-
Notifications
You must be signed in to change notification settings - Fork 0
/
worker-config.json
1032 lines (1032 loc) · 38.8 KB
/
worker-config.json
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
{
"0.5.3": {
"categories": [
{
"title": "LLM Settings",
"settings": [
{
"TOKENIZER": {
"value": "",
"title": "Tokenizer",
"description": "Name or path of the Hugging Face tokenizer to use.",
"required": false,
"type": "text",
"category": "LLM Settings"
},
"TOKENIZER_MODE": {
"value": "auto",
"title": "Tokenizer Mode",
"description": "The tokenizer mode.",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "auto", "label": "auto" },
{ "value": "slow", "label": "slow" }
]
},
"SKIP_TOKENIZER_INIT": {
"value": false,
"title": "Skip Tokenizer Init",
"description": "Skip initialization of tokenizer and detokenizer.",
"required": false,
"type": "toggle",
"category": "LLM Settings"
},
"TRUST_REMOTE_CODE": {
"value": false,
"title": "Trust Remote Code",
"description": "Trust remote code from Hugging Face.",
"required": false,
"type": "toggle",
"category": "LLM Settings"
},
"DOWNLOAD_DIR": {
"value": "",
"title": "Download Directory",
"description": "Directory to download and load the weights.",
"required": false,
"type": "text",
"category": "LLM Settings"
},
"LOAD_FORMAT": {
"value": "auto",
"title": "Load Format",
"description": "The format of the model weights to load.",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "auto", "label": "auto" },
{ "value": "pt", "label": "pt" },
{ "value": "safetensors", "label": "safetensors" },
{ "value": "npcache", "label": "npcache" },
{ "value": "dummy", "label": "dummy" },
{ "value": "tensorizer", "label": "tensorizer" },
{ "value": "bitsandbytes", "label": "bitsandbytes" }
]
},
"DTYPE": {
"value": "auto",
"title": "Data Type",
"description": "Data type for model weights and activations.",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "auto", "label": "auto" },
{ "value": "half", "label": "half" },
{ "value": "float16", "label": "float16" },
{ "value": "bfloat16", "label": "bfloat16" },
{ "value": "float", "label": "float" },
{ "value": "float32", "label": "float32" }
]
},
"KV_CACHE_DTYPE": {
"value": "auto",
"title": "KV Cache Data Type",
"description": "Data type for KV cache storage.",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "auto", "label": "auto" },
{ "value": "fp8", "label": "fp8" }
]
},
"QUANTIZATION_PARAM_PATH": {
"value": "",
"title": "Quantization Param Path",
"description": "Path to the JSON file containing the KV cache scaling factors.",
"required": false,
"type": "text",
"category": "LLM Settings"
},
"MAX_MODEL_LEN": {
"value": "",
"title": "Max Model Length",
"description": "Model context length.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"GUIDED_DECODING_BACKEND": {
"value": "outlines",
"title": "Guided Decoding Backend",
"description": "Which engine will be used for guided decoding by default.",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "outlines", "label": "outlines" },
{ "value": "lm-format-enforcer", "label": "lm-format-enforcer" }
]
},
"DISTRIBUTED_EXECUTOR_BACKEND": {
"value": "",
"title": "Distributed Executor Backend",
"description": "Backend to use for distributed serving.",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "ray", "label": "ray" },
{ "value": "mp", "label": "mp" }
]
},
"WORKER_USE_RAY": {
"value": false,
"title": "Worker Use Ray",
"description": "Deprecated, use --distributed-executor-backend=ray.",
"required": false,
"type": "toggle",
"category": "LLM Settings"
},
"RAY_WORKERS_USE_NSIGHT": {
"value": false,
"title": "Ray Workers Use Nsight",
"description": "If specified, use nsight to profile Ray workers.",
"required": false,
"type": "toggle",
"category": "LLM Settings"
},
"PIPELINE_PARALLEL_SIZE": {
"value": 1,
"title": "Pipeline Parallel Size",
"description": "Number of pipeline stages.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"TENSOR_PARALLEL_SIZE": {
"value": 1,
"title": "Tensor Parallel Size",
"description": "Number of tensor parallel replicas.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"MAX_PARALLEL_LOADING_WORKERS": {
