8000
"Den2MoEE: Domain-Aware Dense-to-MoE Reconstruction for Efficient LLM Embedding Models."
"DelvePO: Direction-Guided Self-Evolving Framework for Flexible Prompt Optimization". This work presents a task-agnostic, component-customizable and flexible framework to optimize the prompts in a self-evolve manner.
The official implementation of paper *Automatic Multi-Task Learning Framework with Neural Architecture Search in Recommendations*
Loading…
Loading…