Commits · 1cb4da5c5f1fbaafa5c48b052b1f05abedd97fe5 · change / sglang

25 Aug, 2024 1 commit
- [Fix] the issue of random order when input is a list (#1199) · 1cb4da5c
  Ying Sheng authored Aug 24, 2024
  
  1cb4da5c
24 Aug, 2024 1 commit
- Cleanup readme, llava examples, usage examples and nccl init (#1194) · f6af3a65
  Lianmin Zheng authored Aug 24, 2024
  
  f6af3a65
23 Aug, 2024 1 commit

[Feat/WIP] add llava-onevision, with support for (1) siglip encoder, (2) qwen2... · a5b14ad0

Kaichen Zhang - NTU authored Aug 24, 2024


[Feat/WIP] add llava-onevision, with support for (1) siglip encoder, (2) qwen2 decoder (3) openai api compatible server. (#1123)
Co-authored-by: Bo Li <drluodian@gmail.com>

a5b14ad0

20 Aug, 2024 2 commits
- [Feat] Support update weights without restart server (#1157) · cd10654e
  Shan Yu authored Aug 20, 2024
  
  cd10654e
- feat: allow streaming for multi-prompt and/or parallel sampling (#1134) · d8476818
  Juwan Yoo authored Aug 20, 2024
  
  d8476818
19 Aug, 2024 1 commit

[Feat]Add support for optional start len of logprobs (#1035) · b997a18d

yichuan~ authored Aug 18, 2024


Co-authored-by: Ying Sheng <sqy1415@gmail.com>
Co-authored-by: Yineng Zhang <me@zhyncs.com>
Co-authored-by: Lianmin Zheng <lianminzheng@gmail.com>
Co-authored-by: Liangsheng Yin <hnyls2002@gmail.com>

b997a18d

17 Aug, 2024 1 commit
- Fix CI accuracy && time out limit (#1133) · 5d0d40d0
  Liangsheng Yin authored Aug 16, 2024
  
  5d0d40d0
16 Aug, 2024 1 commit
- Mixed style of chunked prefill (#1013) · 3694f8f9
  Liangsheng Yin authored Aug 16, 2024
  
  3694f8f9
15 Aug, 2024 2 commits
- Enable chunked prefill by default (#1040) · e86b1ccb
  Lianmin Zheng authored Aug 14, 2024
  
  e86b1ccb
- Support `stop_token_ids` in sglang API (#1092) · 73cf6834
  Liangsheng Yin authored Aug 14, 2024
  
  73cf6834
14 Aug, 2024 3 commits
- Use `dtype` to control generate (#1082) · a34dd86a
  Liangsheng Yin authored Aug 14, 2024
```
Co-authored-by: zhyncs <me@zhyncs.com>
```
  a34dd86a
- ci: update timeout and retry (#1086) · c8423ca3
  Yineng Zhang authored Aug 14, 2024
```
Co-authored-by: Liangsheng Yin <hnyls2002@gmail.com>
```
  c8423ca3
- [Feat] Add window attention for gemma-2 (#1056) · 0909bb0d
  Ying Sheng authored Aug 13, 2024
  
  0909bb0d
13 Aug, 2024 3 commits
- Update the mixtral to use the better FusedMoE layer (#1081) · ad3e4f16
  Lianmin Zheng authored Aug 13, 2024
  
  ad3e4f16
- ci: add accuracy timeout (#1078) · cebd78d8
  Yineng Zhang authored Aug 13, 2024
  
  cebd78d8
- ci: add moe test (#1053) · f7fb68d2
  Yineng Zhang authored Aug 13, 2024
  
  f7fb68d2
12 Aug, 2024 6 commits
- Re-organize CI tests (#1052) · c877292c
  Lianmin Zheng authored Aug 12, 2024
  
  c877292c
- Fix accuracy test (#1051) · 0c1c72a0
  Lianmin Zheng authored Aug 12, 2024
  
  0c1c72a0
- Add longer accuracy test on CI (#1049) · 41598e0d
  Lianmin Zheng authored Aug 12, 2024
  
  41598e0d
- fix: Fix returned prefill logits and add output str test (#1046) · 32f61443
  Ying Sheng authored Aug 11, 2024
  
  32f61443
- Delete the useless test/srt/test_throughput.py (#1045) · 14b64930
  Lianmin Zheng authored Aug 11, 2024
  
  14b64930
- Improve end-to-end throughput test and its coverage (#1039) · 82076370
  Lianmin Zheng authored Aug 11, 2024
  
  82076370
11 Aug, 2024 1 commit
- Test the case when max_new_tokens is very large (#1038) · d84c5e70
  Lianmin Zheng authored Aug 11, 2024
  
  d84c5e70
10 Aug, 2024 3 commits
- Clean up unit tests (#1020) · 54fb1c80
  Lianmin Zheng authored Aug 10, 2024
  
  54fb1c80
- fix: force max new tokens to be 1 for embedding request (#1019) · b68c4c07
  Ying Sheng authored Aug 10, 2024
  
  b68c4c07
- Support embedding input as a list (#1014) · 7599bade
  Ying Sheng authored Aug 10, 2024
  
  7599bade
09 Aug, 2024 5 commits
- Add skip_tokenizer_init args. (#959) · 9cf0a5ba
  gryffindor-rr authored Aug 10, 2024
```
Co-authored-by: lzhang <zhanglei@modelbest.cn>
```
  9cf0a5ba
- Add openai embedding API (#997) · b16e856f
  Ying Sheng authored Aug 09, 2024
  
  b16e856f
- bugfix: penalizers to be merged before reqs (#1001) · 10bca45b
  Juwan Yoo authored Aug 09, 2024
  
  10bca45b
- support models from www.modelscope.cn (#994) · b91a4cb1
  liuyhwangyh authored Aug 09, 2024
```
Co-authored-by: mulin.lyh <mulin.lyh@taobao.com>
```
  b91a4cb1
- test: negative value testing for frequency, presence penalizers (#995) · 95a28019
  Juwan Yoo authored Aug 08, 2024
  
  95a28019
08 Aug, 2024 3 commits
- Add e5-mistral embedding model - step 3/3 (#988) · e040a245
  Ying Sheng authored Aug 08, 2024
  
  e040a245
- feat: frequency, min_new_tokens, presence, and repetition penalties (#973) · ab787594
  Juwan Yoo authored Aug 08, 2024
  
  ab787594
- support more optioin about usage in stream mode (#985) · 3a79613c
  yichuan~ authored Aug 08, 2024
```
Co-authored-by: Ying Sheng <sqy1415@gmail.com>
```
  3a79613c
07 Aug, 2024 3 commits
- chore: update vllm to 0.5.4 (#966) · c31f084c
  Yineng Zhang authored Aug 07, 2024
  
  c31f084c
- misc: simplify test (#964) · 5f6fa04a
  yichuan~ authored Aug 07, 2024
  
  5f6fa04a
- Add support for Batch API test (#936) · 795eab6d
  yichuan~ authored Aug 07, 2024
  
  795eab6d
05 Aug, 2024 3 commits
- Feat: add alternative choices selection methods (#835) · 94e01151
  Aidan Cooper authored Aug 05, 2024
  
  94e01151
- Fix prompt len in parallel sampling (#928) · fd7926e4
  yichuan~ authored Aug 05, 2024
  
  fd7926e4
- Test regex in vision api (#926) · 0a4f5f9b
  Ying Sheng authored Aug 04, 2024
  
  0a4f5f9b