"value": "",
"title": "Max Parallel Loading Workers",
"description": "Load model sequentially in multiple batches.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"ENABLE_PREFIX_CACHING": {
"value": false,
"title": "Enable Prefix Caching",
"description": "Enables automatic prefix caching.",
"required": false,
"type": "toggle",
"category": "LLM Settings"
},
"DISABLE_SLIDING_WINDOW": {
"value": false,
"title": "Disable Sliding Window",
"description": "Disables sliding window, capping to sliding window size.",
"required": false,
"type": "toggle",
"category": "LLM Settings"
},
"USE_V2_BLOCK_MANAGER": {
"value": false,
"title": "Use V2 Block Manager",
"description": "Use BlockSpaceMangerV2.",
"required": false,
"type": "toggle",
"category": "LLM Settings"
},
"NUM_LOOKAHEAD_SLOTS": {
"value": 0,
"title": "Num Lookahead Slots",
"description": "Experimental scheduling config necessary for speculative decoding.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"SEED": {
"value": 0,
"title": "Seed",
"description": "Random seed for operations.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"NUM_GPU_BLOCKS_OVERRIDE": {
"value": "",
"title": "Num GPU Blocks Override",
"description": "If specified, ignore GPU profiling result and use this number of GPU blocks.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"MAX_NUM_BATCHED_TOKENS": {
"value": "",
"title": "Max Num Batched Tokens",
"description": "Maximum number of batched tokens per iteration.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"MAX_NUM_SEQS": {
"value": 256,
"title": "Max Num Seqs",
"description": "Maximum number of sequences per iteration.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"MAX_LOGPROBS": {
"value": 20,
"title": "Max Logprobs",
"description": "Max number of log probs to return when logprobs is specified in SamplingParams.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"DISABLE_LOG_STATS": {
"value": false,
"title": "Disable Log Stats",
"description": "Disable logging statistics.",
"required": false,
"type": "toggle",
"category": "LLM Settings"
},
"QUANTIZATION": {
"value": "",
"title": "Quantization",
"description": "Method used to quantize the weights.",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "None", "label": "None" },
{ "value": "awq", "label": "AWQ" },
{ "value": "squeezellm", "label": "SqueezeLLM" },
{ "value": "gptq", "label": "GPTQ" }
]
},
"ROPE_SCALING": {
"value": "",
"title": "RoPE Scaling",
"description": "RoPE scaling configuration in JSON format.",
"required": false,
"type": "text",
"category": "LLM Settings"
},
"ROPE_THETA": {
"value": "",
"title": "RoPE Theta",
"description": "RoPE theta. Use with rope_scaling.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"TOKENIZER_POOL_SIZE": {
"value": 0,
"title": "Tokenizer Pool Size",
"description": "Size of tokenizer pool to use for asynchronous tokenization.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"TOKENIZER_POOL_TYPE": {
"value": "ray",
"title": "Tokenizer Pool Type",
"description": "Type of tokenizer pool to use for asynchronous tokenization.",
"required": false,
"type": "text",
"category": "LLM Settings"
},
"TOKENIZER_POOL_EXTRA_CONFIG": {
"value": "",
"title": "Tokenizer Pool Extra Config",
"description": "Extra config for tokenizer pool.",
"required": false,
"type": "text",
"category": "LLM Settings"
},
"ENABLE_LORA": {
"value": false,
"title": "Enable LoRA",
"description": "If True, enable handling of LoRA adapters.",
"required": false,
"type": "toggle",
"category": "LLM Settings"
},
"MAX_LORAS": {
"value": 1,
"title": "Max LoRAs",
"description": "Max number of LoRAs in a single batch.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"MAX_LORA_RANK": {
"value": 16,
"title": "Max LoRA Rank",
"description": "Max LoRA rank.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"LORA_EXTRA_VOCAB_SIZE": {
"value": 256,
"title": "LoRA Extra Vocab Size",
"description": "Maximum size of extra vocabulary for LoRA adapters.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"LORA_DTYPE": {
"value": "auto",
"title": "LoRA Data Type",
