v3.1.3

What's Changed

• Feat: IDEFICS models support + Mistral 7B Instruct by @bigadsoleiman in #135

Full Changelog: v3.1.2...v3.1.3

v3.1.2

What's Changed

• Update welcome.tsx by @jawhnycooke in #105
• Enable local user interface by @donatoaz in #107
• feat: embeddings comparison and heat matrix by @massi-ang in #110
• fix: correct number of columns in matrix by @massi-ang in #112
• fix: correct step increment for top p value by @massi-ang in #114
• Update semantic-search-details.tsx by @cpollard0 in #120
• Fix too long resource name by @massi-ang in #124
• fix: display full list of supported file extensions by @massi-ang in #129
• Feat styles for json viewer by @massi-ang in #131

New Contributors

• @jawhnycooke made their first contribution in #105
• @donatoaz made their first contribution in #107
• @cpollard0 made their first contribution in #120

Full Changelog: v.3.1.1...v3.1.2

v.3.1.1

What's Changed

• Kendra Index fixes by @spugachev in #104

Full Changelog: v.3.1.0...v.3.1.1

v.3.1.0

What's Changed

• fix: correct creation of config.json by @massi-ang in #66
• Readme update by @bigadsoleiman in #71
• Updated 'getName' naming pattern requirements by @QuinnGT in #73
• Add ability to deploy with GitHub Codespaces by @bigadsoleiman in #78
• Bedrock Cohere support fixes by @bigadsoleiman in #83
• Allow to access local Kendra Indexs without role by @spugachev in #96
• Workspace deletion by @spugachev in #98

New Contributors

• @QuinnGT made their first contribution in #73

Full Changelog: v.3.0.0...v.3.1.0

v.3.0.0

Updated version with Workspaces and Bedrock Streaming

v2.0.1

• Move to X86_64 architecture

v.2.0.0

1. Added multiple models providers

   • Amazon Bedrock (currently in preview)
   • Amazon SageMaker self hosted models from Foundation, Jumpstart and HuggingFace.
   • External providers via API such as AI21 Labs, Cohere, OpenAI, etc. See available langchain integrations for a comprehensive list.

2. Adding Multiple RAG sources

   • Amazon Aurora with pgvector
   • Amazon OpenSearch VectorSearch
   • Amazon Kendra

3. Replace Lambda Streaming with API Gateway Websockets API

4. Use CDK Constructs instead of NestedStack.

5. Use AWS Cloudscape Design System for user interface.

6. Model Interface logic all in Python.

7. More seamless integration with Langchain framework

v1.5.0

Utilizing Amazon Aurora PostgreSQL with pgvector for Semantic Search

v1.0.0

Merge pull request #6 from jingyan/patch-1

Update README.md