variant: 14B-Instruct
@@ -866,13 +742,11 @@ models:
866
742
  gpus_per_node: 1
867
743
  num_nodes: 1
868
744
  vocab_size: 152064
869
- qos: m2
870
745
  time: 08:00:00
871
- partition: a40
746
+ resource_type: l40s
872
747
  vllm_args:
873
748
  --max-model-len: 32768
874
749
  --max-num-seqs: 256
875
- --compilation-config: 3
876
750
  Qwen2.5-32B-Instruct:
877
751
  model_family: Qwen2.5
878
752
  model_variant: 32B-Instruct
@@ -880,14 +754,12 @@ models:
880
754
  gpus_per_node: 2
881
755
  num_nodes: 1
882
756
  vocab_size: 152064
883
- qos: m2
884
757
  time: 08:00:00
885
- partition: a40
758
+ resource_type: l40s
886
759
  vllm_args:
887
760
  --tensor-parallel-size: 2
888
761
  --max-model-len: 32768
889
762
  --max-num-seqs: 256
890
- --compilation-config: 3
891
763
  Qwen2.5-72B-Instruct:
892
764
  model_family: Qwen2.5
893
765
  model_variant: 72B-Instruct
@@ -895,14 +767,12 @@ models:
895
767
  gpus_per_node: 4
896
768
  num_nodes: 1
897
769
  vocab_size: 152064
898
- qos: m2
899
770
  time: 08:00:00
900
- partition: a40
771
+ resource_type: l40s
901
772
  vllm_args:
902
773
  --tensor-parallel-size: 4
903
774
  --max-model-len: 16384
904
775
  --max-num-seqs: 256
905
- --compilation-config: 3
906
776
  Qwen2.5-Math-1.5B-Instruct:
907
777
  model_family: Qwen2.5
908
778
  model_variant: Math-1.5B-Instruct
@@ -910,13 +780,11 @@ models:
910
780
  gpus_per_node: 1
911
781
  num_nodes: 1
912
782
  vocab_size: 152064
913
- qos: m2
914
783
  time: 08:00:00
915
- partition: a40
784
+ resource_type: l40s
916
785
  vllm_args:
917
786
  --max-model-len: 4096
918
787
  --max-num-seqs: 256
919
- --compilation-config: 3
920
788
  Qwen2.5-Math-7B-Instruct:
921
789
  model_family: Qwen2.5
922
790
  model_variant: Math-7B-Instruct
@@ -924,13 +792,11 @@ models:
924
792
  gpus_per_node: 1
925
793
  num_nodes: 1
926
794
  vocab_size: 152064
927
- qos: m2
928
795
  time: 08:00:00
929
- partition: a40
796
+ resource_type: l40s
930
797
  vllm_args:
931
798
  --max-model-len: 4096
932
799
  --max-num-seqs: 256
933
- --compilation-config: 3
934
800
  Qwen2.5-Math-72B-Instruct:
935
801
  model_family: Qwen2.5
936
802
  model_variant: Math-72B-Instruct
@@ -938,14 +804,12 @@ models:
938
804
  gpus_per_node: 4
939
805
  num_nodes: 1
940
806
  vocab_size: 152064
941
- qos: m2
942
807
  time: 08:00:00
943
- partition: a40
808
+ resource_type: l40s
944
809
  vllm_args:
945
810
  --tensor-parallel-size: 4
946
811
  --max-model-len: 4096
947
812
  --max-num-seqs: 256
948
- --compilation-config: 3
949
813
  Qwen2.5-Coder-7B-Instruct:
950
814
  model_family: Qwen2.5
951
815
  model_variant: Coder-7B-Instruct
@@ -953,13 +817,11 @@ models:
953
817
  gpus_per_node: 1
954
818
  num_nodes: 1
955
819
  vocab_size: 152064
956
- qos: m2
957
820
  time: 08:00:00
958
- partition: a40
821
+ resource_type: l40s
959
822
  vllm_args:
960
823
  --max-model-len: 32768
961
824
  --max-num-seqs: 256
962
- --compilation-config: 3
963
825
  Qwen2.5-Math-RM-72B:
964
826
  model_family: Qwen2.5
965
827
  model_variant: Math-RM-72B
@@ -967,14 +829,12 @@ models:
967
829
  gpus_per_node: 4
968
830
  num_nodes: 1
969
831
  vocab_size: 152064
970
- qos: m2
971
832
  time: 08:00:00
972
- partition: a40
833
+ resource_type: l40s
973
834
  vllm_args:
974
835
  --tensor-parallel-size: 4
975
836
  --max-model-len: 4096
976
837
  --max-num-seqs: 256
