Papers with end-task
Are Shortest Rationales the Best Explanations for Human Understanding? (2022.acl-short)
Copied to clipboard
| Challenge: | Existing models favor extracting the shortest possible rationales to explain model predictions . however, this assumption has yet to be validated . |
| Approach: | They propose a model that extracts rationales at any target length from text inputs . they show that rationale lengths too short do not help humans predict labels better . |
| Outcome: | The proposed model achieves compatible end-task performance and human-annotated rationale agreement compared to baseline models . |
Calibration of Pre-trained Transformers (2020.emnlp-main)
Copied to clipboard
| Challenge: | Pre-trained Transformers dominate benchmark tasks but use a large number of self-attention heads across many layers in a way that is difficult to unpack. |
| Approach: | They analyze pre-trained Transformer models' posterior probabilities to determine whether they are calibrated for three tasks: natural language inference, paraphrase detection, and commonsense reasoning. |
| Outcome: | The models are calibrated in-domain and out-of-domain, and their calibration error out-domain can be as much as 3.5x lower. |
Document Structure in Long Document Transformers (2024.eacl-long)
Copied to clipboard
| Challenge: | Existing long-document Transformers do not learn representations of document structure during pretraining. |
| Approach: | They propose to use long-document Transformers to acquire an internal representation of document structure during pre-training and evaluate the effects of structure infusion on QASPER and Evidence Inference. |
| Outcome: | The proposed models acquire implicit understanding of document structure during pre-training, which can be enhanced by structure infusion, leading to improved end-task performance. |
End-Task Oriented Textual Entailment via Deep Explorations of Inter-Sentence Interactions (P18-2)
Copied to clipboard
| Challenge: | Existing datasets for textual entailment (TE) have been used to study TE. |
| Approach: | They propose a deep explorations of inter-sentence interactions for textual entailment task that uses a convolution to make important words in P and H play a dominant role in learnt representations. |
| Outcome: | Experiments show that the pretrained DEISTE on SciTail gets 5% improvement over prior state of the art and that it generalizes well on RTE-5. |
Unsupervised Adaptation of Question Answering Systems via Generative Self-training (2020.emnlp-main)
Copied to clipboard
| Challenge: | Supervised self-training methods have transformed applied machine learning . however, adapting to target data has received little attention . |
| Approach: | They propose a method to generate synthetic QA pairs for unsupervised self adaptation . they use massive amounts of data to simulate self-supervised tasks . |
| Outcome: | The proposed method improves QA systems significantly by using less data and training computation than existing augmentation approaches. |
“According to . . . ”: Prompting Language Models Improves Quoting from Pre-Training Data (2024.eacl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) may hallucinate and generate false information despite pre-training on factual data. |
| Approach: | They propose a new evaluation metric that measures the extent to which model-produced answers are directly found in underlying text corpora. |
| Outcome: | The proposed evaluation metric measures the extent to which model-produced answers are directly found in underlying text corpora. |
Tracking State Changes in Procedural Text: a Challenge Dataset and Models for Process Paragraph Comprehension (N18-1)
Copied to clipboard
| Challenge: | Using synthetic data, existing models struggle with questions that require inference. |
| Approach: | They propose a dataset and two new neural models that exploit alternative mechanisms for state prediction. |
| Outcome: | The proposed dataset improves accuracy by 19% over previous models. |
Large Language Models Can be Lazy Learners: Analyze Shortcuts in In-Context Learning (2023.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have shown great potential for in-context learning, but their robustness and performance on downstream tasks remains limited. |
| Approach: | They propose to examine the reliance of LLMs on shortcuts or spurious correlations within prompts for downstream tasks and find larger models are more likely to utilize shortcuts in prompts during inference. |
| Outcome: | The proposed model is “lazy learner” and more likely to use shortcuts in prompts during inference. |
When KV Cache Reuse Fails in Multi-Agent Systems: Cross-Candidate Interaction is Crucial for LLM Judges (2026.acl-long)
Copied to clipboard
| Challenge: | Multi-agent LLMs generate multiple candidate responses that are aggregated by an LLM judge. |
| Approach: | They propose to advocate KV cache reuse across partially shared contexts and report substantial speedups for generation agents. |
| Outcome: | The proposed reuse strategies weaken cross-candidate attention, especially for later candidate blocks, and highlight judge-centric inference as a distinct regime that requires dedicated, risk-aware system design. |
Frustratingly Easy Label Projection for Cross-lingual Transfer (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to improve cross-lingual transfer performance are based on word alignment, but no empirical studies have evaluated their effectiveness or limitations. |
| Approach: | They propose a mark-then-translate method that integrates translation and projection by inserting special markers around the labeled spans in the original sentence. |
