[cherry-pick]update docker for 3.2.0 #4496
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Thanks for your contribution! |
* support save markdown content in parsing_res_list * update docs
…PaddlePaddle#4508) * Add missing licenses and update PP-StructureV3 server interface * Update parameter name
* Fix hps bug * Update deps * Fix doc * Fix doc
* Add genai server Co-authored-by: zhang-prog <zhang-prog@users.noreply.github.com> * Underline to hyphen * Fix bugs and complete client code * Add logging info * Support SGLang server Co-authored-by: zhang-prog <zhang-prog@users.noreply.github.com> * to batch image_embeddings * sglang batch * add xformers for genai-sglang-server * Add PaddleOCR-VL paddle code Co-authored-by: boomercat <boomercat@users.noreply.github.com> * Fix bugs * update chat_template file name * Fix bugs * Fix bugs * Fix bugs * Add .contiguous * Use PaddleFormers implementation * Fix bug * support pp-ocr-vl pipeline * Fix default config bug * Align with original version * No version check for paddle2onnx * Fix bugs * Support KV cache * Fix sglang model name * Support basic and high-stability serving * Update docs * Fix chat template and support setting skip_special_tokens * Fix bugs * Add TODO * Support setting max_new_tokens from client side * Fix dtype * fix vllm acc * ensure FD, vLLM, SGLang use the consistent chat template * add run_fastdeploy_server * collect metrics for sglang * set default max_model_len and max_new_tokens * support vl * FastDeploy server temperature * setup.py add chat template * set default max_num_batched_tokens * Fix bug and report batch size * fix: set default max_completion_tokens only for PaddleOCR-VL * fix: use PIL instead of OpenCV to encode ndarray as jpg * make fd use greedy sampling * vllm use 3d rope * Support async processing and asyncio-based genai client * Reset FD gpu-memory-util to 0.3 * Update paddle implementation * sglang use 3d rope * Fix paddle API * support vl * truncate_repetitive_content * support to rebuild table including figures (#1) Co-authored-by: root <root@szzj-acg-tge2-1dc17.szzj.baidu.com> * merge vlm-pipeline-serve * vLLM: set dynamic_arg_dims to support mrope * thread-safe ppocrvl * update pipeline config * Support safetensors * Optimize performance * Update pipeline configs * support ctrl use_layout_detection && use_chart_recognition, support merge formula_number in formula * Update config * Update HPS configs * Bump hps sdk version * Update service interface * Fix flash-attn * vLLM use flash_attn * SGLang use flash_attn * Add flash-attn dep check * clean code * Update for PaddleOCR and fix bugs * Fix setup.py * vllm support mm_kwargs (eg: min_pixels, max_pixels) * Support setting max_concurrency * vllm: improve the robustness of SiglipAttention * sglang: improve the robustness of SiglipAttention * Remove third-party license folder * Remove third-party licenses * add min_pixels, max_pixels params for vlm predictor * sglang: apply image resize correctly * compatible with RGBA images * use stronger chat_template and new image_placeholder * convert local image to RGB mode * use flash-attn==2.8.3 * Support more VLM parameters * merger vlm-pipeline-serve * pretty table and chart * Sync config * support use PP-OCR-VL pipeline in CLI * ctrl show order label * update doc for PP-OCR-VL * fix image save BGR && remove vl crop images in results * Update PP-OCR-VL docs * HPI supports multi-language text rec models * fix image save BGR && remove vl crop images in results * Support return_word_box in server * Fix bugs * fix bug * sglang use flash-attn * Fix bug * Bump PaddleX version * limit installation of safetensors to linux only & replace paddlex image_token * Rename pipeline and model * Fix dep specifier * update model name --------- Co-authored-by: Bobholamovic <bob1998425@hotmail.com> Co-authored-by: zhang-prog <zhang-prog@users.noreply.github.com> Co-authored-by: Bobholamovic <mhlin425@whu.edu.cn> Co-authored-by: zhangyue66 <zhangyue66@baidu.com> Co-authored-by: boomercat <boomercat@users.noreply.github.com> Co-authored-by: Tingquan Gao <35441050@qq.com> Co-authored-by: root <root@szzj-acg-tge2-1dc17.szzj.baidu.com>
Co-authored-by: cuicheng01 <45199522+cuicheng01@users.noreply.github.com>
* support PaddleOCR-VL * remove confusing number characters, 0, 1, 9 * support PP-DocLayoutV2
* Fix bugs * Fix all
* Rename files and classes * Fix doc * Fix bug
* fix PaddleOCR-VL name - local * fix
* add PP-DocLayoutV2 in official models * support concatenate_markdown_pages
* Add missing param * Remove unused extra
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
9 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.