Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Teach your LLM to answer with facts, not fiction

blog.myscale.com

Post reply on HN
Print view View original

11–20 of 149 posts

Re: Teach your LLM to answer with facts, not fiction

#11

Post by jwells89 »

potatoman22 wrote: post #9
wizzwizz4 wrote: post #2

Situation: people try to use these predictive text chatbots as search engines. Problem: LLMs are not search engines. They extrapolate, interpolate, and approximate (so-called “hallucinations”) so they can always produce somewhat-plausible text completions. Solution: Create a search engine so good at returning relevant results that even an LLM can make use of it… then go to significant lengths to plug that search engi…

Well, search engines and my ability to query them aren't good enough for highly specific or poorly worded questions yet. LLMs are sometimes better in this space.

Indeed, a frequent issue with search engines is knowing the right terms plug into them, particularly when researching a topic beyond one's scope of knowledge.

Half the power of LLMs as they currently exist is that they can often extract the intention of the user's question in a way that search engines usually can't, allowing them to provide a more useful answer or at least point the user in the right direction.

Perhaps it would make sense for search engines to utilize LLMs to perform this query extraction and suggest more appropriate search terms, engaging conversational interaction only if the suggestions are wrong and the LLM requires further clarification.

Re: Teach your LLM to answer with facts, not fiction

#12

Post by kmeisthax »

One of the fun things I like to do with these examples is follow along at home. I don't have access to GPT-4 because I refuse to give money to OpenAI, but with GPT-3.5, "What is an LLM hallucination" actually gives:

- The usual knowledge cutoff warning

- An explanation of what a hallucination is

If (in a separate conversation) I give GPT-3.5 the exact prompt that explains what an LLM is, I get gaslighted instead. GPT-3.5 attempts to tell me that LLM stands for "Legal Master of Laws". Then it gives the knowledge cutoff warning, and then the same correct explanation Myscale got.

The rest of this article appears to be trying to turn GPT into a frontend for search engines. I don't know why people keep trying to do this.

Re: Teach your LLM to answer with facts, not fiction

#13

Post by mpsk »

teej wrote: post #8
lqhl wrote: post #6

Earlier quoted context omitted.

This article suggests that LLMs should use a database as a reference for factual information. Rather than asking LLMs to provide their own answers, it is recommended that they summarize based on the facts extracted from the database. This approach reduces the likelihood of hallucinations among LLMs.

> This approach reduces the likelihood of hallucinations among LLMs. This has not been my experience. Did you create any benchmarks as a part of this project?

I am the author of this article. And actually what we tried to do was to replicate the simplest implementation to Retrieval Augmented Language Models by prompting the LLM. There have been many researches on this topic right now like work from Meta(https://arxiv.org/pdf/2208.03299v3.pdf). I think it can give you a picture how those RALMs boost the performance on General QA tasks.

Re: Teach your LLM to answer with facts, not fiction

#15

Post by lqhl »

teej wrote: post #8
lqhl wrote: post #6

Earlier quoted context omitted.

This article suggests that LLMs should use a database as a reference for factual information. Rather than asking LLMs to provide their own answers, it is recommended that they summarize based on the facts extracted from the database. This approach reduces the likelihood of hallucinations among LLMs.

> This approach reduces the likelihood of hallucinations among LLMs. This has not been my experience. Did you create any benchmarks as a part of this project?

This idea is a simplified version of Retrieval-Augmented Generation (RAG), and RAG has been studied in various research papers, such as the one available at https://arxiv.org/abs/2005.11401

Re: Teach your LLM to answer with facts, not fiction

#16

Post by wmf »

wizzwizz4 wrote: post #2

Situation: people try to use these predictive text chatbots as search engines. Problem: LLMs are not search engines. They extrapolate, interpolate, and approximate (so-called “hallucinations”) so they can always produce somewhat-plausible text completions. Solution: Create a search engine so good at returning relevant results that even an LLM can make use of it… then go to significant lengths to plug that search engi…

Yeah, this "fix" just shifts the problem to the vector DB and its embedding algorithm. People keep forgetting to mention that embeddings aren't 100% accurate either. The net accuracy may be better but it's not magic.

Re: Teach your LLM to answer with facts, not fiction

#17

Post by akomtu »

12907835202 wrote: post #14
vouaobrasil wrote: post #5

I believe that LLMs should be banned, but if they have to exist, we should teach them ethics first before anything else.

Whose ethics? Should we tell it that sex is a positive thing or a horrible sin?

Those crusaders that prevail are the most ethical.

Re: Teach your LLM to answer with facts, not fiction

#18

Post by woeirua »

omeze wrote: post #7
wizzwizz4 wrote: post #2

Situation: people try to use these predictive text chatbots as search engines. Problem: LLMs are not search engines. They extrapolate, interpolate, and approximate (so-called “hallucinations”) so they can always produce somewhat-plausible text completions. Solution: Create a search engine so good at returning relevant results that even an LLM can make use of it… then go to significant lengths to plug that search engi…

Yea… my opinion on this is that startups are attempting to force a market for chat bots, instead of accepting that LLM embeddings are best utilized as a search feature, not new product surface area

They really want agents to work, and they just don’t yet.

Re: Teach your LLM to answer with facts, not fiction

#19

Post by lalopalota »

jwells89 wrote: post #11
potatoman22 wrote: post #9

Earlier quoted context omitted.

Well, search engines and my ability to query them aren't good enough for highly specific or poorly worded questions yet. LLMs are sometimes better in this space.

Indeed, a frequent issue with search engines is knowing the right terms plug into them, particularly when researching a topic beyond one's scope of knowledge. Half the power of LLMs as they currently exist is that they can often extract the intention of the user's question in a way that search engines usually can't, allowing them to provide a more useful answer or at least point the user in the right direction. Perha…

I agree. Use the LLM to construct a better search query, then return those results and the query to the user. The user can modify the "optimized" query and repeat until a usefull answer is obtained.

Re: Teach your LLM to answer with facts, not fiction

#20

Post by lqhl »

kmeisthax wrote: post #12

One of the fun things I like to do with these examples is follow along at home. I don't have access to GPT-4 because I refuse to give money to OpenAI, but with GPT-3.5, "What is an LLM hallucination" actually gives: - The usual knowledge cutoff warning - An explanation of what a hallucination is If (in a separate conversation) I give GPT-3.5 the exact prompt that explains what an LLM is, I get gaslighted instead. GPT…

As jwells89 mentioned, LLMs can extract the intention from questions and generate better queries for a search engine or database.
Post reply on HN