josh curry
jdcurry
ยท
AI & ML interests
EM. GPT-2 Small & GPT-Neo 1.3B for fine tuning at the moment. I am working on building relevant datasets for what I'm trying to accomplish. (IA/PA Grants as the initial test) I have completed a fair amount of document scraping but will need to do more. We have the CUDA cores and memory required for both models, but we must focus on batch sizes, data loading pipelines, and optimization.