Langrila

What's Changed
* Update pip install command text in README.md by go5paopao in https://github.com/taikinman/langrila/pull/33
* Change the interface of vector db module for supporting more function of each client by taikinman in https://github.com/taikinman/langrila/pull/34
* Support chroma persistent client and http client by taikinman in https://github.com/taikinman/langrila/pull/34, https://github.com/taikinman/langrila/pull/36

New Contributors
* go5paopao made their first contribution in https://github.com/taikinman/langrila/pull/33

**Full Changelog**: https://github.com/taikinman/langrila/compare/v0.0.7...v0.0.8

What's Changed
* Refactor/qdrant by taikinman in https://github.com/taikinman/langrila/pull/31


**Full Changelog**: https://github.com/taikinman/langrila/compare/v0.0.6...v0.0.7

What's Changed
* Add tqdm for seeing batch processing progress of vector db collection creation and fix log message by taikinman in https://github.com/taikinman/langrila/pull/28
* Add embedding model config to calculate token length for embedding by taikinman in https://github.com/taikinman/langrila/pull/29


**Full Changelog**: https://github.com/taikinman/langrila/compare/v0.0.4...v0.0.6

fix version

What's Changed
Fix/bug by taikinman in https://github.com/taikinman/langrila/pull/26
- Fix import error in Qdrant modules
- Fix contaminating model's output into prompt for gemini modules
- Minor update README.md

**Full Changelog**: https://github.com/taikinman/langrila/compare/v0.0.3...v0.0.4

Main chages
- Supporting Gemini api
- Deprecate chroma support
- Enhance qdrant modules
- automatically collection separation to keep memory error away
- Change return type of `stream()` and `astream()` method to `CompletionResults` instance for all chunks
- Install option for required dependencies.

Breaking changes
- `from langrila import OpenAIChatModule` to `from langrila.openai import OpenAIChatModule`.
- It's the same as other modules related to openai api.
- Return type of `stream()` and `astream()` method. From v0.0.3, all return types of all chunks is CompletionResults.
- The name of retrieval results class : `RetrievalResult` to `RetrievalResults`
- Qdrant vector_size as argument.