Update sample questions in frontend UI

Final updates

fix: enhance claim extraction by including query context in faithfulness evaluation

fix: enhance report filtering options in sidebar and update report content for clarity

fix: adjust top_k parameter in retrieve function to account for blocked section titles

fix: enhance report detection logic for cross-report queries by collecting multiple airlines and flight numbers

fix: LLM as a judge prompt

fix: rename and expose build_faithfulness_context function for UI integration

Upload src/evaluation/evaluate.py with huggingface_hub

Upload src/evaluation/evaluate.py with huggingface_hub

fix: update ablation study documentation and add placeholder handling for missing strategies

fix: ID mismatch and garbage chunks removed

fix: enhance strategy mapping and handle legacy ID prefixes in retrieval

Update evaluation, retrieval query, UI, and add semantic rechunking experiments

Add evaluation and diagnostic scripts for retrieval methods

feat: inject executive summaries as on-the-fly context for LLM and reranking

feat: switch cross-encoder from ms-marco-MiniLM to qnli-distilroberta-base

Update LLM client, retrieval, and UI components

refactor: adjust rrf_fuse_lists parameters for improved document handling and update query event detection logic

New md_recursive chunked and upserted, need to test

Refactor BlackboxRAG documentation and enhance metadata handling

Merge branch 'main' of https://github.com/Abdullah-Khan-Sherwani/BlackboxRAG

fix: always use section chunks for Pinecone text lookup

Merge remote-tracking branch 'origin/main'

fix: remove strategy filter from Pinecone queries + cache semantic embeddings

fix: update embedding task from retrieval to text-matching for improved accuracy

feat: enhance evaluation module with final answer extraction and update relevancy computation prompts

chore: align retrieval artifact resolution and manual evaluation updates

Add simple baseline chunk/upsert script and fix semantic chunking perf

feat: add LLM knowledge document generation for semantic retrieval

fix: resolve merge conflicts keeping local changes, restore HyDE improvements

feat: add HyDE document generation and retrieval strategy tracking in hybrid retrieval

old section aware is back. semantic + BM25 works good but cross encoders hurting . no multi query either

IMPORTANT: Unzip chunks_md_recursive.zip and then test

feat: enhance markdown migration and embedding processes with entity-aware summaries and provenance context

Clean up: Remove temporary debugging scripts, keep production code and enhancements

feat: replace HyDE document generation with multi-query expansion for improved retrieval

feat: enhance LLM integration with GPT support and improve query handling in UI

Merge branch 'main' of https://github.com/Abdullah-Khan-Sherwani/BlackboxRAG

Shantu mantu chunking done

feat: implement parent chunking, diversity tuning, and network resilience

my changes

feat: Added Checkpointed evaluation

feat: llm folder created for ease of use

feat: Changed model from gemini to deepseek

fix: Changed model to gemini 2.5 flash. Will change to Deepseek v3

feat: Added placeholder UI

feat: Added ablation and modified evaluation pipeline

fix: Pinecone metadata limit being reached fixed

feat: Added Hybrid search