You can not get a non-hallucinating AI in 2023.
Could you expand on the reasons and possibly relevant papers?
Ask HN: How do I train a custom LLM/ChatGPT on my own documents in Dec 2023?
181–190 of 247 posts
Re: Ask HN: How do I train a custom LLM/ChatGPT on my own documents in Dec 2023?
#182Re: Ask HN: How do I train a custom LLM/ChatGPT on my own documents in Dec 2023?
#183I have a question for which I haven't found a definitive answer yet: is how can one effectively manage typos and Out-of-Vocabulary (OOV) words in RAG systems? For instance, if I search for a specific product name but accidentally mistype it, the resulting encoded vector might not be similar to the vector for the correctly spelled product name ?
Good embeddings are supposed to handle this natively. For example the embedding for API should be close to the embedding for Application Programming Interface. Purportedly.
Re: Ask HN: How do I train a custom LLM/ChatGPT on my own documents in Dec 2023?
#184You can not get a non-hallucinating AI in 2023.
It seems like there's not much point to LLM+RAG over a good local search engine.
At least, that's what I've rapidly been concluding.
I'm still holding out a bit of hope for there being more though.
Re: Ask HN: How do I train a custom LLM/ChatGPT on my own documents in Dec 2023?
#185You can pay for a ChatGPT account and upload your own documents. I didn't do this myself but my dad uploaded 6 years of sermon transcripts from our church. It sounds exactly like the pastor.
Re: Ask HN: How do I train a custom LLM/ChatGPT on my own documents in Dec 2023?
#186I'm curious about this as well but my data is mostly (95%) numerical metrics. Is there a "RAG" mechanism for numerical data instead of text? My use case is data analysis, insight discovery for example.
im not an expert, but take a look at the llamaindex structured data retrieval examples: https://docs.llamaindex.ai/en/stable/understanding/putting_i...
There are also examples that combine semantic retrieval and structured retrieval.
Re: Ask HN: How do I train a custom LLM/ChatGPT on my own documents in Dec 2023?
#187Earlier quoted context omitted.
Or just using a traditional search engine and "rag" plus literally any ML/AI/LLM term will yield a half dozen results at the top with "Retrieval-augmented generation" in the page title.
Or people could just not use obscure acronyms when discussing specialised topics on an open forum?
Where do you draw the line though?
Re: Ask HN: How do I train a custom LLM/ChatGPT on my own documents in Dec 2023?
#188I'd very much appreciate if someone could clarify what exactly is needed in terms of hardware and software to implement these suggestions.
Would a five years old laptop work? Do you need a beefy GPU? Are you using some prepackaged software?
Re: Ask HN: How do I train a custom LLM/ChatGPT on my own documents in Dec 2023?
#189And what’s the correct answer in December 2023 if one wants to narrow down only to tools and services provided on Azure?
https://github.com/microsoft/semantic-kernel Semantic Kernel is MS's response to LangChain and LlamaIndex - available for .NET, python and Java. Using their Memory support (using Azure Cognitive Search), it gives a powerful RAG quickstart, which you can combined with Azure Document Intelligence to chunk your source documentation into memories that your foundational models can later use. (Disclaimer: Its only very rec…
Thanks for the link and good luck with your contributions to this project
Re: Ask HN: How do I train a custom LLM/ChatGPT on my own documents in Dec 2023?
#190Try https://github.com/SecureAI-Tools/SecureAI-Tools -- it's an open-source application layer for Retrieval-Augmented Generation (RAG). It allows you to use any LLM -- you can use OpenAI APIs, or run models locally with Ollama.