1 paper · 1 filter
Elyas Rashno, Farhana Zulkernine
LLM-based multimodal emotion reasoning systems perform well on standard benchmarks, but improving them through instruction tuning can be computationally expensive. Agentic pipeline…