10 papers
Coverage-Driven KV Cache Eviction for Efficient and Improved Inference of LLM
Shuvendu Roy, Mengyao Zhai, Hossein Hajimirsadeghi +1
Large language models (LLMs) excel at complex tasks like question answering and summarization, thanks to their ability to handle long-context inputs. However, deploying LLMs is cos…
You Need Reasoning to Learn Reasoning: The Limitations of Label-Free RL in Weak Base Models
Shuvendu Roy, Hossein Hajimirsadeghi, Mengyao Zhai +1
Recent advances in large language models have demonstrated the promise of unsupervised reinforcement learning (RL) methods for enhancing reasoning capabilities without external sup…
Consistency-Guided Asynchronous Contrastive Tuning for Few-Shot Class-Incremental Tuning of Foundation Models
Shuvendu Roy, Elham Dolatabadi, Arash Afkanpour +1
We propose Consistency-guided Asynchronous Contrastive Tuning (CoACT), a novel method for continuously tuning foundation models to learn new classes in few-shot settings. CoACT con…
Advancing Medical Representation Learning Through High-Quality Data
Negin Baghbanzadeh, Adibvafa Fallahpour, Yasaman Parhizkar +8
Despite the growing scale of medical Vision-Language datasets, the impact of dataset quality on model performance remains under-explored. We introduce Open-PMC, a high-quality medi…
A Shared Encoder Approach to Multimodal Representation Learning
Shuvendu Roy, Franklin Ogidi, Ali Etemad +2
Multimodal representation learning has demonstrated remarkable potential in enabling models to process and integrate diverse data modalities, such as text and images, for improved…
Task-agnostic Prompt Compression with Context-aware Sentence Embedding and Reward-guided Task Descriptor
Barys Liskavets, Shuvendu Roy, Maxim Ushakov +3
The rise of Large Language Models (LLMs) has led to significant interest in prompt compression, a technique aimed at reducing the length of input prompts while preserving critical…