4 ms·
Updated research paper showing alpha = 2*rank for LoRA finetuning of LLMs gets much better results + other cool results!
by danielhanchen 2y ago
Updated research paper showing alpha = 2*rank for LoRA finetuning of LLMs gets much better results + other cool results!