implement optional MRR through Qdrant (cosine)

re-order prompt modules to increase cache hit rate

add MRR for diverse chunks

re distribute GPT 4 and 5 series

Make ML models globally imported at the Start Up

HADNOVER: clean logging

HADNOVER: clean logging

add warm up options for GPT5, auto-cache, and API signature determination

add warm up options for GPT5, auto-cache, and API signature determination

Fix context passing among Resolver x Other agents

Fix context passing among Resolver x Other agents

revert (GPT 5 does not suupport temperature, and runs longer, because it's reaosning model)

dump old gpt4 model

make ReAct style

adjust Source x Distrcit filters, relax follow ups

HANDOVER: delete old utils

HANDOVER: delete visual rag components

HANDOVER: delete visual rag components

HANDOVER: delete visual rag components

HANDOVER: delete visual rag components

HANDOVER: delete visual rag components

HANDOVER: delete visual rag components

HANDOVER: refactor / decouple UI components

HANDOVER: refactor pipeline, enhance logging

HANDOVER: refacotor utils

HANDOVER: refactor filters, resolving functionalities

HANDOVER: remove visual RAG strategy

HANDOVER: remove GFS strategy

HANDOVER: re-enable Feedback DataBase

HANDOVER: re-enable Feedback DataBase

HANDOVER: update configurations

HADNOVER: refactor agentic modules

HANDOVER cleanup

downgrade top-k values

add dynamic hardware based reranking routing

add timers for the retrieval

fix ui filter resolve

include Uganda Parliament Structure (analysis doc)

add entity type (UI)

add entity type (UI)

add entity type (infer)

update filters

fix comma separation issue and filter check

remove extra langchain sub packages

integrate the updates to the pipeline

multi agent level routing

adjsut the feedback

update configs (model routing)

add various helper tools and context / prompt enrichment for the agents

revert visual document (streamlit deprications)