HowItWorks: fix step 02 title 'Dual Pipeline' -> '3-Pipeline Activation'

Home page: fix 8 stale/wrong facts

Benchmark: add LLM-as-a-Judge + BERTScore (hackathon 30% accuracy criterion)

Branding: replace Claude with TigerGraph logo and remove Claude references

Explorer live query: better error, science chips, fallback answer

Explorer: 4 science scenarios, BFS hop filter, live query section

Fix UI: 3-pipeline playground, science examples, correct docs/architecture

Fix README: remove HF frontmatter, correct dataset/eval/quickstart/latency

Improve latency: parallel LLM calls, embedding cache, client reuse

Add HuggingFace repo card YAML metadata to README

Add corpus.jsonl dataset via git-lfs (478 Wikipedia science docs, 2.5M tokens)

Add .gitignore, dataset metadata, retrieval layer, and latest web/graphrag updates

Add dataset preparation script — downloads Wikipedia/arxiv/BBC, verifies 2M+ tokens, ingests into TigerGraph

Address feedback: add benchmark results table, ablation study, demo GIF section, 2M+ dataset plan

Fix #10: Rewrite README — 3-pipeline system, TG GraphRAG integration, LLM-Judge + BERTScore, NoveltyEngine wiring

Fix #8: Update web API compare route — 3-pipeline (LLM-Only + Basic RAG + GraphRAG) with side-by-side metrics

Fix #7: Update dashboard.py — 3-column layout (LLM-Only / Basic RAG / GraphRAG), fix _get_demo_passages() query matching, add LLM-Judge + BERTScore display

Fix #9: Update benchmark runner for 3-pipeline comparison + LLM-as-a-Judge + BERTScore evaluation

Update __init__.py with new exports (TGGraphRAGClient, TripleComparisonResult, LLM judge, BERTScore, etc.)

Fix #5: Install advanced GSQL queries (PPR, paths, activation) in setup_tigergraph.py

Fix #3: Add LLM-as-a-Judge (PASS/FAIL) + BERTScore evaluation — the two hackathon-required accuracy metrics

Fix #2+#4+#6: Add LLM-Only pipeline, wire NoveltyEngine, integrate TG GraphRAG client, 3-pipeline comparison

Fix #1: Add TigerGraph GraphRAG integration layer wrapping official repo REST APIs

Deep research update: comprehensive README with 12 cited papers, evaluation methodology, architecture deep-dives, and hackathon-aligned benchmarking strategy

🎨 Ultra premium UI: 6 separate pages with animations, bento grid, premium design

🎨 Add home page sections: Hero, Stats, Features, HowItWorks, Bento, Testimonials, CTA

🎨 Stats bar component

🎨 Premium navbar with page routing and mobile menu

🎨 Updated footer with page links

🎨 New multi-section home page

🎨 Update layout with new title

🎨 Ultra premium UI: enhanced design system with animations, gradients, bento grid

Massive README: 14 novelties from 10 papers, full architecture, paper citations, test counts"

Add unit tests for all 6 novelty techniques"

Add advanced GSQL queries: PPR scoring, shortest paths, spreading activation, community detection"

Add GraphRAG novelties engine: PPR scoring, token budget, path pruning, spreading activation, hybrid router, incremental updates

Add .npmrc for peer dep compat

Fix TS strict mode: remove unused imports, fix Tooltip formatter types in CostAnalysis"

Fix build: add openai dep, add overrides for recharts/react19 peer deps, add .npmrc

Final README update with Docker deployment, test instructions, live benchmark, provider selector docs

Update Benchmark tab with live 'Run Benchmark Now' that calls API and populates real data"

Wire UniversalLLM into layers __init__ + add benchmark API endpoint for live runs

Add benchmark API route for live benchmark runs from dashboard

Add comprehensive unit tests for core functions

Add Dockerfile for one-command deployment

Add provider selector to LiveCompare + fix UI to support all 12 providers

Massive README update with 12 LLM providers, OpenClaw integration, Ollama support, full architecture docs

Update .env.example with all 12 LLM providers + Ollama + TigerGraph

Add Python universal LLM layer with LiteLLM supporting 12 providers + Ollama

Add multi-provider compare API route + providers listing endpoint