"description": "Data type for LoRA.",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "auto", "label": "auto" },
{ "value": "float16", "label": "float16" },
{ "value": "bfloat16", "label": "bfloat16" },
{ "value": "float32", "label": "float32" }
]
},
"LONG_LORA_SCALING_FACTORS": {
"value": "",
"title": "Long LoRA Scaling Factors",
"description": "Specify multiple scaling factors for LoRA adapters.",
"required": false,
"type": "text",
"category": "LLM Settings"
},
"MAX_CPU_LORAS": {
"value": "",
"title": "Max CPU LoRAs",
"description": "Maximum number of LoRAs to store in CPU memory.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"FULLY_SHARDED_LORAS": {
"value": false,
"title": "Fully Sharded LoRAs",
"description": "Enable fully sharded LoRA layers.",
"required": false,
"type": "toggle",
"category": "LLM Settings"
},
"DEVICE": {
"value": "auto",
"title": "Device",
"description": "Device type for vLLM execution.",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "auto", "label": "auto" },
{ "value": "cuda", "label": "cuda" },
{ "value": "neuron", "label": "neuron" },
{ "value": "cpu", "label": "cpu" },
{ "value": "openvino", "label": "openvino" },
{ "value": "tpu", "label": "tpu" },
{ "value": "xpu", "label": "xpu" }
]
},
"SCHEDULER_DELAY_FACTOR": {
"value": 0.0,
"title": "Scheduler Delay Factor",
"description": "Apply a delay before scheduling next prompt.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"ENABLE_CHUNKED_PREFILL": {
"value": false,
"title": "Enable Chunked Prefill",
"description": "Enable chunked prefill requests.",
"required": false,
"type": "toggle",
"category": "LLM Settings"
},
"SPECULATIVE_MODEL": {
"value": "",
"title": "Speculative Model",
"description": "The name of the draft model to be used in speculative decoding.",
"required": false,
"type": "text",
"category": "LLM Settings"
},
"NUM_SPECULATIVE_TOKENS": {
"value": "",
"title": "Num Speculative Tokens",
"description": "The number of speculative tokens to sample from the draft model.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"SPECULATIVE_DRAFT_TENSOR_PARALLEL_SIZE": {
"value": "",
"title": "Speculative Draft Tensor Parallel Size",
"description": "Number of tensor parallel replicas for the draft model.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"SPECULATIVE_MAX_MODEL_LEN": {
"value": "",
"title": "Speculative Max Model Length",
"description": "The maximum sequence length supported by the draft model.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"SPECULATIVE_DISABLE_BY_BATCH_SIZE": {
"value": "",
"title": "Speculative Disable by Batch Size",
"description": "Disable speculative decoding if the number of enqueue requests is larger than this value.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"NGRAM_PROMPT_LOOKUP_MAX": {
"value": "",
"title": "Ngram Prompt Lookup Max",
"description": "Max size of window for ngram prompt lookup in speculative decoding.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"NGRAM_PROMPT_LOOKUP_MIN": {
"value": "",
"title": "Ngram Prompt Lookup Min",
"description": "Min size of window for ngram prompt lookup in speculative decoding.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"SPEC_DECODING_ACCEPTANCE_METHOD": {
"value": "rejection_sampler",
"title": "Speculative Decoding Acceptance Method",
"description": "Specify the acceptance method for draft token verification in speculative decoding.",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "rejection_sampler", "label": "rejection_sampler" },
{ "value": "typical_acceptance_sampler", "label": "typical_acceptance_sampler" }
]
},
"TYPICAL_ACCEPTANCE_SAMPLER_POSTERIOR_THRESHOLD": {
"value": "",
"title": "Typical Acceptance Sampler Posterior Threshold",
"description": "Set the lower bound threshold for the posterior probability of a token to be accepted.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"TYPICAL_ACCEPTANCE_SAMPLER_POSTERIOR_ALPHA": {
"value": "",
"title": "Typical Acceptance Sampler Posterior Alpha",
"description": "A scaling factor for the entropy-based threshold for token acceptance.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"MODEL_LOADER_EXTRA_CONFIG": {
