Reaction
Generalist LLMs are claimed to be competitive with Jev zero-shot
A member of the paper’s team says a little fine-tuning takes the models further.
TLDR
A member of the paper’s team claims generalist LLMs are already competitive with Jev zero-shot and says a little fine-tuning takes them further. In a reply, another user asks whether fine-tuning an LLM, possibly with LoRA, and reading its last-state logits would suffice—or whether zero-shot performance is enough.
Combined views
78
1 Source, first seen ago