v0.0.5

What's Changed

• Add option to select models

Full Changelog: v0.0.4...v0.0.5

v0.0.4

What's Changed

• Use PyMuPDF for faster PDF processing
• Implemented better file validation logic
• Added Fakes for embedding, vector store, and chat model classes for testing

Full Changelog: v0.0.3...v0.0.4

v0.0.3

What's Changed

• Refactored whole code base
• Added better UI validation logic
• Used token count for chunk size. Issues #8 #15
• Load API key from env variables in local devlopement
• Implemented better error logging
• Added unit tests

Full Changelog: v0.0.2...v0.0.3

v0.0.2

TL;DR

• Switched to OpenAI chat models for faster and cheaper completions
• Embedding takes much less time due to batched requests

What's Changed

• Used OpenAI chat model instead of completion model by @mmz-001 in #42
• Updated dependecies by @mmz-001 in #42
• Refactors and small bug fixes by @mmz-001 in #42
• Remove custom embedding module by @mmz-001 in #42

Full Changelog: v0.0.1...v0.0.2

v0.0.1

What's Changed

• [Fix] add exponential backoff for embeddings by @mmz-001 in #6
• [Ops] limit max file size to 15 MB by @mmz-001 in #10
• [Refactor] move sidebar to components by @mmz-001 in #11
• [Refactor] lint by @mmz-001 in #12
• added dockerfile and update sidebar by @shantanuo in #16
• Flake8 on embeddings.py by @umeed-salman in #19
• add FAQ section and fix imports by @mmz-001 in #23
• [Deps] add flake8 by @mmz-001 in #24
• [Fix] fix imports by @mmz-001 in #25
• bump streamlit to 1.24.0 and use new caching methods by @mmz-001 in #41

New Contributors

• @shantanuo made their first contribution in #16
• @umeed-salman made their first contribution in #19

Full Changelog: https://github.com/mmz-001/knowledge_gpt/commits/v0.0.1