Recent research has shed light on the impressive capabilities of Large Language Models (LLMs) trained on Code for various software engineering tasks. These models can be classified into three main paradigms, namely code LLMs specialized in code completion, task-specific code LLMs fine-tuned for specific tasks, and instruction-tuned code LLMs adept at following human instructions and excelling in new tasks without additional fine-tuning.… Read the rest
The Impact of Instruction-Tuned Code Language Models on Software Engineering Tasks
