Shubhra K. Karmaker ("Santu")
146 posts
@karmake2
Assistant Professor of Computer Science at Auburn University. Communication Chair of ACL Rolling Reviews (ARR)
Do large language models think like humans? Are they also prone to human-like cognitive biases? We just launched a new ranking system of LLMs based on their ability to resist cognitive biases with a large-scale study of 2.8M+ responses across 8 well-known biases.
A big thanks to @eduardo_nlp for his collaboration and significant contributions to this work. Our paper introduces ALIGN-SIM, a novel task-free framework for evaluating and interpreting sentence embeddings based on various semantic similarity alignment criteria. 2/2
We are excited to collaborate with BangaLLM to introduce BongLLaMA, the first ever and, of course, the state-of-the-art Bangla LLM (Based on Meta LLaMA). Try it out. @ReviewAcl #EMNLP #ACL #NLP #BanglaAI @UCFComputerSci arxiv.org/abs/2410.21200
We are happy to release our text summarization metric, Sem-F1, for public use. huggingface.co/spaces/BridgeA… Sem-F1 achieves a 25% higher correlation with human judgments than Bertscore on the Overlap Summarization task. Benchmarking paper coming soon. @ReviewAcl
