Yi-Coder: A Small but Mighty LLM for Code

DeepMind’s GenRM improves LLM accuracy by having models verify their own outputs

Code in pre-training data improves LLM performance at non-coding tasks

The historical distinction between coder and programmer