Fine-tune small foundational LLM on typical large enterprise use-case and compare results with pre-trained and large scale models. Instruction dataset will be generated artificially with a SOTA LLM.
2
stars
0
forks
source link
Theory: Performing three categorisation tasks at once limits fine-tuning performance #8
-> Try doing three single classifications per sample