977
- --compilation-config: 3
978
838
  Qwen2.5-Math-PRM-7B:
979
839
  model_family: Qwen2.5
980
840
  model_variant: Math-PRM-7B
@@ -982,28 +842,24 @@ models:
982
842
  gpus_per_node: 1
983
843
  num_nodes: 1
984
844
  vocab_size: 152064
985
- qos: m2
986
845
  time: 08:00:00
987
- partition: a40
846
+ resource_type: l40s
988
847
  vllm_args:
989
848
  --max-model-len: 4096
990
849
  --max-num-seqs: 256
991
- --compilation-config: 3
992
- QwQ-32B-Preview:
850
+ QwQ-32B:
993
851
  model_family: QwQ
994
- model_variant: 32B-Preview
852
+ model_variant: 32B
995
853
  model_type: LLM
996
854
  gpus_per_node: 2
997
855
  num_nodes: 1
998
856
  vocab_size: 152064
999
- qos: m2
1000
857
  time: 08:00:00
1001
- partition: a40
858
+ resource_type: l40s
1002
859
  vllm_args:
1003
860
  --tensor-parallel-size: 2
1004
861
  --max-model-len: 32768
1005
862
  --max-num-seqs: 256
1006
- --compilation-config: 3
1007
863
  Pixtral-12B-2409:
1008
864
  model_family: Pixtral
1009
865
  model_variant: 12B-2409
@@ -1011,13 +867,11 @@ models:
1011
867
  gpus_per_node: 1
1012
868
  num_nodes: 1
1013
869
  vocab_size: 131072
1014
- qos: m2
1015
870
  time: 08:00:00
1016
- partition: a40
871
+ resource_type: l40s
1017
872
  vllm_args:
1018
873
  --max-model-len: 8192
1019
874
  --max-num-seqs: 256
1020
- --compilation-config: 3
1021
875
  e5-mistral-7b-instruct:
1022
876
  model_family: e5
1023
877
  model_variant: mistral-7b-instruct
@@ -1025,13 +879,11 @@ models:
1025
879
  gpus_per_node: 1
1026
880
  num_nodes: 1
1027
881
  vocab_size: 32000
1028
- qos: m2
1029
882
  time: 08:00:00
1030
- partition: a40
883
+ resource_type: l40s
1031
884
  vllm_args:
1032
885
  --max-model-len: 4096
1033
886
  --max-num-seqs: 256
1034
- --compilation-config: 3
1035
887
  bge-base-en-v1.5:
1036
888
  model_family: bge
1037
889
  model_variant: base-en-v1.5
@@ -1039,13 +891,11 @@ models:
1039
891
  gpus_per_node: 1
1040
892
  num_nodes: 1
1041
893
  vocab_size: 30522
1042
- qos: m2
1043
894
  time: 08:00:00
1044
- partition: a40
895
+ resource_type: l40s
1045
896
  vllm_args:
1046
897
  --max-model-len: 512
1047
898
  --max-num-seqs: 256
1048
- --compilation-config: 3
1049
899
  all-MiniLM-L6-v2:
1050
900
  model_family: all-MiniLM
1051
901
  model_variant: L6-v2
@@ -1053,13 +903,11 @@ models:
1053
903
  gpus_per_node: 1
1054
904
  num_nodes: 1
1055
905
  vocab_size: 30522
1056
- qos: m2
1057
906
  time: 08:00:00
1058
- partition: a40
907
+ resource_type: l40s
1059
908
  vllm_args:
1060
909
  --max-model-len: 512
1061
910
  --max-num-seqs: 256
1062
- --compilation-config: 3
1063
911
  Llama-3.3-70B-Instruct:
1064
912
  model_family: Llama-3.3
1065
913
  model_variant: 70B-Instruct
@@ -1067,14 +915,12 @@ models:
1067
915
  gpus_per_node: 4
1068
916
  num_nodes: 1
1069
917
  vocab_size: 128256
1070
- qos: m2
1071
918
  time: 08:00:00
1072
- partition: a40
919
+ resource_type: l40s
1073
920
  vllm_args:
1074
921
  --tensor-parallel-size: 4
1075
922
  --max-model-len: 65536
1076
923
  --max-num-seqs: 256
1077
- --compilation-config: 3
1078
924
  InternVL2_5-26B:
1079
925
  model_family: InternVL2_5
1080
926
  model_variant: 26B
@@ -1082,14 +928,12 @@ models:
1082
928
  gpus_per_node: 2
1083
929
  num_nodes: 1
1084
930
  vocab_size: 92553
1085
- qos: m2
1086
931
  time: 08:00:00
1087
- partition: a40
932
+ resource_type: l40s
1088
