Yingda Chen
3e13cc899b
add transformer support for Qwen2vl ( #1106 )
...
* add qwen2vlconfig
* rearrange
---------
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-28 20:08:14 +08:00
Yingda Chen
e2bd302175
fix potential double definition for ocr pipeline ( #1102 )
...
* fix potential double definition issue
---------
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-28 00:43:24 +08:00
Yunlin Mao
a7856a5995
add multi-thread download ( #1095 )
...
* add thread download
* add thread download
* fix print
* change default workers to 8
* fix return cache path
* manage tqdm progress bars
---------
Co-authored-by: DaozeZhang <zdz408@126.com >
2024-11-26 20:22:36 +08:00
Yingda Chen
6d9e6d57c0
More automodel ( #1098 )
...
* add more hf alias
---------
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-25 22:16:05 +08:00
Yingda Chen
4a3b255d53
change warning to debug ( #1099 )
...
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-25 22:15:03 +08:00
Yingda Chen
5ca12c6cc4
Llamafile support gpu flag ( #1097 )
...
* add gpu flag when gpu is detected
* fix typo
* fix typo
* add printout prompt
---------
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-25 16:31:52 +08:00
Jintao
e3f63fd1ea
lazy print ast logs ( #1089 )
2024-11-25 10:32:16 +08:00
Yingda Chen
2b1c839918
Format llm pipeline ( #1094 )
...
* format llm pipeline
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-22 20:04:59 +08:00
Yingda Chen
ddc5fab311
Add AutoModelForImageSegmentation and T5EncoderModel; Support from subfolder option ( #1096 )
...
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-22 19:43:47 +08:00
Jintao
d687c9c514
fix tensorflow warning ( #1093 )
2024-11-22 09:52:56 +08:00
Mashiro
18de9c079e
refactor(models/audio/ans): Optimize zipformer and scaling layers ope… ( #1088 )
...
* refactor(models/audio/ans): Optimize zipformer and scaling layers operations
This change modifies some operations in `zipformer` and `scaling` to use an in-place method for efficiency.
---------
Co-authored-by: Haoxu Wang <wanghaoxu.whx@alibaba-inc.com >
2024-11-20 16:18:33 +08:00
Yingda Chen
63febc58a6
add llamafile support to command line ( #1087 )
...
* add llamafile support to command line
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-20 09:24:58 +08:00
Yingda Chen
91d85f23c2
Fix dependency ( #1085 )
...
* fix yaml dependency for hub
---------
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-14 16:41:24 +08:00
Yingda Chen
d25b27f6d5
Log reduce ( #1081 )
...
* do not print log for symbolic link creation failure due to existing ones
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-13 18:43:38 +08:00
Yingda Chen
d322854c6f
fix log for downloading to local dir ( #1080 )
...
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-12 12:15:29 +08:00
Jintao
da47da41bc
Fix facial 68ldk detection ( #1078 )
2024-11-11 16:52:59 +08:00
Jintao
e961f1671d
Update docker from release/1.20 ( #1077 )
2024-11-11 14:56:36 +08:00
Jintao
61c2fd97e4
Update llm docker ( #1076 )
2024-11-08 18:09:05 +08:00
Jintao
a393b6ef24
update docker ( #1075 )
2024-11-08 15:49:50 +08:00
Jintao
d02d887918
update docker ( #1073 )
2024-11-08 10:56:39 +08:00
Jintao
125f44fb20
update docker evalscope version ( #1071 )
2024-11-07 18:56:17 +08:00
suluyana
19fc9dc3f1
feat ollama template: llama3.2-vision ( #1070 )
2024-11-07 16:09:12 +08:00
Jintao
a51f25abdf
fix docker numpy version ( #1069 )
2024-11-07 13:58:08 +08:00
tastelikefeet
33308cedd4
fix numpy build error ( #1068 )
2024-11-06 20:40:58 +08:00
tastelikefeet
6856b157ed
try to reduce the image size of llm ( #1067 )
2024-11-06 15:52:08 +08:00
Jintao
9cf74e01c0
Fix the missing __init__.py file ( #1066 )
2024-11-06 13:22:06 +08:00
Yingda Chen
1a9ada4e06
create a symbolic link for special models that has been masked to avoid dot in model name ( #1063 )
2024-11-04 20:44:54 +08:00
tastelikefeet
43536064d5
default install tf-keras ( #1064 )
2024-11-04 20:15:18 +08:00
suluyana
5e78f92e24
Feat(multimodal model):ovis vl pipeline ( #1057 )
2024-11-04 14:24:07 +08:00
suluyana
ae1541846b
Template.to_ollama: add new argument split ( #1039 )
...
feat: 1. add new argument `split` 2. ollama template
2024-11-04 13:46:07 +08:00
Yingda Chen
0b47683f69
improve upload model, remove requirment for configuration.json ( #1062 )
...
