GitLab

docs: update README files with improved instructions for sglang acceleration using tensor parallelism

master->dev

Release 2.0.5

update docs

fix: update version to 2.0.5 and clarify changelog entries in README files

Dev

Fix: 表格内容中的HTML Entity会导致表格内容错乱 #2694

chore: speed up "mineru --help"

master->dev

Release 2.0.4

Dev

fix: include model_path in key for backend model retrieval in vlm_analyze.py

fix: support NPU device in UnimernetModel initialization

correct the chat template of sglang_server 

MinerU2.0's chat template during training has additional "\n" compared with chatml's. This difference may slightly affect the performance under server mode. (because sglang_server use sglang's chat template).

add new enum values and improve MIN_BATCH_INFERENCE_SIZE documentation in pipeline_analyze.py

fix: include server_url in model key for backend model retrieval in vlm_analyze.py