Tags: docqai/docq
Tags
refactor: switch shared as to use a Llama Index pipeline implementati… …on and adopt HyDE. (#259) * refactor(query): switch to implementation that uses Llama Index pipelines. Switch to HyDE rather than n query rewrite technique. Still running BM25 retrieval also and using reciprocal rank fusion as reranker. * refactor: fix some undefined type errors * build: update to Ruff v0.2.2 -> v0.4.4 * build: update llama-index-core from v0.10.39 to v0.10.39 * refactor: split things into files a bit more to avoid circular deps. * docs(readme): update 2nd level tag line to make it clearer. * tests: fix failing test from moving some indexing and storage functions to a different file. * feat: add GPT-4o support * refactor: move azure env var strings to config file and reference by py variable in code * update azure_openai_latest setting to use gpt-4o
fix: handle org setting enabled feat keys that are no longer valid. that's a feature that used to exist that was saved as a setting for an org in the DB. When the feature is removed from the avail features list (enum) because we no longer want it, we want the admin UI code to just ignore it.
PreviousNext