"value": "",
"title": "Model Loader Extra Config",
"description": "Extra config for model loader.",
"required": false,
"type": "text",
"category": "LLM Settings"
},
"PREEMPTION_MODE": {
"value": "",
"title": "Preemption Mode",
"description": "If 'recompute', the engine performs preemption-aware recomputation. If 'save', the engine saves activations into the CPU memory as preemption happens.",
"required": false,
"type": "text",
"category": "LLM Settings"
},
"PREEMPTION_CHECK_PERIOD": {
"value": 1.0,
"title": "Preemption Check Period",
"description": "How frequently the engine checks if a preemption happens.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"PREEMPTION_CPU_CAPACITY": {
"value": 2,
"title": "Preemption CPU Capacity",
"description": "The percentage of CPU memory used for the saved activations.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"MAX_LOG_LEN": {
"value": "",
"title": "Max Log Length",
"description": "Max number of characters or ID numbers being printed in log.",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"DISABLE_LOGGING_REQUEST": {
"value": false,
"title": "Disable Logging Request",
"description": "Disable logging requests.",
"required": false,
"type": "toggle",
"category": "LLM Settings"
}
}
]
},
{
"title": "Tokenizer Settings",
"settings": [
{
"TOKENIZER_NAME": {
"value": "",
"title": "Tokenizer Name",
"description": "Tokenizer repo to use a different tokenizer than the model's default",
"required": false,
"type": "text",
"category": "Tokenizer Settings"
},
"TOKENIZER_REVISION": {
"value": "",
"title": "Tokenizer Revision",
"description": "Tokenizer revision to load",
"required": false,
"type": "text",
"category": "Tokenizer Settings"
},
"CUSTOM_CHAT_TEMPLATE": {
"value": "",
"title": "Custom Chat Template",
"description": "Custom chat jinja template",
"required": false,
"type": "text",
"category": "Tokenizer Settings"
}
}
]
},
{
"title": "System Settings",
"settings": [
{
"GPU_MEMORY_UTILIZATION": {
"value": "0.95",
"title": "GPU Memory Utilization",
"description": "Sets GPU VRAM utilization",
"required": false,
"type": "number",
"category": "System Settings"
},
"MAX_PARALLEL_LOADING_WORKERS": {
"value": "",
"title": "Max Parallel Loading Workers",
"description": "Load model sequentially in multiple batches. Leave empty for auto",
"required": false,
"type": "number",
"category": "System Settings"
},
"BLOCK_SIZE": {
"value": "16",
"title": "Block Size",
"description": "Token block size for contiguous chunks of tokens",
"required": false,
"type": "number",
"category": "System Settings"
},
"SWAP_SPACE": {
"value": "4",
"title": "Swap Space",
"description": "CPU swap space size (GiB) per GPU",
"required": false,
"type": "number",
"category": "System Settings"
},
"ENFORCE_EAGER": {
"value": false,
"title": "Enforce Eager",
"description": "Always use eager-mode PyTorch. If False (0), will use eager mode and CUDA graph in hybrid for maximal performance and flexibility",
"required": false,
"type": "toggle",
"category": "System Settings"
},
"MAX_SEQ_LEN_TO_CAPTURE": {
"value": "8192",
"title": "CUDA Graph Max Content Length",
"description": "Maximum context length covered by CUDA graphs. If a sequence has context length larger than this, we fall back to eager mode",
"required": false,
"type": "number",
"category": "System Settings"
},
"DISABLE_CUSTOM_ALL_REDUCE": {
"value": false,
"title": "Disable Custom All Reduce",
"description": "Enables or disables custom all reduce",
"required": false,
"type": "toggle",
"category": "System Settings"
}
}
]
},
{
"title": "Streaming Settings",
"settings": [
{
"DEFAULT_BATCH_SIZE": {
"value": "50",
"title": "Default Final Batch Size",
"description": "Default and Maximum batch size for token streaming to reduce HTTP calls",
"required": false,
"type": "number",
"category": "Streaming Settings"
},
"DEFAULT_MIN_BATCH_SIZE": {
"value": "1",
"title": "Default Starting Batch Size",
"description": "Batch size for the first request, which will be multiplied by the growth factor every subsequent request",
"required": false,
"type": "number",
