Model Directions, Not Words: Mechanistic Topic Models Using Sparse Autoencoders

• URL: http://arxiv.org/abs/2507.23220v1
• Date: Thu, 31 Jul 2025 03:17:43 GMT
• Title: Model Directions, Not Words: Mechanistic Topic Models Using Sparse Autoencoders
• Authors: Carolina Zheng, Nicolas Beltran-Velez, Sweta Karlekar, Claudia Shi, Achille Nazaret, Asif Mallik, Amir Feder, David M. Blei,
• Abstract summary: Mechanistic Topic Models (MTMs) operate on interpretable features learned by sparse autoencoders (SAEs)<n>MTMs enable controllable text generation using topic-based steering vectors.<n>To properly evaluate MTM topics against word-list-based approaches, we propose textittopic judge, an LLM-based pairwise comparison evaluation framework.
• Score: 21.650904669171524
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Traditional topic models are effective at uncovering latent themes in large text collections. However, due to their reliance on bag-of-words representations, they struggle to capture semantically abstract features. While some neural variants use richer representations, they are similarly constrained by expressing topics as word lists, which limits their ability to articulate complex topics. We introduce Mechanistic Topic Models (MTMs), a class of topic models that operate on interpretable features learned by sparse autoencoders (SAEs). By defining topics over this semantically rich space, MTMs can reveal deeper conceptual themes with expressive feature descriptions. Moreover, uniquely among topic models, MTMs enable controllable text generation using topic-based steering vectors. To properly evaluate MTM topics against word-list-based approaches, we propose \textit{topic judge}, an LLM-based pairwise comparison evaluation framework. Across five datasets, MTMs match or exceed traditional and neural baselines on coherence metrics, are consistently preferred by topic judge, and enable effective steering of LLM outputs.

Related papers

• Neural Topic Modeling with Large Language Models in the Loop [12.142323482188056]
  We propose a novel framework that integrates Large Language Models (LLMs) with Neural Topic Models (NTMs)<n>With the flexibility of being integrated into many existing NTMs, the proposed approach enhances the interpretability of topics while preserving the efficiency of NTMs in learning topics and document representations.
  arXiv  Detail & Related papers  (2024-11-13T11:31:02Z)
• Align-SLM: Textless Spoken Language Models with Reinforcement Learning from AI Feedback [50.84142264245052]
  This work introduces the Align-SLM framework to enhance the semantic understanding of textless Spoken Language Models (SLMs)<n>Our approach generates multiple speech continuations from a given prompt and uses semantic metrics to create preference data for Direct Preference Optimization (DPO)<n>We evaluate the framework using ZeroSpeech 2021 benchmarks for lexical and syntactic modeling, the spoken version of the StoryCloze dataset for semantic coherence, and other speech generation metrics, including the GPT4-o score and human evaluation.
  arXiv  Detail & Related papers  (2024-11-04T06:07:53Z)
• Enhancing Short-Text Topic Modeling with LLM-Driven Context Expansion and Prefix-Tuned VAEs [25.915607750636333]
  We propose a novel approach that leverages large language models (LLMs) to extend short texts into more detailed sequences before applying topic modeling. Our method significantly improves short-text topic modeling performance, as demonstrated by extensive experiments on real-world datasets with extreme data sparsity.
  arXiv  Detail & Related papers  (2024-10-04T01:28:56Z)
• Interactive Topic Models with Optimal Transport [75.26555710661908]
  We present EdTM, as an approach for label name supervised topic modeling. EdTM models topic modeling as an assignment problem while leveraging LM/LLM based document-topic affinities.
  arXiv  Detail & Related papers  (2024-06-28T13:57:27Z)
• A Large Language Model Guided Topic Refinement Mechanism for Short Text Modeling [10.589126787499973]
  Existing topic models often struggle to accurately capture the underlying semantic patterns of short texts.<n>This paper introduces a novel model-agnostic mechanism, termed Topic Refinement.<n>We show that Topic Refinement boosts the topic quality and improves the performance in topic-related text classification tasks.
  arXiv  Detail & Related papers  (2024-03-26T13:50:34Z)
• Large Language Models Offer an Alternative to the Traditional Approach of Topic Modelling [0.9095496510579351]
  We investigate the untapped potential of large language models (LLMs) as an alternative for uncovering the underlying topics within extensive text corpora. Our findings indicate that LLMs with appropriate prompts can stand out as a viable alternative, capable of generating relevant topic titles and adhering to human guidelines to refine and merge topics.
  arXiv  Detail & Related papers  (2024-03-24T17:39:51Z)
• A General and Flexible Multi-concept Parsing Framework for Multilingual Semantic Matching [60.51839859852572]
  We propose to resolve the text into multi concepts for multilingual semantic matching to liberate the model from the reliance on NER models. We conduct comprehensive experiments on English datasets QQP and MRPC, and Chinese dataset Medical-SM.
  arXiv  Detail & Related papers  (2024-03-05T13:55:16Z)
• TopicGPT: A Prompt-based Topic Modeling Framework [77.72072691307811]
  We introduce TopicGPT, a prompt-based framework that uses large language models to uncover latent topics in a text collection. It produces topics that align better with human categorizations compared to competing methods. Its topics are also interpretable, dispensing with ambiguous bags of words in favor of topics with natural language labels and associated free-form descriptions.
  arXiv  Detail & Related papers  (2023-11-02T17:57:10Z)
• FLIP: Fine-grained Alignment between ID-based Models and Pretrained Language Models for CTR Prediction [49.510163437116645]
  Click-through rate (CTR) prediction plays as a core function module in personalized online services. Traditional ID-based models for CTR prediction take as inputs the one-hot encoded ID features of tabular modality. Pretrained Language Models(PLMs) has given rise to another paradigm, which takes as inputs the sentences of textual modality. We propose to conduct Fine-grained feature-level ALignment between ID-based Models and Pretrained Language Models(FLIP) for CTR prediction.
  arXiv  Detail & Related papers  (2023-10-30T11:25:03Z)
• Dialogue Meaning Representation for Task-Oriented Dialogue Systems [51.91615150842267]
  We propose Dialogue Meaning Representation (DMR), a flexible and easily extendable representation for task-oriented dialogue. Our representation contains a set of nodes and edges with inheritance hierarchy to represent rich semantics for compositional semantics and task-specific concepts. We propose two evaluation tasks to evaluate different machine learning based dialogue models, and further propose a novel coreference resolution model GNNCoref for the graph-based coreference resolution task.
  arXiv  Detail & Related papers  (2022-04-23T04:17:55Z)
• Topic Discovery via Latent Space Clustering of Pretrained Language Model Representations [35.74225306947918]
  We propose a joint latent space learning and clustering framework built upon PLM embeddings. Our model effectively leverages the strong representation power and superb linguistic features brought by PLMs for topic discovery.
  arXiv  Detail & Related papers  (2022-02-09T17:26:08Z)
• Keyword Assisted Embedded Topic Model [1.9000421840914223]
  Probabilistic topic models describe how words in documents are generated via a set of latent distributions called topics. Recently, the Embedded Topic Model (ETM) has extended LDA to utilize the semantic information in word embeddings to derive semantically richer topics. We propose the Keyword Assisted Embedded Topic Model (KeyETM), which equips ETM with the ability to incorporate user knowledge in the form of informative topic-level priors.
  arXiv  Detail & Related papers  (2021-11-22T07:27:17Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.