Fact Checking Rocks! - How to build a fact checking system for rock music
summary
A working baseline fact‑checker for rock music that indexes Wikipedia passages, encodes them with sentence‑transformers, performs semantic retrieval via FAISS and scores claims with an entailment classifier built from Hugging Face Transformers. It returns the top passages, an entailment/contradiction/neutral probability and a summary score, and runs as a web demo on Hugging Face Spaces demo. Uses Haystack + FAISS + Transformers, accepts a single claim, and is limited to Wikipedia sources and no automated claim detection.
AI-written summary · about
chapters (11)
- 0:00Introduction to fact checking
- 1:35Overview of the system
- 3:55Building the fact checking pipeline
- 9:47Custom entailment checker
- 13:38Demo and results
- 15:23Using large language models
- 18:41Future improvements
- 20:08Agents with prompt nodes
- 23:28Privacy with open models
- 25:25Retrieval model importance
- 28:44Q&A and conclusion
related talks
35mYou've got questions. We've got answers! #bbuzz
44mThe Limits of Imagination : An Open Source journey
34mApplying Machine Learning Techniques to Detect and Understand Fake News
14mPlease (Don't) Stop the Music: Adversarial Red-Teaming of AI...
22mThe Contradiction bot
38mSentence embeddings for automated factchecking
42mApplying Responsible AI with Open-Source Tools
27mPythonic Metal