933
  vllm_args:
1089
934
  --tensor-parallel-size: 2
1090
935
  --max-model-len: 32768
1091
936
  --max-num-seqs: 256
1092
- --compilation-config: 3
1093
937
  InternVL2_5-38B:
1094
938
  model_family: InternVL2_5
1095
939
  model_variant: 38B
@@ -1097,14 +941,12 @@ models:
1097
941
  gpus_per_node: 4
1098
942
  num_nodes: 1
1099
943
  vocab_size: 92553
1100
- qos: m2
1101
944
  time: 08:00:00
1102
- partition: a40
945
+ resource_type: l40s
1103
946
  vllm_args:
1104
947
  --tensor-parallel-size: 4
1105
948
  --max-model-len: 32768
1106
949
  --max-num-seqs: 256
1107
- --compilation-config: 3
1108
950
  Aya-Expanse-32B:
1109
951
  model_family: Aya-Expanse
1110
952
  model_variant: 32B
@@ -1112,14 +954,12 @@ models:
1112
954
  gpus_per_node: 2
1113
955
  num_nodes: 1
1114
956
  vocab_size: 256000
1115
- qos: m2
1116
957
  time: 08:00:00
1117
- partition: a40
958
+ resource_type: l40s
1118
959
  vllm_args:
1119
960
  --tensor-parallel-size: 2
1120
961
  --max-model-len: 8192
1121
962
  --max-num-seqs: 256
1122
- --compilation-config: 3
1123
963
  DeepSeek-R1-Distill-Llama-70B:
1124
964
  model_family: DeepSeek-R1
1125
965
  model_variant: Distill-Llama-70B
@@ -1127,14 +967,12 @@ models:
1127
967
  gpus_per_node: 4
1128
968
  num_nodes: 1
1129
969
  vocab_size: 128256
1130
- qos: m2
1131
970
  time: 08:00:00
1132
- partition: a40
971
+ resource_type: l40s
1133
972
  vllm_args:
1134
973
  --tensor-parallel-size: 4
1135
974
  --max-model-len: 65536
1136
975
  --max-num-seqs: 256
1137
- --compilation-config: 3
1138
976
  DeepSeek-R1-Distill-Llama-8B:
1139
977
  model_family: DeepSeek-R1
1140
978
  model_variant: Distill-Llama-8B
@@ -1142,13 +980,11 @@ models:
1142
980
  gpus_per_node: 1
1143
981
  num_nodes: 1
1144
982
  vocab_size: 128256
1145
- qos: m2
1146
983
  time: 08:00:00
1147
- partition: a40
984
+ resource_type: l40s
1148
985
  vllm_args:
1149
986
  --max-model-len: 131072
1150
987
  --max-num-seqs: 256
1151
- --compilation-config: 3
1152
988
  DeepSeek-R1-Distill-Qwen-32B:
1153
989
  model_family: DeepSeek-R1
1154
990
  model_variant: Distill-Qwen-32B
@@ -1156,14 +992,12 @@ models:
1156
992
  gpus_per_node: 2
1157
993
  num_nodes: 1
1158
994
  vocab_size: 152064
1159
- qos: m2
1160
995
  time: 08:00:00
1161
- partition: a40
996
+ resource_type: l40s
1162
997
  vllm_args:
1163
998
  --tensor-parallel-size: 2
1164
999
  --max-model-len: 65536
1165
1000
  --max-num-seqs: 256
1166
- --compilation-config: 3
1167
1001
  DeepSeek-R1-Distill-Qwen-14B:
1168
1002
  model_family: DeepSeek-R1
1169
1003
  model_variant: Distill-Qwen-14B
@@ -1171,13 +1005,11 @@ models:
1171
1005
  gpus_per_node: 1
1172
1006
  num_nodes: 1
1173
1007
  vocab_size: 152064
1174
- qos: m2
1175
1008
  time: 08:00:00
1176
- partition: a40
1009
+ resource_type: l40s
1177
1010
  vllm_args:
1178
1011
  --max-model-len: 65536
1179
1012
  --max-num-seqs: 256
1180
- --compilation-config: 3
1181
1013
  DeepSeek-R1-Distill-Qwen-7B:
1182
1014
  model_family: DeepSeek-R1
1183
1015
  model_variant: Distill-Qwen-7B
@@ -1185,13 +1017,11 @@ models:
1185
1017
  gpus_per_node: 1
1186
1018
  num_nodes: 1
1187
1019
  vocab_size: 152064
1188
- qos: m2
1189
1020
  time: 08:00:00
1190
- partition: a40
1021
+ resource_type: l40s
1191
1022
  vllm_args:
1192
1023
  --max-model-len: 131072
1193
1024
  --max-num-seqs: 256
1194
- --compilation-config: 3
