HackerPapers

arXiv:2406.17642cs.CL

Banishing LLM Hallucinations Requires Rethinking Generalization

Johnny Li, Saksham Consul, Eda Zhou, James Wong, Naila Farooqui, Yuxin Ye, Nithyashree Manohar, Zhuxiaona Wei, Tian Wu, Ben Echols, Sharon Zhou, Gregory Diamos

Abstract

Despite their powerful chat, coding, and reasoning abilities, Large Language Models (LLMs) frequently hallucinate. Conventional wisdom suggests that hallucinations are a consequence of a balance between creativity and factuality, which can be mitigated, but not eliminated, by grounding the LLM in external knowledge sources. Through extensive systematic experiments, we show that these traditional approaches fail to explain why LLMs hallucinate in practice. Specifically, we show that LLMs augmented with a massive Mixture of Memory Experts (MoME) can easily memorize large datasets of random numbers. We corroborate these experimental findings with a theoretical construction showing that simple neural networks trained to predict the next token hallucinate when the training loss is above a threshold as it usually does in practice when training on internet scale data. We interpret our findings by comparing against traditional retrieval methods for mitigating hallucinations. We use our findings to design a first generation model for removing hallucinations -- Lamini-1 -- that stores facts in a massive mixture of millions of memory experts that are retrieved dynamically.

arXiv comments
I want to revisit some of the experiments in this paper, specifically figure 5
Journal reference
DOI
Subjects
cs.CLComputation and Language cs.AIArtificial Intelligence

On Hacker News

Linked by 3 identifiable commenters in 5 comments across 4 threads, from Aug 2024 to Jan 2025.

Comments linking this paper, per year.

Discussed in comments4 threads

  1. Memory makes computation universal, remember? 1 comment

    • I forgot one tech that’s actually commercialized by lamini.ai: https://arxiv.org/abs/2406.17642 https://medium.com/pythons-gurus/mixture-of-memory-experts-l... Re your reply I’ll add two things to “drive us to models that have emergent, reasoning abilities:” 1. It didn’t have…

      nickpsecurityView on HNParent context

  2. OpenAI, Google and Anthropic are struggling to build more advanced AI 2 comments

    • The brain solves that problem. It seems to involve memory and specialized regions. I found a few groups building hippocampus-like, research models. One had content-addressable memory. There was another one that claimed to get rid of hallucinations. They also said it takes 50-100…

      nickpsecurityView on HNParent context

    • Comments on that paper? PDF: [1] What they are measuring, it seems, is whether LLMs can be built which will retrieve a reliable known correct answer on request. That's an information retrieval problem, and, in fact, they solve it by adding "Memory Experts" which are basically…

      AnimatsView on HNParent context

  3. LLMs Will Always Hallucinate, and We Need to Live with This 1 comment

    • Disagree - https://arxiv.org/abs/2406.17642 We cover halting problem and intractable problems in the related work. Of course LLMs cannot give answers to intractable problems. I also don’t see why you should call an answer of “I cannot compute that” to a halting problem question…

      gdiamosView on HNParent context

  4. Anthropic publishes the 'system prompts' that make Claude tick 1 comment

    • We know that LLMs hallucinate, but we can also remove them. I’d love to see a future generation of a model that doesn’t hallucinate on key facts that are peer and expert reviewed. Like the Wikipedia of LLMs https://arxiv.org/pdf/2406.17642 That’s a paper we wrote digging into…

      gdiamosView on HNParent context

Submitted to Hacker News

Connected through shared HN threads, not inferred similarity or a recommendation.