Spaces:
Sleeping
LLM backbone evaluation: cost + hallucination for web QA models
Hi 👋
Web QA is a great benchmark for real-world LLM capability. For teams deploying web QA systems, hallucination rate is critical — a model that fabricates web content is worse than no model.
I built an open source LLM Evaluation Framework that measures:
→ 🔍 Hallucination Rate — detects fabricated facts and overconfident wrong answers
→ 🧠 Reasoning Quality — scores whether models show their web reasoning process
→ 💰 Cost per 1K tokens — web QA can be token-heavy with context
→ ⚡ Latency p95 — for real-time web search + answer use cases
→ 🎯 Accuracy — 4-strategy cascade scorer
Live demo: https://huggingface.co/spaces/vigneshwar234/llm-eval-demo
GitHub: https://github.com/vignesh2027/LLM-Evaluation-Framework
Open source. Free forever. Happy to discuss web QA evaluation approaches!