v2.0.5

What's Changed

Add extra CORS origins.

Pull requests

Bugfix related to CORS origins by @zaldivards in #47.

Full Changelog: v2.0.4...v2.0.5

v2.0.4

What's Changed

Add -h option to the CLI. It will be very helpful when running contextqa in Docker

Pull requests

Dynamic hostname or ip address by @zaldivards in #46

Full Changelog: v2.0.3...v2.0.4

v2.0.3

What's Changed

• Fix ingestion errors related to file pointers
• Update layout of message adder (question input text)

Pull requests

• Sources ingestion bugfix by @zaldivards in #45

Full Changelog: v2.0.2...v2.0.3

v2.0.2

What's Changed

• Update and fix the SQLAlchemy query that handles the deletion of sources.
• Fix the rendering of sources' hashes.

Pull requests

• Sources removal bugfix by @zaldivards in #44

Full Changelog: v2.0.1...v2.0.2

v2.0.1

What's Changed

• Enhance the parsing of the agent's final answer.
• Update the searcher tool to return an "error message".

Pull requests

• Agent responses bugfix by @zaldivards in #43

Full Changelog: v2.0.0...v2.0.1

v2.0.0

What's Changed

• Now, users will be able to use ContextQA as a python utility using the pip install command
• Add github workflow for automating the package publication
• Refine development tools
• UI integration when building and publishing a release
• sentence-transformers/all-mpnet-base-v2 local encoder. More information about the model is available here

Pull requests

• Installable package by @zaldivards in #40
• Prevent errors in settings initialization in the dev environent by @zaldivards in #41
• HuggingFace encoder by @zaldivards in #42

Full Changelog: v1.7.0...v2.0.0

v1.7.0

What's Changed

• Notification sound on assistant response.
• Added button so that users can directly report a bug or request a new feature.
• Pretty format assistant responses, including code highlighting.

Pull requests

• Notification sound on assistant response by @zaldivards in #36
• UI Refinement by @zaldivards in #37

Full Changelog: v1.6.0...v1.7.0

v1.6.0

What's Changed

• Updated layouts
• Added Chips of common queries in chat sessions
• Improved responsive design
• Improved initialization logic: Redirect to home if ContextQA has not been initialized
• Added empty message in chat and qa sessions
• Users can now copy assistant responses (CopyButton component)
• Fixed line break bug in main textarea

Pull requests

• UI improvements by @zaldivards in #33

Full Changelog: v1.5.0...v1.6.0

v1.5.0

What's Changed

• Added stepper guidance when ContextQA is opened for the first time
• Support for OpenAI's models and Google Gemini
• Support for ChromaDB and Pinecone
• Support for Local and Redis memory
• Enhanced data sources handling
• Added Status view to check the health of the components used by ContextQA
• Agents, QA and chat sessions are now using LCEL
• Fixed bugs in streaming by using Langchain Runnable's astream_events function in agents and chat sessions
• Added persistent ContextQA settings
• Improved responsiveness
• Added usage of relational DB
• Now, digests of data sources are calculated
• Enhanced QA sessions by retrieving the relevant sources used to answer the user query

Pull requests

• Advanced settings by @zaldivards in #30

Full Changelog: v1.4.0...v1.5.0

v1.4.0

What's Changed

UI

• Incremental rendering of tokens (streaming)
• Rename of sections: Now the settings section is called Source ingestion
• Context manager now accepts up to 10 files to be ingested

API

• Now, the /qa/ and /bot/ endpoints return streaming responses
• The /qa/ingest/ endpoint can now handle up to 10 concurrent ingestions. The response contains the number of successful operations and the names of the sources that were skipped because they already existed and their content has not changed since the previous ingestion

Pull requests

• Streaming by @zaldivards in #24
• Source rendering (QA) by @zaldivards in #26

Full Changelog: v1.3.1...v1.4.0