GPT Neo: open-source GPT model, with pretrained 1.3B & 2.7B weight models

Re:GPT Neo: open-source GPT model, with pretrained 1.3B & 2.... by monkeydust on Sunday March 21, 2021 @06:16PM (#26534639)

If I wanted to build an support Q&A system using texts from support logs, training docs, transcribed videos etc etc (basically as much text about my product as I can get) would this model be a good start ?

[ Parent | on HN ]

Re:GPT Neo: open-source GPT model, with pretrained 1.3B & 2.... by malka on Sunday March 21, 2021 @06:53PM (#26534915)

You should look on hugging faces https://huggingface.co/

https://huggingface.co/transformers/task_summary.html#extrac... more precisely

[ Parent | on HN ]

Re:GPT Neo: open-source GPT model, with pretrained 1.3B & 2.... by kristoo on Tuesday March 23, 2021 @12:10PM (#26556680)

That would be one incredibly expensive Q&A system. As rightly suggested by others, some search + a finetuned Q&A system is what you need.

I've been building a library that makes solving tasks like conversational Q&A easy.

https://github.com/backprop-ai/backprop

If you need some pointers, we have example notebooks and I'm happy to chat about it as well. Would love to know if it helps!

[ Parent | on HN ]

Re:GPT Neo: open-source GPT model, with pretrained 1.3B & 2.... by FL33TW00D on Sunday March 21, 2021 @07:07PM (#26534990)

Depending on how much content you've got, this blog post from HuggingFace might be interesting: https://yjernite.github.io/lfqa.html

[ Parent | on HN ]

Re:GPT Neo: open-source GPT model, with pretrained 1.3B & 2.... by ggnore7452 on Sunday March 21, 2021 @08:00PM (#26535283)

also, for some quick and simple Q&A system. Haystack https://github.com/deepset-ai/haystack (essentially dense vector similarity on Elastic Search) looks pretty promising and supports whole pipeline.

[ Parent | on HN ]

Re:GPT Neo: open-source GPT model, with pretrained 1.3B & 2.... by monkeydust on Tuesday March 23, 2021 @12:11PM (#26556707)

Thanks that looks decent for quick start.

[ Parent | on HN ]