EvalYaks: Instruction Tuning Datasets and LoRA Fine-tuned Models for Automated Scoring of CEFR B2 Speaking Assessment Transcripts

Nicy Scaria, Silvester John Joseph Kennedy, Thomas Latinovich +1

Relying on human experts to evaluate CEFR speaking assessments in an e-learning environment creates scalability challenges, as it limits how quickly and widely assessments can be c…