GitLab

- Add support for Apple M1 chips (mps device)
- Refactor image processing for better performance and compatibility
- Update model loading and inference for various devices
- Adjust batch processing and memory management

- Set NPUDTCompile to false for better performance on NPU
- Adjust batch ratio

- Increase batch ratio to 8 for GPU memory >=16GB
- Improve inference performance on systems with higher GPU memory

- Simplify batch ratio logic for GPU memory >= 16GB
- Remove unnecessary conditions for 20GB and 40GB memory

- Simplify batch ratio logic for GPU memory >= 16GB
- Remove unnecessary conditions for 20GB and 40GB memory

- Move batch model initialization outside the loop
- Collect page dimensions before analyzing- Update page info dictionary structure
- Add null dimensions for non-analyzed pages

- Update the logic for determining `end_page_id` to handle negative values
- This change ensures proper behavior when `end_page_id` is set to -1 or other negative values

- Move environment variable settings for NPU, MPS, and other configurations to the global scope in doc_analyze_by_custom_model.py
- Remove redundant environment variable settings in pdf_extract_kit.py
- This change ensures consistent configuration across the application and avoids potential conflicts or duplicate settings

- Remove unused utility functions
- Update import statements for better readability
- Add conditional imports for Ascend plugin
- Refactor table model initialization to support NPU

- Update batch ratio calculation logic to better utilize available GPU memory
- Improve logging for all GPU memory sizes

- Update batch ratio calculation for GPU memory range
- Increase upper limit for batch ratio 16 from 24 to 32 GB

- Update conditions for batch ratio assignment:
  -8 <= gpu_memory < 10: batch_ratio = 2 - 10 <= gpu_memory <= 12: batch_ratio =4
- This fix ensures proper batch ratio selection for GPU memory sizes

- Improve batch ratio calculation based on GPU memory
- Enhance performance for devices with 8GB or more VRAM

- Reduce batch_ratio by 1 for better performance and stability
- This change ensures more consistent memory usage when processing documents

refactor(magic_pdf): adjust VRAM allocation and MFR batch size- Update VRAM allocation logic to use 'VIRTUAL_VRAM_SIZE' environment variable
- Reduce MFR (Math Formula Recognition) batch size from 64 to 32

- Update GPU memory check and batch ratio calculation logic
- Add support for virtual VRAM size environment variable
- Improve logging for GPU memory and batch ratio

- Reduce YOLO_LAYOUT_BASE_BATCH_SIZE from 4 to 1
- Simplify batch ratio calculation for formula detection
- Remove unused conditional logic in batch ratio determination

- Commented out the original batch ratio calculation
- Set a fixed batch ratio of 2 for GPUs with less than 8 GB memory
- Increased batch ratio to 4 for GPUs with 8 GB or more memory

- Adjust end_page_id calculation to prevent IndexError when accessing pages
- Enhance error handling in LLM post-processing by specifically catching JSONDecodeError

- Add support for NPU (Neural Processing Unit) when available
- Implement batch analysis for GPU and NPU devices
- Optimize memory usage and improve performance
- Update logging and error handling

- Update clean_memory function to support both CUDA and NPU devices
- Implement get_device function to centralize device selection logic
- Modify model initialization and memory cleaning to use the selected device
- Update RapidTableModel to support both RapidOCR and PaddleOCR engines

- Add YOLOv11 language detection model for PDF documents
- Implement language detection in PymuDocDataset
- Update app.py to include 'auto' language option
- Create language detection utilities and constants

- Remove YOLO_VERBOSE environment variable from multiple files
- Set verbose=False in YOLOv8 prediction method to suppress logger output

- Import paddle module and disable its signal handler to prevent interference with other components
- This change addresses potential conflicts between PaddlePaddle and other libraries or system signals

- Add environment variables to disable albumentations and yolo updates
- Import torchtext and disable deprecation warnings
- Update unimernet to 0.2.2
- Specify ultralytics version as >=8.3.48
- Remove upper version limit for torch

- Remove concurrency limit logic from app.py
- Update model initialization process in various modules
- Remove unused VRAM check for concurrency limit
- Refactor OCR model initialization in pdf_extract_kit.py
- Update txt_spans_extract_v2 function to use lang parameter instead of ocr_model

- Remove usage of ModelSingleton class
- Initialize model directly using custom_model_init function
- Add self._lock attribute to PDFExtractKit class for thread safety- Replace local lock with self._lock for OCR processing

- Decrease the maximum image size threshold from 9000 to 4500 pixels
- This change aims to improve performance and reduce memory usage
- Affects the custom model document analysis process

- Move page total time logging to doc_analyze_by_custom_model.py
- Remove page total time logging from pdf_extract_kit.py
- Add page_start timing variable to custom model analysis
- Update logger output format for page total time

- Add new layout model option: DocLayout-YOLO
- Implement model initialization and prediction for DocLayout-YOLO
- Update configuration options to include new model- Modify existing code to support both LayoutLMv3 and DocLayout-YOLO models
- Update Gradio app to support more Custom Switch

- Reintegrate para_split_v3 into the pdf_parse_union_core_v2 process
- Add support for specifying page range in doc_analyze_by_custom_model
- Implement garbage collection and memory cleaning after processing
- Refine image loading from PDF, including handling out-of-range pages