"category": "Streaming Settings"
},
"DEFAULT_BATCH_SIZE_GROWTH_FACTOR": {
"value": "3",
"title": "Default Batch Size Growth Factor",
"description": "Growth factor for dynamic batch size",
"required": false,
"type": "number",
"category": "Streaming Settings"
}
}
]
},
{
"title": "OpenAI Settings",
"settings": [
{
"RAW_OPENAI_OUTPUT": {
"value": true,
"title": "Raw OpenAI Output",
"description": "Raw OpenAI output instead of just the text",
"required": false,
"type": "toggle",
"category": "OpenAI Settings"
},
"OPENAI_RESPONSE_ROLE": {
"value": "assistant",
"title": "OpenAI Response Role",
"description": "Role of the LLM's Response in OpenAI Chat Completions",
"required": false,
"type": "text",
"category": "OpenAI Settings"
},
"OPENAI_SERVED_MODEL_NAME_OVERRIDE": {
"value": "",
"title": "OpenAI Served Model Name Override",
"description": "Overrides the name of the served model from model repo/path to specified name, which you will then be able to use the value for the `model` parameter when making OpenAI requests",
"required": false,
"type": "text",
"category": "OpenAI Settings"
}
}
]
},
{
"title": "Serverless Settings",
"settings": [
{
"MAX_CONCURRENCY": {
"value": "300",
"title": "Max Concurrency",
"description": "Max concurrent requests per worker. vLLM has an internal queue, so you don't have to worry about limiting by VRAM, this is for improving scaling/load balancing efficiency",
"required": false,
"type": "number",
"category": "Serverless Settings"
},
"DISABLE_LOG_STATS": {
"value": true,
"title": "Disable Log Stats",
"description": "Enables or disables vLLM stats logging",
"required": false,
"type": "toggle",
"category": "Serverless Settings"
},
"DISABLE_LOG_REQUESTS": {
"value": true,
"title": "Disable Log Requests",
"description": "Enables or disables vLLM request logging",
"required": false,
"type": "toggle",
"category": "Serverless Settings"
}
}
]
}
]
},
"0.4.2": {
"categories": [
{
"title": "LLM Settings",
"settings": [
{
"MODEL_REVISION": {
"value": "",
"title": "Model Revision",
"description": "Model revision (branch) to load",
"required": false,
"type": "text",
"category": "LLM Settings"
},
"MAX_MODEL_LEN": {
"value": "",
"title": "Max Model Length",
"description": "Maximum number of tokens for the engine to handle per request",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"BASE_PATH": {
"value": "/runpod-volume",
"title": "Base Path",
"description": "Storage directory for Huggingface cache and model",
"required": false,
"type": "text",
"category": "LLM Settings"
},
"LOAD_FORMAT": {
"value": "auto",
"title": "Load Format",
"description": "Format to load model in",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "auto", "label": "auto" },
{ "value": ".safetensors", "label": ".safetensors" },
{ "value": ".bin", "label": ".bin" },
{ "value": ".pt", "label": ".pt" }
]
},
"QUANTIZATION": {
"value": "None",
"title": "Quantization",
"description": "Quantization of given model. The model must already be quantized",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "None", "label": "None" },
{ "value": "awq", "label": "AWQ" },
{ "value": "squeezellm", "label": "SqueezeLLM" },
{ "value": "gptq", "label": "GPTQ" }
]
},
"TRUST_REMOTE_CODE": {
"value": "0",
"title": "Trust Remote Code",
"description": "Trust remote code for HuggingFace models",
"required": false,
"type": "toggle",
"category": "LLM Settings"
},
"SEED": {
"value": "",
"title": "Seed",
"description": "Sets random seed for operations",
"required": false,
"type": "number",
"category": "LLM Settings"
},
"KV_CACHE_DTYPE": {
"value": "auto",
"title": "KV Cache Data Type",
"description": "Data type for kv cache storage. Uses DTYPE if set to auto",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "auto", "label": "auto" },
{ "value": "fp8_e5m2", "label": "fp8_e5m2" }
]
},
"DTYPE": {
"value": "auto",
"title": "Weights Datatype/Precision",
"description": "Sets datatype/precision for model weights and activations",
"required": false,
"type": "select",
"category": "LLM Settings",
"options": [
{ "value": "auto", "label": "auto" },
{ "value": "half", "label": "half" },
