Papers by Dhruv Kumar
Optimizing Deeper Transformers on Small Datasets (2021.acl-long)
Copied to clipboard
Peng Xu, Dhruv Kumar, Wei Yang, Wenjie Zi, Keyi Tang, Chenyang Huang, Jackie Chi Kit Cheung, Simon J.D. Prince, Yanshuai Cao
| Challenge: | a common belief that training deep transformers from scratch requires large datasets is wrong . however, with proper initialization and optimization, the benefits of very deep transformer can carry over to challenging tasks with small datasets. |
| Approach: | They train 48 layers of transformers from pre-trained RoBERTa and 24 relation-aware layers from scratch. |
| Outcome: | The proposed scheme achieves state-of-the-art performance on a text-to-sql parsing benchmark . it uses 24 fine-tuned layers from pre-trained RoBERTa and 24 relation-aware layers from scratch . |
mEdIT: Multilingual Text Editing via Instruction Tuning (2024.naacl-long)
Copied to clipboard
| Challenge: | mEdIT is a multi-lingual extension to CoEdit for writing assistance. |
| Approach: | They propose to train multi-lingual large language models (LLMs) by fine-tuning them via instruction tuning. |
| Outcome: | The proposed model performs well on multilingual text editing benchmarks and generalizes well to new languages. |
A Hybrid Supervised-LLM Pipeline for Actionable Suggestion Mining in Unstructured Customer Reviews (2026.eacl-industry)
Copied to clipboard
| Challenge: | Existing approaches to extract actionable suggestions from customer reviews are often mixed-intent, unstructured text. |
| Approach: | They propose a hybrid pipeline that uses a RoBERTa classifier and a precision–recall surrogate to extract actionable suggestions from customer reviews. |
| Outcome: | The proposed pipeline outperforms prompt-only, rule-based, and classifier-only baselines in extraction accuracy and cluster coherence. |
ContraDoc: Understanding Self-Contradictions in Documents with Large Language Models (2024.naacl-long)
Copied to clipboard
| Challenge: | Detecting contradictions in texts is often regarded as determining relation between hypothesis and piece of premise. |
| Approach: | They propose a human-annotated dataset to study self-contradictions in long documents . they analyze the capabilities of four open-source and commercially available LLMs . |
| Outcome: | The proposed dataset outperforms open-source LLMs on document-level tasks but struggles with self-contradictions that require more nuance and context. |
CoEdIT: Text Editing by Task-Specific Instruction Tuning (2023.findings-emnlp)
Copied to clipboard
| Challenge: | We present a large language model for writing assistance that is fine-tuned on task-specific instructions. |
| Approach: | They propose a large language model that is fine-tuned on task-specific instructions and outputs the edited text. |
| Outcome: | The proposed model performs better than other state-of-the-art models on various editing benchmarks while being 60x smaller. |
Speakerly: A Voice-based Writing Assistant for Text Composition (2023.emnlp-industry)
Copied to clipboard
Dhruv Kumar, Vipul Raheja, Alice Kaiser-Schatzlein, Robyn Perry, Apurva Joshi, Justin Hugues-Nuger, Samuel Lou, Navid Chowdhury
| Challenge: | Speakerly TM is a voice-based writing assistance system that works across the different stages of writing. |
| Approach: | They propose a voice-based writing assistance system that helps users with text composition across various use cases such as emails, instant messages, and notes. |
| Outcome: | The proposed system can be used for email, instant messages, and notes. |
Improving Iterative Text Revision by Learning Where to Edit from Other Revision Tasks (2022.emnlp-main)
Copied to clipboard
| Challenge: | Iterative text revision improves text quality by fixing grammatical errors, rephrasing for better readability or contextual appropriateness. |
| Approach: | They propose to build an end-to-end text revision system that can iteratively generate helpful edits by explicitly detecting editable spans with their corresponding edit intents. |
| Outcome: | The proposed system outperforms baselines on other text revision tasks and human evaluations. |
Iterative Edit-Based Unsupervised Sentence Simplification (2020.acl-main)
Copied to clipboard
| Challenge: | Sentence simplification is relevant in various real-world and downstream applications. |
| Approach: | They propose an edit-based approach to unsupervised sentence simplification that uses a scoring function to score fluency, simplicity, and meaning preservation to perform edits. |
| Outcome: | The proposed model is more controllable and interpretable than state-of-the-art models on newsela and WikiLarge datasets. |
Understanding Iterative Revision from Human-Written Text (2022.acl-long)
Copied to clipboard
| Challenge: | This work describes IteraTeR: the first large-scale, multi-domain, edit-intention annotated corpus of iteratively revised text. |
| Approach: | They propose to annotate iteratively revised text using a multi-domain annotated corpus that generalizes to a variety of domains, edit intentions, revision depths, and granularities. |
| Outcome: | The proposed model improves automatic evaluations by integrating edit intentions with writing quality. |
Investigating Pedagogical Teacher and Student LLM Agents: Genetic Adaptation Meets Retrieval-Augmented Generation Across Learning Styles (2025.emnlp-main)
Copied to clipboard
Debdeep Sanyal, Agniva Maiti, Umakanta Maharana, Dhruv Kumar, Ankur Mali, C. Lee Giles, Murari Mandal
| Challenge: | Existing models for large language models neglect comprehensive student modeling beyond basic knowledge states and lack mechanisms for teachers to dynamically adapt their approach based on student feedback and collective performance. |
| Approach: | They propose a framework that integrates LLM-based diverse student agents with a self-evolving teacher agent to optimize teacher's pedagogical parameters based on simulated student performance. |
| Outcome: | The proposed framework integrates diverse student agents with a self-evolving teacher agent to optimize teacher pedagogical parameters based on simulated student performance. |
From Trust to Compromise: Outcome-Verified LLM Phishing Simulation and Real-Time Defense (2026.acl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) excel as conversational agents, but existing simulators focus on PII requests within the chat. |
| Approach: | They propose a large language model that generates human-like language and maintains conversational context to automate social engineering attacks. |
| Outcome: | The proposed model improves dialogue-level detection over a real-time baseline. |
ReviewEval: An Evaluation Framework for AI-Generated Reviews (2025.findings-emnlp)
Copied to clipboard
| Challenge: | escalating volume of academic research necessitates innovative approaches to peer review . authors propose reviewEval, ReviewAgent and ReviewEval to improve on existing reviews . |
| Approach: | They propose a framework for AI-generated reviews that measures alignment with human assessments . they propose 'reviewAgent' that iteratively optimizes its intermediate outputs and external improvement loops . |
| Outcome: | The proposed framework improves actionable insights and analytical depth by 6.78% and 47.62% over baselines and expert reviews. |
GRS: Combining Generation and Revision in Unsupervised Sentence Simplification (2022.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for sentence simplification are supervised or unsupervised . paraphrasing captures complex edit operations, while revision-based methods provide more control and interpretability. |
| Approach: | They propose an unsupervised approach to sentence simplification that combines text generation and text revision. |
| Outcome: | The proposed method improves on the Newsela and ASSET datasets. |