LayerCake: Token-Aware Contrastive Decoding within Large Language Model Layers
Paper
•
2507.04404
•
Published
•
22
To read
-
-
70% Size, 100% Accuracy: Lossless LLM Compression for Efficient GPU Inference via Dynamic-Length Float
Paper • 2504.11651 • Published • 31 -
A Token is Worth over 1,000 Tokens: Efficient Knowledge Distillation through Low-Rank Clone
Paper • 2505.12781 • Published • 2 -
A Survey of Context Engineering for Large Language Models
Paper • 2507.13334 • Published • 261 -
Nougat: Neural Optical Understanding for Academic Documents
Paper • 2308.13418 • Published • 42 -
Efficient Agents: Building Effective Agents While Reducing Cost
Paper • 2508.02694 • Published • 86 -
Souper-Model: How Simple Arithmetic Unlocks State-of-the-Art LLM Performance
Paper • 2511.13254 • Published • 136