1 paper
Josh McGiff, Salma Mekaoui, Robert Shanahan +1
Instruction-tuned LLMs are deployed into environments where domains evolve, yet extending a fine-tuned model's capabilities without full retraining remains an unsolved practical ch…