Extrinsic Hallucinations in LLMs

1. 1Identify cases of hallucination in LLM output
2. 2Distinguish between in-context and extrinsic hallucinations
3. 3Analyze the role of context and world knowledge in mitigating hallucinations
4. 4Develop strategies to mitigate extrinsic hallucinations in LLMs

NLP researchers and AI engineers working with LLMs can benefit from understanding hallucinations to improve model performance and reliability

💡 Extrinsic hallucinations are a specific type of hallucination that can be addressed by improving model understanding of context and world knowledge

🤖 LLMs can hallucinate! Extrinsic hallucinations occur when output is fabricated & not grounded in context or world knowledge