Update dependencies and refactor embedding models #150
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
Former-commit-id: c2a8558 Former-commit-id: 1c96aa8bcd74bd0f97a255cfcc0c29d5d25b3c30
Former-commit-id: 1753ccf Former-commit-id: 3d6f6989f2f8522b326aea858e356a166d64b7e7
Former-commit-id: 6448cf0 Former-commit-id: 6867ccda9759df9ffacde7a02616b1ae9f1f067c
- Bumped the project version to 0.6.2 in `Cargo.toml`. - Updated `processors-rs` dependency version to 0.6.2. - Refined `Cargo.lock` to include specific versions for `candle-core`, `candle-flash-attn`, and `candle-kernels`, ensuring compatibility across the project. - Added `strip` option in `pyproject.toml` for optimized builds. - Removed optional `cudarc` dependency from `python/Cargo.toml` to streamline CUDA integration.
- Removed specific versioning for `candle-core`, `candle-flash-attn`, and `candle-kernels` to allow for more flexible dependency resolution. - Added source and checksum information for several packages to ensure integrity and compatibility. - Cleaned up redundant package entries to enhance clarity and maintainability of the dependency list.
- Updated existing dependencies including `onnxruntime` to version 1.22.0 in `pyproject.toml`. - Refactored embedding models to use `RwLock` for session management, improving thread safety in `colbert.rs`, `bert.rs`, `jina.rs`, and `reranker/model.rs`. - Adjusted model IDs in `colpali.rs` for consistency with new architecture. - Removed deprecated `ort_qwen3.rs` file.
…colpali_ort.rs - Updated the 'ort' dependency in Cargo.toml to include the 'half' feature for improved performance. - Refactored tensor creation in colpali_ort.rs to use ort::value::Value instead of ort::value::TensorRef, enhancing clarity and functionality. - Adjusted the output extraction to use f16 instead of f32 for better precision.
- Changed the ONNX model reference in `colpali.py` from "akshayballal" to "starlight-ai". - Adjusted the way top pages are selected in `colpali.py` for better performance. - Updated the model reference in `late_chunking.py` to use "jinaai/jina-embeddings-v2-small-en". - Simplified the file path in `Qwen.py` for embedding. - Removed unnecessary print statement in `splade.rs`. - Refactored error handling in `ort_bert.rs` and `ort_jina.rs` to use the `?` operator for better readability.
- Implemented logic to automatically add common CUDA installation paths on Windows. - Added a warning message if no CUDA DLL directory is found, guiding users to ensure proper installation. - This enhancement improves compatibility for users working with CUDA on Windows systems.
…ath formatting in embed_anything - Updated the `onnxruntime-gpu` dependency in `pyproject.toml` to version 1.22.0 for improved performance and compatibility. - Changed backslashes to forward slashes in CUDA paths for consistency and better cross-platform compatibility in `embed_anything/__init__.py`.
- Changed the project name from "embed_anything_gpu" to "embed_anything" for clarity. - Updated the `onnxruntime-gpu` dependency to `onnxruntime` in `pyproject.toml` to streamline compatibility. - Removed CUDA path handling code from `embed_anything/__init__.py` to simplify the implementation.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
onnxruntimeto version 1.22.0 inpyproject.toml.RwLockfor session management, improving thread safety incolbert.rs,bert.rs,jina.rs, andreranker/model.rs.colpali.rsfor consistency with new architecture.ort_qwen3.rsfile.