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-04 13:33:23 +08:00
Yingda Chen
83f1e20e80
add repo existence check hub-api ( #1060 )
...
* add repo existence check api
* update
---------
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-02 08:04:40 +08:00
Yingda Chen
06108d105a
add log for download location ( #1061 )
...
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-02 08:02:50 +08:00
Mark
250b72fce7
fix: text error correction batch run bug ( #1052 )
...
Co-authored-by: Mark <smartmark-pro@qq.com >
2024-11-01 09:35:19 +08:00
Yingda Chen
fac865fd97
OCR pipeline shall depend on TF only when necessary ( #1059 )
...
* ocr pipeline tf as optional
Co-authored-by: Yingda Chen <yingda.chen@alibaba-inc.com >
2024-11-01 09:10:37 +08:00
Mashiro
3de8430c03
fix(audio ans pipeline): Restore file reading from string input in ANSZip… ( #1055 )
...
* fix(audio pipeline): Restore file reading from string input in ANSZipEnhancerPipeline
Fix the code to support various types of input, including local or remote URLs.
Co-authored-by: Haoxu Wang <wanghaoxu.whx@alibaba-inc.com >
2024-10-30 16:34:35 +08:00
tastelikefeet
328bbc0494
Fix some bugs ( #1056 )
...
* fix some bugs
* remove install tf-kersa
* fix
* use bin bash as default
2024-10-30 15:49:38 +08:00
tastelikefeet
3762ba3318
fix tests ( #1053 )
2024-10-28 19:50:33 +08:00
tastelikefeet
62ffc04215
Fix the slow downloading ( #1051 )
...
* fix slow download
* fix
* fix audio installation
* fix build
* fix
* test
* test
* test
* test
* fix
* fix
* fix
* fix
2024-10-25 13:37:01 +08:00
Mashiro
e71057cacf
feat(audio/ans): Add ZipEnhancer and related layers for acoustic nois… ( #1019 )
...
* feat(audio): Add acoustic noise suppression pipeline and tests for zipenhancer model
* Introduce `ZipEnhancer` module and associated layers (`ZipEnhancerLayer`, `Generator`, `ZipFormer`, ...).
* Add `speech_zipenhancer_ans_multiloss_16k_base` pipeline for `ZipEnhancer` module.
* Add new test cases and update metainfo.
Co-authored-by: Haoxu Wang <wanghaoxu.whx@alibaba-inc.com >
2024-10-24 15:38:09 +08:00
tastelikefeet
ae98067485
Fix timestamp in docker build ( #1049 )
...
* update flow name
* fix
* fix
* update docker builder
* lint
* fix build
* fix cpu build
* fix ts
2024-10-24 14:04:10 +08:00
tastelikefeet
acc60bab1c
Fix pypi mirror ( #1048 )
...
* update flow name
* fix
* fix
* update docker builder
* lint
* fix build
* fix cpu build
2024-10-24 10:35:10 +08:00
tastelikefeet
a6d583b6cf
Fix build error ( #1047 )
2024-10-23 18:44:11 +08:00
Xingjun.Wang
134fe72f06
hotfix for datasets 3.0.2 ( #1046 )
2024-10-23 16:25:53 +08:00
tastelikefeet
cda7a6f04a
Update docker scripts ( #1044 )
2024-10-23 15:18:40 +08:00
tastelikefeet
136d8c0a9d
Add docker workflow name ( #1043 )
2024-10-23 13:43:41 +08:00
tastelikefeet
7a57ee418c
Refine dockerfile ( #1042 )
2024-10-23 09:59:51 +08:00
tastelikefeet
eac004d7f2
Refine docker file ( #1041 )
2024-10-22 11:13:17 +08:00
tastelikefeet
0026729f7a
Fix dockerfile ( #1038 )
2024-10-21 19:47:11 +08:00
tastelikefeet
c4e561ec6b
fix file mode ( #1037 )
2024-10-21 19:31:08 +08:00
