Papers by Momchil Hardalov
Exploring Fine-Tuning for In-Context Retrieval and Efficient KV-Caching in Long-Context Language Models (2026.eacl-short)
Copied to clipboard
| Challenge: | Long-Context Language Models (LCLMs) can encode entire document collections, offering a strong alternative to retrieval-augmented generation (RAG). |
| Approach: | They propose to use LCLMs to encode documents with context windows of millions of tokens to improve their performance. |
| Outcome: | The proposed training strategies improve long-context performance and their robustness under compression techniques. |
A Survey on Stance Detection for Mis- and Disinformation Identification (2022.findings-naacl)
Copied to clipboard
| Challenge: | Understanding attitudes expressed in texts plays an important role in systems for detecting false information online, be it misinformation (unintentionally false) or disinformation (intentional false information). |
| Approach: | They examine the relationship between stance detection and mis- and disinformation detection online and examine the results of previous studies. |
| Outcome: | The proposed task is a component of fact-checking, rumour detection, and detecting previously fact- checked claims, and is compared with other related tasks such as argumentation mining and sentiment analysis. |
Correct, Concise and Complete: Multi-stage Training For Adaptive Reasoning (2026.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) increase test-time computation, often in the form of chain-of-thought (CoT) however, reasoning traces can become unnecessarily long, increasing computation costs without improving accuracy and sometimes even degrading performance. |
| Approach: | They propose a multi-stage efficient reasoning method that combines supervised fine-tuning with reinforcement learning using an adaptive length penalty. |
| Outcome: | The proposed method reduces response length by an average of 28% for 8B models and 40% for 32B models while incurring only minor performance drops of 1.6 and 2.5 points, respectively. |
bgGLUE: A Bulgarian General Language Understanding Evaluation Benchmark (2023.acl-long)
Copied to clipboard
Momchil Hardalov, Pepa Atanasova, Todor Mihaylov, Galia Angelova, Kiril Simov, Petya Osenova, Veselin Stoyanov, Ivan Koychev, Preslav Nakov, Dragomir Radev
| Challenge: | bgGLUE is a benchmark for evaluating language models on natural language understanding (NLU) tasks in Bulgarian. |
| Approach: | They propose to use a benchmark to evaluate language models on NLU tasks in Bulgarian. |
| Outcome: | The proposed model performs well on sequence labeling tasks, but there is room for improvement for tasks that require more complex reasoning. |
DEM: Distribution Edited Model for Training with Mixed Data Distributions (2024.emnlp-main)
Copied to clipboard
| Challenge: | Recent fine-tuning approaches for large language models require supervised finetun on diverse datasets and follow different distributions. |
| Approach: | They propose a distribution edited model that integrates models individually trained on each data source with the base model using basic element-wise vector operations. |
| Outcome: | The proposed model outperforms baseline models on a variety of benchmarks and is cheaper than standard data mixing methods. |
Diable: Efficient Dialogue State Tracking as Operations on Tables (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing systems for dialogue state tracking use the full dialogue history as input and generate the entire state from scratch at each dialogue turn. |
| Approach: | They propose a task formalisation that represents the dialogue state as a table and formalises it as 'table manipulation task' they represent the dialogue as if it were a list with all the slots and generate the entire state from scratch at each dialogue turn. |
| Outcome: | The proposed system outperforms existing systems while maintaining competitive accuracy. |
EXAMS: A Multi-subject High School Examinations Dataset for Cross-lingual and Multilingual Question Answering (2020.emnlp-main)
Copied to clipboard
| Challenge: | EXAMS is a benchmark dataset for cross-lingual and multilingual question answering for high school examinations. |
| Approach: | They propose to use EXAMS to evaluate cross-lingual and multilingual question answering for high school examinations. |
| Outcome: | The proposed model can be used to explore multilingual reasoning and knowledge transfer methods and pre-trained models in schools in different languages, which was not possible by now. |
CrowdChecked: Detecting Previously Fact-Checked Claims in Social Media (2022.aacl-main)
Copied to clipboard
| Challenge: | Existing systems to automate fact-checking lack credibility in the eyes of the users. |
| Approach: | They propose to perform automatic fact-checking by verifying whether an input claim has been fact- checked by professional fact- checkers and to return back an article that explains their decision. |
| Outcome: | The proposed method improves on the CLEF’21 CheckThat! test set by two points absolute. |
A Neighborhood Framework for Resource-Lean Content Flagging (2022.tacl-1)
Copied to clipboard
Sheikh Muhammad Sarwar, Dimitrina Zlatkova, Momchil Hardalov, Yoan Dinkov, Isabelle Augenstein, Preslav Nakov
| Challenge: | Existing approaches to cross-lingual content flagging with limited target language data are lacking in many languages. |
| Approach: | They propose a framework for cross-lingual content flagging with limited target- language data based on a nearest-neighbor architecture and a transformer representation in all its components. |
| Outcome: | The proposed framework outperforms previous work in terms of predictive performance on eight languages from two different datasets. |
Cross-Domain Label-Adaptive Stance Detection (2021.emnlp-main)
Copied to clipboard
| Challenge: | Stance detection is a task that focuses on the classification of a writer’s viewpoint towards a target. |
| Approach: | They propose an end-to-end unsupervised framework for out-of-domain prediction of unseen, user-defined labels. |
| Outcome: | The proposed framework shows that it can be used to predict unseen labels over strong baselines. |
Understanding and Improving Information Preservation in Prompt Compression for LLMs (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Recent advances in large language models have enabled their successful application to a broad range of tasks. |
| Approach: | They propose a framework that allows for in-depth analysis of prompt compression methods. |
| Outcome: | The proposed framework analyzes state-of-the-art soft and hard compression methods . it shows that some fail to preserve key details from the original prompt, limiting performance on complex tasks. |
Factual Confidence of LLMs: on Reliability and Robustness of Current Estimators (2024.acl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) tend to be unreliable on fact-based answers. |
| Approach: | They propose a framework for comparing LLMs' confidence over fact-based answers with hidden-state probes that are more reliable than hidden-status probes. |
| Outcome: | The proposed methods show that hidden-state probes provide the most reliable confidence estimates despite requiring access to weights and supervision data. |
DeepFact: Co-Evolving Benchmarks and Agents for Deep Research Factuality (2026.acl-long)
Copied to clipboard
Yukun Huang, Leonardo F. R. Ribeiro, Momchil Hardalov, Bhuwan Dhingra, Markus Dreyer, Venkatesh Saligrama
| Challenge: | Existing fact-checkers usually target general-domain atomic claims . citation-grounded fact- checking ignores claims without explicit citations . |
| Approach: | They propose to use a benchmark to test whether claim-level factuality is transferable . they instantiate **Audit-then-Score** as a versioned DRR factualism benchmark . |
| Outcome: | The proposed benchmark outperforms the best prior deep-research and traditional fact-checkers by 14.3 and 24.9 points. |