729 Commits

Author SHA1 Message Date
Gen Sato
dbe7a7c4fd
Fix: Add a INFO-level log when fallback to gpt2tokenizer (#12508) 2025-01-09 14:37:46 +08:00
-LAN-
0a49d3dd52
fix: tiktoken cannot be loaded without internet (#12478)
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-01-08 14:49:44 +08:00
crazywoola
6222179a57
Revert "fix:deepseek tool call not working correctly" (#12463) 2025-01-08 10:50:34 +08:00
Infinitnet
4e6c86341d
Add 'document' feature to Sonnet 3.5 through OpenRouter (#12444) 2025-01-07 19:51:38 +08:00
呆萌闷油瓶
9677144015
fix:deepseek tool call not working correctly (#12437) 2025-01-07 17:25:38 +08:00
SiliconFlow, Inc
15797c556f
add fish-speech-1.5 from siliconflow (#12425) 2025-01-07 15:27:34 +08:00
-LAN-
d3f5b1cbb6
refactor: use tiktoken for token calculation (#12416)
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-01-07 13:32:30 +08:00
SiliconFlow, Inc
dc650c5368
Fixes #12414: Add cheaper model and long context model for Qwen2.5-72B-Instruct from siliconflow (#12415) 2025-01-07 11:28:24 +08:00
Alex Chen
2bb521b135
Support TTS and Speech2Text for Model Provider GPUStack (#12381) 2025-01-07 09:42:11 +08:00
SiliconFlow, Inc
409cc7d9b0
mark deprecated models in siliconflow #12399 (#12405)
Co-authored-by: crazywoola <427733928@qq.com>
2025-01-07 09:08:58 +08:00
Warren Chen
147d578922
[Fix] revert sagemaker llm to support model hub (#12378) 2025-01-06 18:01:45 +08:00
方程
6df17a334c
fix: Update the API call address for the text_embedding model (#12342)
Co-authored-by: 方程 <fangcheng@oschina.cn>
2025-01-03 19:19:17 +08:00
jifei
3c2e30f348
fix: #12143 support streaming mode content start with "data:" (#12171) 2025-01-03 16:33:37 +08:00
丹枫染秋色
7c1961e618
feat: Add response format support to GLM-4 (#12252) 2025-01-03 09:38:50 +08:00
xander-art
baeddd4d15
feat:Add support for stop parameter in hunyuan model #12313 (#12315)
Co-authored-by: xander-art <xander-art@gmail.com>
2025-01-03 09:15:04 +08:00
-LAN-
6f5a8a33d9
refactor: replace gevent threadpool with ProcessPoolExecutor in GPT2Tokenizer (#12316)
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-01-03 09:13:18 +08:00
Giovanny Gutiérrez
d7c0bc8c23
feat: Add response format support for openai compat models (#12240)
Co-authored-by: Gio Gutierrez <giovannygutierrez@gmail.com>
2025-01-02 09:59:34 +08:00
yihong
f30bf08580
fix: close #12215 for yi special case (#12222)
Signed-off-by: yihong0618 <zouzou0208@gmail.com>
2025-01-02 09:58:34 +08:00
Warren Chen
9954ddb780
[Fix] modify sagemaker llm (#12274) 2025-01-02 09:49:11 +08:00
-LAN-
6a85960605
feat: implement asynchronous token counting in GPT2Tokenizer (#12239)
Signed-off-by: -LAN- <laipz8200@outlook.com>
2024-12-31 17:02:08 +08:00
Kepler
2a909e634b
feat: support Ernie-lite-pro-128k (#12161)
Co-authored-by: bigfish49 <bigfish49@126.com>
2024-12-27 20:23:46 +08:00
jiangbo721
c98d91e44d
fix: o1 model error, use max_completion_tokens instead of max_tokens. (#12037)
Co-authored-by: 刘江波 <jiangbo721@163.com>
2024-12-25 13:29:43 +08:00
yihong
56e15d09a9
feat: mypy for all type check (#10921) 2024-12-24 18:38:51 +08:00
yihong
6a0ff3686c
fix: fix typo (#12034)
Signed-off-by: yihong0618 <zouzou0208@gmail.com>
2024-12-24 15:23:27 +08:00
-LAN-
af2888d394
fix: remove json_schema if response format is disabled. (#12014)