| Outcome: | The proposed method outperforms word alignment-based methods in 57 languages and three tasks. |
Effective Attention Sheds Light On Interpretability (2021.findings-acl)
Copied to clipboard
| Challenge: | Using a subset of the GLUE tasks and BERT, we compare the two attention matrices and show that their interpretations differ. |
| Approach: | They propose to use visualizing effective attention to interpret a transformer's behavior since it is more pertinent to the model output by design. |
| Outcome: | The proposed method is more relevant to the model output by design than visualizing attention weights. |
RL4F: Generating Natural Language Feedback with Reinforcement Learning for Repairing Model Outputs (2023.acl-long)
Copied to clipboard
| Challenge: | Despite their success, even the largest language models make mistakes. |
| Approach: | They propose a framework where one language model can generate critiques to improve its peer's performance. |
| Outcome: | The proposed framework improves the performance of a fixed model 200 times its size by 10% over other models. |
Time Waits for No One! Analysis and Challenges of Temporal Misalignment (2022.naacl-main)
Copied to clipboard
| Challenge: | a pretrained model is optionally adapted through domain-specific pretraining, followed by task-specific finetuning. |
| Approach: | They establish a suite of eight tasks across different domains to quantify the effects of temporal misalignment in modern NLP systems. |
| Outcome: | The proposed tasks are based on eight domains and periods of time spanning five years or more and show that they have stronger effects than previous studies. |
FiD-ICL: A Fusion-in-Decoder Approach for Efficient In-Context Learning (2023.acl-long)
Copied to clipboard
| Challenge: | Large pre-trained models are capable of few-shot in-context learning (ICL) however, concatenated demonstrations are often excessively long and require additional computation. |
| Approach: | They propose to apply fusion-in-decoder (FiD) models to perform few-shot in-context learning (ICL) they propose to use concatenation-based, early-fusion, intermediate- and late-fusion methods to improve efficiency . |
| Outcome: | The proposed methods outperform concatenation-based models on 11 held-out tasks. |
Efficient k-NN Search with Cross-Encoders using Adaptive Multi-Round CUR Decomposition (2023.findings-emnlp)
Copied to clipboard
| Challenge: | ANNCUR uses a cross-encoder only to perform k-NN search, but the approximation of the distances is often detrimental to the retrieval of top-k items. |
| Approach: | They propose a method that minimizes approximation error for k-nearest neighbor searches . they propose to use a cross-encoder only to perform k NN search . |
| Outcome: | The proposed method reduces approximation error for top-k neighbors by up to 70% . iteratively performs k-NN search using the available anchors, then adds them to the next set . |
Breakpoint Transformers for Modeling and Tracking Intermediate Beliefs (2022.emnlp-main)
Copied to clipboard
| Challenge: | a framework for language understanding models to track and improve beliefs through intermediate points in text is needed . breakpoint modeling is an efficient and end-to-end learning approach that trains models to train beliefs . understanding the behavior of models remains a formidable challenge for model safety, authors say . |
| Approach: | They propose a framework that trains models to track beliefs through intermediate points in text . their framework allows for efficient and robust learning of this type of model . |
| Outcome: | The proposed model outperforms strong representation learning approaches on a variety of NLU tasks. |
BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension (2020.acl-main)
Copied to clipboard
Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Veselin Stoyanov, Luke Zettlemoyer
| Challenge: | Recent work has shown gains by improving the distribution of masked tokens and the order in which mucked tokens are predicted. |
| Approach: | They propose a denoising autoencoder for pretraining sequence-to-sequence models that corrupts text with an arbitrary noising function and learns a model to reconstruct the original text. |
| Outcome: | The proposed model outperforms RoBERTa on GLUE and SQUAD and provides a 1.1 BLEU increase over a back-translation system for machine translation. |
Verify Before You Commit: Towards Faithful Reasoning in LLM Agents via Self-Auditing (2026.acl-long)
Copied to clipboard
| Challenge: | Existing approaches to reasoning faithfulness violate constraints, authors say . a science fantasy series and companion books are among the books . |
| Approach: | They propose a framework that enforces verification over internal belief states within the agent before action commitment, achieving faithful reasoning. |
| Outcome: | The proposed framework improves reasoning faithfulness while preserving competitive end-task performance. |
LLM-Based Multi-Agent Systems for Clinical Workflows: A Survey of AI Hospitals (2026.acl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are moving from isolated text generation toward agentic work inside clinical workflows. |
| Approach: | They propose a workflow-level taxonomy for LLM-based multi-agent systems for clinical and healthcare workflows . they propose integration readiness levels, task-level instrumentation requirements and recurring workflow failure modes as a practical framework for comparing, evaluating and deploying clinical LLM agents and AI hospitals. |
| Outcome: | The proposed systems should be compared at the workflow level, rather than only by model components or end-task accuracy. |