Papers by Kaushik Roy
Segmented Recurrent Transformer: An Efficient Sequence-to-Sequence Model (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Transformers have shown dominant performance across a range of domains including language and vision, but their computational cost grows quadratically with the sequence length, making their usage prohibitive for resource-constrained applications. |
| Approach: | They propose a segmented recurrent transformer that combines segmente recursion with recursive attention to reduce the computational cost. |
| Outcome: | The proposed model achieves higher ROUGE1 scores and lower computational complexity than current approaches. |
Prompt-Based Bias Calibration for Better Zero/Few-Shot Learning of Language Models (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Prompt-based learning is susceptible to intrinsic bias present in pre-trained language models (LMs), leading to sub-optimal performance in prompt-based zero/few-shot settings. |
| Approach: | They propose a null-input prompting method to calibrate intrinsic bias encoded in pre-trained language models (LMs) they leverage a diverse set of auto-selected null meaning inputs generated from GPT-4 to probe intrinsic bias. |
| Outcome: | The proposed method significantly improves zero/few-shot learning performance of LMs for both in-context learning and prompt-based fine-tuning (on average 9% and 2%, respectively). |
LogicTree: Structured Proof Exploration for Coherent and Rigorous Logical Reasoning with Large Language Models (2025.emnlp-main)
Copied to clipboard
| Challenge: | Large language models (LLMs) have remarkable multi-step reasoning capabilities, but they still face challenges in complex logical reasoning. |
| Approach: | They propose an algorithm-guided search framework that automates structured proof exploration and ensures logical coherence. |
| Outcome: | The proposed framework outperforms o3-mini and chain-of-thought with average gains of 23.6% and 12.5% on five datasets. |
Eigen Attention: Attention in Low-Rank Space for KV Cache Compression (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Large language models (LLMs) have been increasing context lengths to enhance their performance, but at long context length, the KV cache becomes the new bottleneck in memory usage during inference. |
| Approach: | They propose an approach which performs the attention operation in a low-rank space and reduces the KV cache memory overhead. |
| Outcome: | The proposed approach reduces the KV cache memory overhead and reduces memory usage with minimal drop in performance over OPT, MPT, and Llama model families. |
ELLA: Efficient Lifelong Learning for Adapters in Large Language Models (2026.eacl-long)
Copied to clipboard
| Challenge: | Existing approaches to training Large Language Models (LLMs) suffer from catastrophic forgetting when adapted sequentially to new tasks in a continual learning (CL) setting. Existing methods are impractical and could potentially violate privacy. |
| Approach: | They propose a training framework built on the principle of selective subspace de-correlation that characterizes the structure of past updates and penalizes alignments along their high-energy, task-specific directions. |
| Outcome: | The proposed training framework achieves state-of-the-art CL performance on three popular benchmarks spanning both classification and generative tasks with relative accuracy gains of up to 9.6% and a 35 smaller memory footprint. |