9 papers
Accelerated Test-Time Scaling with Model-Free Speculative Sampling
Woomin Song, Saket Dingliwal, Sai Muralidhar Jayanthi +4
Language models have demonstrated remarkable capabilities in reasoning tasks through test-time scaling techniques like best-of-N sampling and tree search. However, these approaches…
Compress, Gather, and Recompute: REFORMing Long-Context Processing in Transformers
Woomin Song, Sai Muralidhar Jayanthi, Srikanth Ronanki +5
As large language models increasingly gain popularity in real-world applications, processing extremely long contexts, often exceeding the model's pre-trained context limits, has em…
LAWCAT: Efficient Distillation from Quadratic to Linear Attention with Convolution across Tokens for Long Context Modeling
Zeyu Liu, Souvik Kundu, Lianghao Jiang +5
Although transformer architectures have achieved state-of-the-art performance across diverse domains, their quadratic computational complexity with respect to sequence length remai…
Context Length Alone Hurts LLM Performance Despite Perfect Retrieval
Yufeng Du, Minyang Tian, Srikanth Ronanki +7
Large language models (LLMs) often fail to scale their performance on long-context tasks performance in line with the context lengths they support. This gap is commonly attributed…
Think Clearly: Improving Reasoning via Redundant Token Pruning
Daewon Choi, Jimin Lee, Jihoon Tack +7
Recent large language models have shown promising capabilities in long-form reasoning, following structured chains of thought before arriving at a final answer. However, we observe…
Mamba Drafters for Speculative Decoding
Daewon Choi, Seunghyuk Oh, Saket Dingliwal +9
Speculative decoding has emerged as a promising approach to accelerating large language model (LLM) generation using a fast drafter while maintaining alignment with the target mode…