{ "value": "float16", "label": "float16" },
{ "value": "bfloat16", "label": "bfloat16" },
{ "value": "float", "label": "float" },
{ "value": "float32", "label": "float32" }
]
}
}
]
},
{
"title": "Tokenizer Settings",
"settings": [
{
"TOKENIZER_NAME": {
"value": "",
"title": "Tokenizer Name",
"description": "Tokenizer repo to use a different tokenizer than the model's default",
"required": false,
"type": "text",
"category": "Tokenizer Settings"
},
"TOKENIZER_REVISION": {
"value": "",
"title": "Tokenizer Revision",
"description": "Tokenizer revision to load",
"required": false,
"type": "text",
"category": "Tokenizer Settings"
},
"CUSTOM_CHAT_TEMPLATE": {
"value": "",
"title": "Custom Chat Template",
"description": "Custom chat jinja template",
"required": false,
"type": "text",
"category": "Tokenizer Settings"
}
}
]
},
{
"title": "System Settings",
"settings": [
{
"GPU_MEMORY_UTILIZATION": {
"value": "0.95",
"title": "GPU Memory Utilization",
"description": "Sets GPU VRAM utilization",
"required": false,
"type": "number",
"category": "System Settings"
},
"MAX_PARALLEL_LOADING_WORKERS": {
"value": "",
"title": "Max Parallel Loading Workers",
"description": "Load model sequentially in multiple batches. Leave empty for auto",
"required": false,
"type": "number",
"category": "System Settings"
},
"BLOCK_SIZE": {
"value": "16",
"title": "Block Size",
"description": "Token block size for contiguous chunks of tokens",
"required": false,
"type": "number",
"category": "System Settings"
},
"SWAP_SPACE": {
"value": "4",
"title": "Swap Space",
"description": "CPU swap space size (GiB) per GPU",
"required": false,
"type": "number",
"category": "System Settings"
},
"ENFORCE_EAGER": {
"value": "0",
"title": "Enforce Eager",
"description": "Always use eager-mode PyTorch. If False (0), will use eager mode and CUDA graph in hybrid for maximal performance and flexibility",
"required": false,
"type": "toggle",
"category": "System Settings"
},
"MAX_SEQ_LEN_TO_CAPTURE": {
"value": "8192",
"title": "CUDA Graph Max Content Length",
"description": "Maximum context length covered by CUDA graphs. If a sequence has context length larger than this, we fall back to eager mode",
"required": false,
"type": "number",
"category": "System Settings"
},
"DISABLE_CUSTOM_ALL_REDUCE": {
"value": "0",
"title": "Disable Custom All Reduce",
"description": "Enables or disables custom all reduce",
"required": false,
"type": "toggle",
"category": "System Settings"
}
}
]
},
{
"title": "Streaming Settings",
"settings": [
{
"DEFAULT_BATCH_SIZE": {
"value": "50",
"title": "Default Final Batch Size",
"description": "Default and Maximum batch size for token streaming to reduce HTTP calls",
"required": false,
"type": "number",
"category": "Streaming Settings"
},
"DEFAULT_MIN_BATCH_SIZE": {
"value": "1",
"title": "Default Starting Batch Size",
"description": "Batch size for the first request, which will be multiplied by the growth factor every subsequent request",
"required": false,
"type": "number",
"category": "Streaming Settings"
},
"DEFAULT_BATCH_SIZE_GROWTH_FACTOR": {
"value": "3",
"title": "Default Batch Size Growth Factor",
"description": "Growth factor for dynamic batch size",
"required": false,
"type": "number",
"category": "Streaming Settings"
}
}
]
},
{
"title": "OpenAI Settings",
"settings": [
{
"RAW_OPENAI_OUTPUT": {
"value": "1",
"title": "Raw OpenAI Output",
"description": "Raw OpenAI output instead of just the text",
"required": false,
"type": "toggle",
"category": "OpenAI Settings"
},
"OPENAI_RESPONSE_ROLE": {
"value": "assistant",
"title": "OpenAI Response Role",
"description": "Role of the LLM's Response in OpenAI Chat Completions",
"required": false,
"type": "text",
"category": "OpenAI Settings"
},
"OPENAI_SERVED_MODEL_NAME_OVERRIDE": {
"value": "",
"title": "OpenAI Served Model Name Override",
"description": "Overrides the name of the served model from model repo/path to specified name, which you will then be able to use the value for the `model` parameter when making OpenAI requests",
"required": false,
"type": "text",
"category": "OpenAI Settings"
}
}
]
},
{
"title": "Serverless Settings",