5.4.0

[kyteinsky]

5.4.0 - 2026-06-24

Highlights

• The indexing direction has been reversed now. Instead of the context_chat PHP app sending documents to the context_chat_backend ExApp, the ExApp downloads the documents from the server according to a list obtained from the PHP app. This also means that the occ context_chat:scan command serves no purpose and has been removed. Indexing should be smoother and run continuously now.
• Kubernetes support to scale the CPU computation
• Separate docker images for CPU, CUDA and ROCM (uses Vulkan) instead of one heavy CUDA/CPU image
• CUDA 12.8 is shipped in the CUDA image so the host drivers should be updated to this at the minimum.

Added

• add network embedding batching (Add network embedding batching #276) @fcharlaix-opendsi
• add kubernetes support and reverse content/indexing flow (feat: add kubernetes support #284) @kyteinsky @marcelklehr
• add gh workflows for docker builds and do separate cpu, cuda and rocm (vulkan) images (feat: add gh workflows for docker builds #295) @kyteinsky

Changed

• update readme according to the latest changes (chore: update readme according to the latest changes #300) @kyteinsky
• bump llama_cpp_python to 0.3.23 (chore: bump llama_cpp_python to 0.3.23 #301) @kyteinsky
• move task types to the backend (chore: move task types to the backend #321) @kyteinsky
• adjust comment in Dockerfile regarding RTX5090 support (chore: adjust comment in Dockerfile regarding RTX5090 support #316) @kyteinsky

Fixed

• improve loadSources error handling (fix: improve loadSources error handling #288) @kyteinsky
• fix(pgvector): add chunking to prevent long list of args in queries (fix(pgvector): add chunking to prevent long list of args in queries #290) @kyteinsky
• fix(pgvector): make doc deletion query faster (fix(pgvector): make doc deletion query faster #289) @kyteinsky
• drop latin-1 decode in source title and userIds (fix: drop latin-1 decode in source title and userIds #306) @kyteinsky
• handle validation errors of files and content providers individually (fix: handle validation errors of files and content providers individually #308) @kyteinsky
• prevent race condition in vectordb tables creation (fix: handle validation errors of files and content providers individually #308) @kyteinsky
• pass actual error in the error object (fix: handle validation errors of files and content providers individually #308) @kyteinsky
• add container hostname to /etc/hosts to silence sudo warning (fix: add container hostname to /etc/hosts to silence sudo warning #311) @sanzakicesarr

🤖 AI (if applicable)

• The content of this PR was partly or fully generated using AI