LLM Compression Comparison Hub
3 papers - avg viability 4.3
Top Papers
- Shorter Thoughts, Same Answers: Difficulty-Scaled Segment-Wise RL for CoT Compression(7.0)
Compress chain-of-thought reasoning traces with difficulty-aware reinforcement learning to reduce token cost without sacrificing answer quality.
- Leech Lattice Vector Quantization for Efficient LLM Compression(3.0)
Leech Lattice Vector Quantization offers a novel approach to efficiently compress large language models using high-dimensional lattice structures.
- Only relative ranks matter in weight-clustered large language models(3.0)
A novel approach to compress large language models by focusing on the relative rank of weights.