| Challenge: | Recent work shows that models trained on held-out data perform poorly on hard instances . previous methods have resorted to manual methods of encouraging models not to overfit to superficial cues . |
| Approach: | They propose to explicitly learn a model that does well on both easy and hard tests . they use Choice of Plausible Alternatives and Commonsense Explanation to evaluate the model . |
| Outcome: | The proposed model performs well on easy and hard tests with superficial cues but performs poorly on hard ones without superficial cuings. |
Similar Papers
When Choosing Plausible Alternatives, Clever Hans can be Clever (D19-60)
Copied to clipboard
| Challenge: | Pretrained language models have shown large improvements in the commonsense reasoning benchmark COPA, but recent work has identified superficial cues in benchmark datasets which are predictive of the correct answer. |
| Approach: | They propose an extension of COPA that does not suffer from easy-to-exploit single token cues and exploits them. |
| Outcome: | The proposed extension of COPA does not suffer from easy-to-exploit single token cues. |
Exploring Strategies for Generalizable Commonsense Reasoning with Pre-trained Models (2021.emnlp-main)
Copied to clipboard
| Challenge: | Recent work proposes lightweight updates to improve commonsense reasoning models . fine-tuning can cause models to overfit to task-specific data and forget knowledge gained during training . |
| Approach: | They propose to use lightweight models to update pre-trained language models to learn commonsense background knowledge. |
| Outcome: | The proposed models learn from commonsense reasoning datasets, but they are overfitted and limited generalized. |
Weak-to-Strong Reasoning (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing approaches to supervise large language models (LLMs) exceed human capabilities, but the effectiveness of this approach is still unexplored. |
| Approach: | They propose a weak-to-strong reasoning framework that enables strong models to refine training data . they use supervised fine-tuning and preference optimization to optimize weak models . |
| Outcome: | The proposed framework improves the reasoning capabilities of a language model using three weak models. |
The Unreasonable Effectiveness of Easy Training Data for Hard Tasks (2024.acl-long)
Copied to clipboard
| Challenge: | Existing pretrained language models perform well on hard data, but hard data is noisier and costlier to collect. |
| Approach: | They propose to use in-context learning, linear classifier heads, and QLoRA to show that pretrained language models generalize relatively well from easy to hard data. |
| Outcome: | The proposed model generalizes well from easy to hard data even better than oracle models finetuned on hard data. |
Teaching Small Language Models to Learn Logic through Meta-Learning (2026.eacl-long)
Copied to clipboard
| Challenge: | Large language models are increasingly evaluated on reasoning tasks, yet their logical abilities remain contested. |
| Approach: | They propose to apply few-shot meta-learning to large language models' reasoning domain to enable them to acquire abstract inference patterns that generalize to novel structures. |
| Outcome: | The proposed model outperforms GPT-4o and o3-mini on a syllogistic reasoning task. |
Keep Learning: Self-supervised Meta-learning for Learning from Inference (2021.eacl-main)
Copied to clipboard
| Challenge: | A common approach to improve performance of machine learning algorithms involves self-supervised learning on large unlabeled data before fine-tuning on downstream tasks. |
| Approach: | They propose to use model's own class-balanced predictions to back-propagate the loss from the model''s class-balancing predictions (pseudo-labels) this method improves performance of standard backbones such as BERT, Electra, and ResNet-50 on a wide variety of tasks, including question answering on SQuAD and NewsQA . |
| Outcome: | The proposed method outperforms previous approaches on a wide variety of tasks including question answering on SQuAD and NewsQA, benchmark task SuperGLUE, conversation response selection on Ubuntu Dialog corpus v2.0, and image classification on MNIST and ImageNet. |
MERIt: Meta-Path Guided Contrastive Learning for Logical Reasoning (2022.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to infer logical relations with annotated training data suffer from over-fitting and poor generalization problems due to the dataset sparsity. |
| Approach: | They propose a MEta-path guided contrastive learning method for logical ReasonIng of text that performs self-supervised pre-training on abundant unlabeled text data. |
| Outcome: | The proposed method outperforms the baselines on two logical reasoning benchmarks with significant improvements. |
Know Where You’re Going: Meta-Learning for Parameter-Efficient Fine-Tuning (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on parameter-efficient fine-tuning methods require additional measures after pre-training and before fine-uning. |
| Approach: | They propose to take parameter-efficient fine-tuning into consideration after pre-training and before fine-uning and use meta-learning to prime a model specifically for parameter-efficiency. |
| Outcome: | The proposed method improves on a pre-trained model with certain modifications and achieves 4.96 points on cross-lingual NER fine-tuning. |
MetaICL: Learning to Learn In Context (2022.naacl-main)
Copied to clipboard
| Challenge: | Large language models can do in-context learning by conditioning on a few training examples with no parameter updates or task-specific templates. |
| Approach: | They propose a meta-training framework where a pretrained language model is tuned to do in-context learning on a large set of training tasks. |
| Outcome: | The proposed framework outperforms baseline models on 142 NLP datasets and a range of target tasks with domain shifts. |
Extracting and Understanding the Superficial Knowledge in Alignment (2025.naacl-long)
Copied to clipboard
Runjin Chen, Gabriel Jacob Perin, Xuxi Chen, Xilun Chen, Yan Han, Nina S. T. Hirata, Junyuan Hong, Bhavya Kailkhura
| Challenge: | Recent studies have shown that alignment of large language models with human values and preferences requires substantial data and computation resources. |
| Approach: | They propose a method to extract and isolate superficial knowledge from aligned models by focusing on the shallow modifications to the final token selection process. |
| Outcome: | The proposed method extracts and isolates superficial knowledge from aligned models, focusing on the shallow modifications to the final token selection process. |