1195
1025
  DeepSeek-R1-Distill-Qwen-1.5B:
1196
1026
  model_family: DeepSeek-R1
1197
1027
  model_variant: Distill-Qwen-1.5B
@@ -1199,13 +1029,11 @@ models:
1199
1029
  gpus_per_node: 1
1200
1030
  num_nodes: 1
1201
1031
  vocab_size: 152064
1202
- qos: m2
1203
1032
  time: 08:00:00
1204
- partition: a40
1033
+ resource_type: l40s
1205
1034
  vllm_args:
1206
1035
  --max-model-len: 131072
1207
1036
  --max-num-seqs: 256
1208
- --compilation-config: 3
1209
1037
  Phi-3.5-vision-instruct:
1210
1038
  model_family: Phi-3.5-vision
1211
1039
  model_variant: instruct
@@ -1213,14 +1041,12 @@ models:
1213
1041
  gpus_per_node: 2
1214
1042
  num_nodes: 1
1215
1043
  vocab_size: 32064
1216
- qos: m2
1217
1044
  time: 08:00:00
1218
- partition: a40
1045
+ resource_type: l40s
1219
1046
  vllm_args:
1220
1047
  --tensor-parallel-size: 2
1221
1048
  --max-model-len: 65536
1222
1049
  --max-num-seqs: 256
1223
- --compilation-config: 3
1224
1050
  InternVL2_5-8B:
1225
1051
  model_family: InternVL2_5
1226
1052
  model_variant: 8B
@@ -1228,13 +1054,11 @@ models:
1228
1054
  gpus_per_node: 1
1229
1055
  num_nodes: 1
1230
1056
  vocab_size: 92553
1231
- qos: m2
1232
1057
  time: 08:00:00
1233
- partition: a40
1058
+ resource_type: l40s
1234
1059
  vllm_args:
1235
1060
  --max-model-len: 32768
1236
1061
  --max-num-seqs: 256
1237
- --compilation-config: 3
1238
1062
  glm-4v-9b:
1239
1063
  model_family: glm-4v
1240
1064
  model_variant: 9b
@@ -1242,13 +1066,11 @@ models:
1242
1066
  gpus_per_node: 1
1243
1067
  num_nodes: 1
1244
1068
  vocab_size: 151552
1245
- qos: m2
1246
1069
  time: 08:00:00
1247
- partition: a40
1070
+ resource_type: l40s
1248
1071
  vllm_args:
1249
1072
  --max-model-len: 8192
1250
1073
  --max-num-seqs: 256
1251
- --compilation-config: 3
1252
1074
  Molmo-7B-D-0924:
1253
1075
  model_family: Molmo
1254
1076
  model_variant: 7B-D-0924
@@ -1256,27 +1078,23 @@ models:
1256
1078
  gpus_per_node: 1
1257
1079
  num_nodes: 1
1258
1080
  vocab_size: 152064
1259
- qos: m2
1260
1081
  time: 08:00:00
1261
- partition: a40
1082
+ resource_type: l40s
1262
1083
  vllm_args:
1263
1084
  --max-model-len: 4096
1264
1085
  --max-num-seqs: 256
1265
- --compilation-config: 3
1266
1086
  deepseek-vl2:
1267
1087
  model_family: deepseek-vl2
1268
1088
  model_type: VLM
1269
1089
  gpus_per_node: 2
1270
1090
  num_nodes: 1
1271
1091
  vocab_size: 129280
1272
- qos: m2
1273
1092
  time: 08:00:00
1274
- partition: a40
1093
+ resource_type: l40s
1275
1094
  vllm_args:
1276
1095
  --tensor-parallel-size: 2
1277
1096
  --max-model-len: 4096
1278
1097
  --max-num-seqs: 256
1279
- --compilation-config: 3
1280
1098
  deepseek-vl2-small:
1281
1099
  model_family: deepseek-vl2
1282
1100
  model_variant: small
@@ -1284,10 +1102,20 @@ models:
1284
1102
  gpus_per_node: 1
1285
1103
  num_nodes: 1
1286
1104
  vocab_size: 129280
1287
- qos: m2
1288
1105
  time: 08:00:00
1289
- partition: a40
1106
+ resource_type: l40s
1290
1107
  vllm_args:
1291
1108
  --max-model-len: 4096
1292
1109
  --max-num-seqs: 256
1293
- --compilation-config: 3
1110
+ Qwen3-14B:
1111
+ model_family: Qwen3
1112
+ model_variant: 14B
1113
+ model_type: LLM
1114
+ gpus_per_node: 1
1115
+ num_nodes: 1
1116
+ vocab_size: 151936
1117
+ time: 08:00:00
1118
+ resource_type: l40s
1119
+ vllm_args:
1120
+ --max-model-len: 40960
1121
+ --max-num-seqs: 256