Papers by Julia Hockenmaier
A Framework for Bidirectional Decoding: Case Study in Morphological Inflection (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing encoder-decoders that generate sequences from left to right are prone to errors due to the "snowballing" effect. |
| Approach: | They propose a transformer-based encoder-decoder framework that produces sequences from the "outside-in" they argue that this approach is more principled than prior bidirectional decoders . |
| Outcome: | The proposed model beats the current system by over 4.7 and 2.7 points in accuracy on 2022 and 2023 tasks. |
Toward Efficient Sparse Autoencoder-Guided Steering for Improved In-Context Learning in Large Language Models (2025.emnlp-main)
Copied to clipboard
| Challenge: | Sparse autoencoders (SAEs) have emerged as a powerful analytical tool in mechanistic interpretability for large language models (LLMs). |
| Approach: | They propose a novel approach that leverages SAEs to enhance the general in-context learning performance of large language models (LLMs). |
| Outcome: | The proposed method yields a 3.5% improvement across diverse text classification tasks and exhibits greater robustness to hyperparameter variations compared to standard steering approaches. |
Entailment-Preserving First-order Logic Representations in Natural Language Entailment (2025.acl-long)
Copied to clipboard
| Challenge: | First-order logic (FOL) is often used to represent logical entailment, but determining natural language (NL) enanglement using FOL remains a challenge. |
| Approach: | They propose an Entailment-Preserving FOL representations task and a method which trains an NL-to-FOL translator by using the natural language entailment labels as verifiable rewards. |
| Outcome: | The proposed method achieves 1.8–2.7% improvement in EPR and 17.4–20.6% increase in E PR@16 compared to baselines in three datasets. |
Scaling Evaluation-Time Compute with Reasoning Models as Evaluators (2026.findings-acl)
Copied to clipboard
Seungone Kim, Ian Wu, Jinu Lee, Xiang Yue, Seongyun Lee, Minkyeong Moon, Carolin Lawrence, Kiril Gashteovski, Julia Hockenmaier, Graham Neubig, Sean Welleck
| Challenge: | Language model (LM) evaluators that generate chain-of-thought reasoning are widely used for the assessment of LM responses. |
| Approach: | They investigate whether increasing LMs' "thinking" time through scaling test-time compute can improve an LM's evaluation capability. |
| Outcome: | The proposed reasoning models improve evaluation performance monotonically with the number of reasoning tokens generated, mirroring trends seen in LM reasoning. |
Tutor-ICL: Guiding Large Language Models for Improved In-Context Learning Performance (2024.findings-emnlp)
Copied to clipboard
| Challenge: | In-context learning (ICL) is a dominant paradigm in natural language processing. |
| Approach: | They propose a prompting method for classification tasks using exemplar answers in a *comparative format' they also propose introducing a test instance before the exemplars to improve performance . |
| Outcome: | The proposed method achieves up to 13.76% increase in accuracy on classification tasks across decoder-only and encoder-decoder LLMs. |
A Multi-Perspective Architecture for Semantic Code Search (2020.acl-main)
Copied to clipboard
| Challenge: | Existing models do not model interactions between code and description until the final step when their global similarity is calculated. |
| Approach: | They propose a multi-perspective cross-lingual neural framework for code–text matching that captures both global and local similarities. |
| Outcome: | The proposed model performs better on the CoNaLa dataset than previous approaches that map code and text to a single joint embedding space. |
On the Versatility of Sparse Autoencoders for In-Context Learning (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Sparse autoencoders (SAEs) are emerging as a key analytical tool in interpretability for large language models. |
| Approach: | They propose to use SAEs to extract knowledge from billions of tokens for sparse reconstruction. |
| Outcome: | The proposed model can extract knowledge from billions of tokens for sparse reconstruction. |
Collaborative Dialogue in Minecraft (P19-1)
Copied to clipboard
| Challenge: | Using computer games to simulate grounded situations, we want to develop interactive agents that can communicate with humans to solve tasks in grounded scenarios. |
| Approach: | They propose a Minecraft-based collaborative building task in which one player is shown a building structure and needs to instruct the other player to build it. |
| Outcome: | The proposed agent can communicate with humans to solve a building task in a Minecraft-based environment without the need for physical robots. |
Evaluating Step-by-step Reasoning Traces: A Survey (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing evaluation practices are inconsistent, resulting in fragmented progress across evaluator design and benchmark development. |
| Approach: | a survey provides a comprehensive overview of step-by-step reasoning evaluation . existing evaluation practices are inconsistent, resulting in fragmented progress . |
| Outcome: | The proposed evaluation criteria are based on four top-level categories . the results are presented in a systematic review of the literature. |
Tackling Distractor Documents in Multi-Hop QA with Reinforcement and Curriculum Learning (2026.findings-eacl)
Copied to clipboard
| Challenge: | Existing work on retrieval-augmented generation systems has shown that retrievers exhibit imperfect recall and precision, limiting downstream performance. |
| Approach: | They propose a retrieval-augmented generation model that generates answers from larger sets of retrieved contexts. |
| Outcome: | The proposed model generates answers and cites relevant information from larger sets of retrieved contexts. |
