# CARMA: Enhanced Compositionality in LLMs via Advanced Regularisation and Mutual Information Alignment

arXiv: https://arxiv.org/abs/2502.11066 · HTML: https://mlsanity.com/paper/2502.11066/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 27 of its 52 references that are in the map.
- First posted on arXiv 16 Feb 2025.

## Builds on (most cited first, 27 of 27)

- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Representation Learning with Contrastive Predictive Coding](https://mlsanity.com/paper/1807.03748/) (2018)
- [Gemma 2: Improving Open Language Models at a Practical Size](https://mlsanity.com/paper/2408.00118/) (2024)
- [Faith and Fate: Limits of Transformers on Compositionality](https://mlsanity.com/paper/2305.18654/) (2023)
- [Regularization With Stochastic Transformations and Perturbations for Deep Semi-Supervised Learning](https://mlsanity.com/paper/1606.04586/) (2016)
- [COGS: A Compositional Generalization Challenge Based on Semantic Interpretation](https://mlsanity.com/paper/2010.05465/) (2020)
- [Constituency Parsing with a Self-Attentive Encoder](https://mlsanity.com/paper/1805.01052/) (2018)
- [Stabilizing Transformer Training by Preventing Attention Entropy Collapse](https://mlsanity.com/paper/2303.06296/) (2023)
- [Compositional Generalization and Natural Language Variation: Can a Semantic Parsing Approach Handle Both?](https://mlsanity.com/paper/2010.12725/) (2020)
- [Multilingual Constituency Parsing with Self-Attention and Pre-Training](https://mlsanity.com/paper/1812.11760/) (2018)
- [The Devil is in the Detail: Simple Tricks Improve Systematic Generalization of Transformers](https://mlsanity.com/paper/2108.12284/) (2021)
- [The paradox of the compositionality of natural language: a neural machine translation case study](https://mlsanity.com/paper/2108.05885/) (2021)
- [On Compositional Generalization of Neural Machine Translation](https://mlsanity.com/paper/2105.14802/) (2021)
- [Decoding Probing: Revealing Internal Linguistic Structures in Neural Language Models using Minimal Pairs](https://mlsanity.com/paper/2403.17299/) (2024)
- [Compositional Capabilities of Autoregressive Transformers: A Study on Synthetic, Interpretable Tasks](https://mlsanity.com/paper/2311.12997/) (2023)
- [Pushdown Layers: Encoding Recursive Structure in Transformer Language Models](https://mlsanity.com/paper/2310.19089/) (2023)
- [Token Erasure as a Footprint of Implicit Vocabulary Items in LLMs](https://mlsanity.com/paper/2406.20086/) (2024)
- [A Survey on Compositional Learning of AI Models: Theoretical and Experimental Practices](https://mlsanity.com/paper/2406.08787/) (2024)
- [Transformer Working Memory Enables Regular Language Reasoning and Natural Language Length Extrapolation](https://mlsanity.com/paper/2305.03796/) (2023)
- [When Can Transformers Ground and Compose: Insights from Compositional Generalization Benchmarks](https://mlsanity.com/paper/2210.12786/) (2022)
- [Learning to Substitute Spans towards Improving Compositional Generalization](https://mlsanity.com/paper/2306.02840/) (2023)
- [Can LLM find the green circle? Investigation and Human-guided tool manipulation for compositional generalization](https://mlsanity.com/paper/2312.07763/) (2023)
- [Data Factors for Better Compositional Generalization](https://mlsanity.com/paper/2311.04420/) (2023)
- [Benchmarking and Improving Compositional Generalization of Multi-aspect Controllable Text Generation](https://mlsanity.com/paper/2404.04232/) (2024)
- [Interpreting token compositionality in LLMs: A robustness analysis](https://mlsanity.com/paper/2410.12924/) (2024)
- [Learning to Compose Representations of Different Encoder Layers towards Improving Compositional Generalization](https://mlsanity.com/paper/2305.12169/) (2023)

## Built on by (most cited first, 3 of 3)

- [Revisiting Compositional Generalization Capability of Large Language Models Considering Instruction Following Ability](https://mlsanity.com/paper/2506.15629/) (2025)
- [TRACE for Tracking the Emergence of Semantic Representations in Transformers](https://mlsanity.com/paper/2505.17998/) (2025)
- [Emergence and Localisation of Semantic Role Circuits in LLMs](https://mlsanity.com/paper/2511.20910/) (2025)

## Abstract (arXiv)

Large language models (LLMs) struggle with compositional generalisation, limiting their ability to systematically combine learned components to interpret novel inputs. While architectural modifications, fine-tuning, and data augmentation improve compositionality, they often have limited adaptability, face scalability constraints, or yield diminishing returns on real data. To address this, we propose CARMA, an intervention that enhances the stability and robustness of compositional reasoning in LLMs while preserving fine-tuned performance. CARMA employs mutual information regularisation and layer-wise stability constraints to mitigate feature fragmentation, ensuring structured representations persist across and within layers. We evaluate CARMA on inverse dictionary modelling and sentiment classification, measuring its impact on semantic consistency, performance stability, and robustness to lexical perturbations. Results show that CARMA reduces the variability introduced by fine-tuning, stabilises token representations, and improves compositional reasoning. While its effectiveness varies across architectures, CARMA's key strength lies in reinforcing learned structures rather than introducing new capabilities, making it a scalable auxiliary method. These findings suggest that integrating CARMA with fine-tuning can improve compositional generalisation while maintaining task-specific performance in LLMs.