search

asahi417 / lm-question-generation

Multilingual/multidomain question generation datasets, models, and python library for question generation.
https://www.autoqg.net
MIT License
324 stars 31 forks source link

LM vs LLM Generation Comparison #13

Closed xiaoxixxxx closed 1 year ago

xiaoxixxxx commented 1 year ago

Hey, thank you for your impressive work.

I wonder whether you have considered comparing the generation with LM to the generation with LLM?

AntonOfTheWoods commented 1 year ago

@asahi417 , put another way, is there anything stopping us from fine-tuning say, llama 2, for QAG with the provided datasets? I am interested in doing this as part of my PhD... :-).

asahi417 commented 1 year ago

Hi there, i'm quite open for the latest LLM fine-tuning, yet LMQG currently support encoder-decoder LM fine-tuning only, and cannot be used to fine-tune decoder-only LMs including llama and gpt families.