Signed-off-by: -LAN- <laipz8200@outlook.com>
2024-12-23 17:53:57 +08:00
-LAN-
10caab1729
fix: change CredentialsValidateFailedError to inherit from ValueError (#11950)
Signed-off-by: -LAN- <laipz8200@outlook.com>
2024-12-22 10:43:31 +08:00
非法操作
366857cd26
fix: gemini system prompt with variable raise error (#11946) 2024-12-21 23:14:05 +08:00
-LAN-
455791b710
fix(model_runtime): make invoke as ValueError (#11929)
Signed-off-by: -LAN- <laipz8200@outlook.com>
2024-12-21 21:22:14 +08:00
Kalo Chin
2681bafb76
fix: handle document fetching from URL in Anthropic LLM model, solving base64 decoding error (#11858) 2024-12-20 18:23:42 +08:00
yihong
7b03a0316d
fix: better memory usage from 800+ to 500+ (#11796)
Signed-off-by: yihong0618 <zouzou0208@gmail.com>
2024-12-20 14:51:43 +08:00
-LAN-
996a9135f6
feat(llm_node): support order in text and files (#11837)
Signed-off-by: -LAN- <laipz8200@outlook.com>
2024-12-20 14:12:50 +08:00
Dr.MerdanBay
bb2f46d7cc
fix: add safe dictionary access for bedrock credentials (#11860) 2024-12-20 12:13:39 +09:00
yihong
463fbe2680
fix: better gard nan value from numpy for issue #11827 (#11864)
Signed-off-by: yihong0618 <zouzou0208@gmail.com>
2024-12-20 09:28:32 +08:00
非法操作
9d93ad1f16
feat: add gemini-2.0-flash-thinking-exp-1219 (#11863) 2024-12-20 09:26:31 +08:00
yihong
12d45e9114
fix: silicon change its model fix #11844 (#11847)
Signed-off-by: yihong0618 <zouzou0208@gmail.com>
2024-12-19 20:50:09 +08:00
barabicu
d057067543
fix: remove ruff ignore SIM300 (#11810) 2024-12-19 18:30:51 +08:00
sino
560d375e0f
feat(ark): add doubao-pro-256k and doubao-embedding-large (#11831) 2024-12-19 17:49:31 +08:00
Agung Besti
3388d6636c
add-model-azure-gpt-4o-2024-11-20 (#11803)
Co-authored-by: agungbesti <agung.besti@insignia.co.id>
2024-12-19 12:36:11 +08:00
xander-art
56434db4f5
feat:add hunyuan model(hunyuan-role, hunyuan-large, hunyuan-large-rol… (#11766)
Co-authored-by: xanderdong <xanderdong@tencent.com>
2024-12-18 15:25:53 +08:00
-LAN-
a5db7c9acb
feat: add openai o1 & update pricing and max_token of other models (#11780)
Signed-off-by: -LAN- <laipz8200@outlook.com>
2024-12-18 12:15:11 +08:00
非法操作
9048832a9a
chore: improve gemini models (#11745) 2024-12-17 17:42:21 +08:00
Shota Totsuka
7d5a385811
feat: use Gemini response metadata for token counting (#11743) 2024-12-17 17:42:05 +08:00
sino
99430a5931
feat(ark): support doubao vision series models (#11740) 2024-12-17 15:43:11 +08:00
非法操作
c9b4029ce7
chore: the consistency of MultiModalPromptMessageContent (#11721) 2024-12-17 15:01:38 +08:00
呆萌闷油瓶
cd4310df25
chore:update azure api version (#11711) 2024-12-17 13:39:56 +08:00
非法操作
74fdc16bd1
feat: enhance gemini models (#11497) 2024-12-17 12:05:13 +08:00
方程
fc8fdbacb4
feat: add gitee ai vl models (#11697)
Co-authored-by: 方程 <fangcheng@oschina.cn>
2024-12-16 18:45:26 +08:00
zhongliliu-butterfly
daccb10d8c
fix: volcengine_maas and baichuan message error (#11625)
Co-authored-by: zhongliliu <liuzlx@digitalchina.com>
2024-12-16 13:05:27 +08:00
zhaobingshuang
79801f5c30
fix: deepseek reports an error when using Response Format #11677 (#11678)
Co-authored-by: zhaobs <zhaobs@cailian.net>
2024-12-16 12:58:03 +08:00
非法操作
cf0ff88120
feat: add grok-2-1212 and grok-2-vision-1212 (#11672) 2024-12-15 21:18:24 +08:00