8 papers
MultihopSpatial: Multi-hop Compositional Spatial Reasoning Benchmark for Vision-Language Model
Youngwan Lee, Soojin Jang, Yoorhim Cho +3
Spatial reasoning is foundational for Vision-Language Models (VLMs), particularly when deployed as Vision-Language-Action (VLA) agents in physical environments. However, existing b…
HoliSafe: Holistic Safety Benchmarking and Modeling for Vision-Language Model
Youngwan Lee, Kangsan Kim, Kwanyong Park +5
Despite emerging efforts to enhance the safety of Vision-Language Models (VLMs), current approaches face two main shortcomings. 1) Existing safety-tuning datasets and benchmarks on…
Delta Attention: Fast and Accurate Sparse Attention Inference by Delta Correction
Jeffrey Willette, Heejun Lee, Sung Ju Hwang
The attention mechanism of a transformer has a quadratic complexity, leading to high inference costs and latency for long sequences. However, attention matrices are mostly sparse,…
Training-Free Exponential Context Extension via Cascading KV Cache
Jeffrey Willette, Heejun Lee, Youngwan Lee +2
The transformer's context window is vital for tasks such as few-shot learning and conditional generation as it preserves previous tokens for active memory. However, as the context…
InfiniteHiP: Extending Language Model Context Up to 3 Million Tokens on a Single GPU
Heejun Lee, Geon Park, Jaduk Suh +1
In modern large language models (LLMs), handling very long context lengths presents significant challenges as it causes slower inference speeds and increased memory costs. Addition…
A Training-free Sub-quadratic Cost Transformer Model Serving Framework With Hierarchically Pruned Attention
Heejun Lee, Geon Park, Youngwan Lee +7
In modern large language models (LLMs), increasing the context length is crucial for improving comprehension and coherence in long-context, multi-modal, and retrieval-augmented lan…