Analyzing the Performance of Large Language Models on Code Summarization (2024.lrec-main)
Copied to clipboard
| Challenge: | Large language models perform very well on tasks that involve both natural language and source code. |
| Approach: | They show that large language models perform very well on tasks that involve both natural language and source code. |
| Outcome: | The proposed models perform very well on tasks that involve both natural language and source code. |
Rating Roulette: Self-Inconsistency in LLM-As-A-Judge Frameworks (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Using large language models (LLMs) for evaluating natural language generation has gained traction . lm judges have low intra-rater reliability in their assigned scores, making it difficult to measure how good their judgments actually are. |
| Approach: | They show that large language models align more closely with human preferences than n-grams . they quantify this variance and compare them to other NLG tasks and benchmarks based on the results . |
| Outcome: | The proposed models align more closely with human preferences than n-gram or embedding-based metrics. |
Phrase Grounding by Soft-Label Chain Conditional Random Field (D19-1)
Copied to clipboard
| Challenge: | Existing methods to ground entities depend on inference or non-differentiable losses. |
| Approach: | They propose a phrase grounding task that grounds entities to corresponding regions in an image . they use neural chain Conditional Random Fields to model dependencies among regions . |
| Outcome: | The proposed method is based on a dataset of the Flickr30k Entities dataset. |
Contextual Relevance and Adaptive Sampling for LLM-Based Document Reranking (2026.acl-long)
Copied to clipboard
| Challenge: | identifying relevant documents for Reasoning-intensive queries remains a challenge . large language models have shown strong performance in zero-shot document reranking . |
| Approach: | They propose a reranking algorithm that estimates contextual relevance by aggregating LLMs' relevance judgments across batches. |
| Outcome: | The proposed algorithm improves nDCG@10 over retrieval and reranking baselines by 15% and 6–21% respectively. |
HySPA: Hybrid Span Generation for Scalable Text-to-Graph Extraction (2021.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to extract information graphs are difficult to scale to datasets with longer input texts because of their secondorder space/time complexities. |
| Approach: | They propose a Hybrid SPan GenerAtor that invertibly maps the information graph to an alternating sequence of nodes and edge types and generates them via a hybrid span decoder. |
| Outcome: | The proposed method outperforms state-of-the-art methods on the ACE05 dataset. |
SIR-ABSC: Incorporating Syntax into RoBERTa-based Sentiment Analysis Models with a Special Aggregator Token (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods to integrate syntactic dependency information into language models capture syntax . aspect-based sentiment classification tasks require a complex model to handle different aspects of a sentence . |
| Approach: | They propose a method to incorporate syntactic dependency information directly into transformer-based language models for Aspect-Based Sentiment Classification. |
| Outcome: | The proposed model outperforms existing models for aspect-based sentiment analysis tasks. |
Evaluating Legal Reasoning Traces with Legal Issue Tree Rubrics (2026.acl-long)
Copied to clipboard
| Challenge: | Evaluating the quality of LLM-generated reasoning traces in expert domains is essential for ensuring credibility and explainability, yet remains challenging due to the inherent complexity of such reasoning tasks. |
| Approach: | They propose a large-scale legal reasoning dataset with an emphasis on reasoning trace evaluation that converts court judgments into hierarchical trees of opposing parties’ arguments and the court’s conclusions. |
| Outcome: | The proposed model improves the quality of LLM-generated reasoning traces in legal domains, whereas RL improves correctness albeit with reduced coverage. |
The Power of Bullet Lists: A Simple Yet Effective Prompting Approach to Enhancing Spatial Reasoning in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are currently dominating the field of natural language processing, but spatial reasoning ability is lacking in LLMs. |
| Approach: | They propose a prompting technique that integrates bullet lists, coordinates, and visualizations into the reasoning process and integrates them into planning tasks. |
| Outcome: | The proposed technique boosts LLMs' spatial reasoning abilities compared to previous prompting techniques. |
Multimedia Generative Script Learning for Task Planning (2023.findings-acl)
Copied to clipboard
| Challenge: | Goal-oriented generative script learning aims to generate subsequent steps to reach a specific goal . ability to capture historical states in visual modalities provides detailed information not covered by text . |
| Approach: | They propose a goal-oriented generative script learning task to generate subsequent steps by tracking historical states in both text and vision modalities. |
| Outcome: | The proposed task outperforms baselines in three aspects of the current task. |
Learning to execute instructions in a Minecraft dialogue (2020.acl-main)
Copied to clipboard
| Challenge: | Existing attempts to build interactive agents that can communicate with humans about and operate within the physical world are either completely ungrounded, focus on slot-value filling tasks, or operate within static environments, such as images or videos. |
| Approach: | They define the subtask of predicting correct action sequences in a given game context and capture B’s past actions as well as B’ s perspective leads to a significant improvement in performance. |
| Outcome: | The proposed task improves the performance of the two-player game and its corresponding Minecraft Dialogue Corpus. |