FailSAE: Towards Interpretable Failure Prediction for Vision-Language Models via Sparse Autoencoders

• URL: http://arxiv.org/abs/2609.04276v1
• Date: Wed, 02 Sep 2026 23:52:47 GMT
• Title: FailSAE: Towards Interpretable Failure Prediction for Vision-Language Models via Sparse Autoencoders
• Abstract summary: We investigate the use of Sparse Autoencoders (SAEs) for interpretable failure prediction in vision-language models (VLMs)<n>We formulate failure prediction as a classification task over sparse SAE latent activations and introduce a three-stage failure-aware training pipeline.<n>Our experiments show that the resulting framework outperforms the evaluated baselines in failure prediction.
• Score: 12.030216781017558
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Vision-language models (VLMs), such as CLIP, have achieved strong performance across multimodal tasks by aligning visual and textual representations in a shared embedding space. As VLMs are increasingly used for high-stakes domains, failure prediction becomes critical for risk-aware deployment and human intervention. Existing failure prediction methods typically rely on confidence scores or auxiliary classifiers. Although these methods are effective on predicting VLM failures, they provide limited interpretability. In this work, we investigate the use of Sparse Autoencoders (SAEs) for interpretable failure prediction in VLMs. We formulate failure prediction as a classification task over sparse SAE latent activations and introduce a three-stage failure-aware training pipeline that encourages the learned latent directions to remain interpretable while becoming more informative for failure prediction. Our experiments show that the resulting framework outperforms the evaluated baselines in failure prediction. Further analysis suggests that failure-aware training encourages SAE latent directions to capture more class-specific concepts. We also use the SAE to provide a concept-level analysis of how model representations change during failures, revealing a shift from class-specific concepts toward more ambiguous or style-related concepts. Finally, we explore how the learned SAE latent directions can support runtime failure recovery.

Related papers

• Uncertainty Quantification for Flow-Based Vision-Language-Action Models [33.28454469934064]
  Vision-language-action models (VLAs) combine vision-language backbones with expressive generative action heads trained via flow matching on large-scale robotic datasets.<n>Despite their strong empirical performance in robotic manipulation, VLAs lack mechanisms to quantify confidence in their predictions and to detect when their actions may be unreliable.<n>We propose SAVE, a framework for uncertainty-guided active fine-tuning that reduces the number of costly expert demonstrations required to adapt VLAs to new tasks.
  arXiv  Detail & Related papers  (2026-06-16T15:19:09Z)
• QuoVLA: Quotient Space for Vision-Language-Action Models [51.02329790939691]
  Vision-Language-Action (VLA) models commonly adapt pretrained VLM latents to robot control by mapping visual observations and language instructions to continuous actions.<n>Our textitQuotient Theory for VLA shows that pretrained VLM latents are not action-insufficient but action-sufficient.<n>We propose QuoVLA, a quotient-space framework for VLA that compresses pretrained VLM latents into action-sufficient representations.
  arXiv  Detail & Related papers  (2026-05-24T06:28:53Z)
• Unleashing VLA Potentials in Autonomous Driving via Explicit Learning from Failures [14.313346858887286]
  Vision-Language-Action (VLA) models for autonomous driving often hit a performance plateau during Reinforcement Learning (RL) optimization.<n>This stagnation arises from exploration capabilities constrained by previous Supervised Fine-Tuning (SFT)<n>We propose Explicit Learning from Failures (ELF-VLA), a framework that augments RL with structured diagnostic feedback.
  arXiv  Detail & Related papers  (2026-03-01T11:41:22Z)
• VAUQ: Vision-Aware Uncertainty Quantification for LVLM Self-Evaluation [22.921677603408188]
  Large Vision-Language Models (LVLMs) frequently hallucinate, limiting their safe deployment in real-world applications.<n>We propose VAUQ, a vision-aware uncertainty quantification framework for LVLM self-evaluation.<n>VAUQ explicitly measures how strongly a model's output depends on visual evidence.
  arXiv  Detail & Related papers  (2026-02-24T16:11:14Z)
• Understanding Degradation with Vision Language Model [56.09241449206817]
  Understanding visual degradations is a critical yet challenging problem in computer vision.<n>We introduce DU-VLM, a multimodal chain-of-thought model trained with supervised fine-tuning and reinforcement learning.<n>We also introduce textbfDU-110k, a large-scale dataset comprising 110,000 clean-degraded pairs with grounded physical annotations.
  arXiv  Detail & Related papers  (2026-02-04T13:51:15Z)
• Reshaping Action Error Distributions for Reliable Vision-Language-Action Models [69.38615670891038]
  In robotic manipulation, vision-language-action (VLA) models have emerged as a promising paradigm for learning generalizable and scalable robot policies.<n>We focus on continuous-action VLA models and move beyond conventional MSE-based regression by reshaping action error distributions during training.<n>We evaluate our approaches across standard, few-shot, and noisy settings on multiple representative VLA architectures.
  arXiv  Detail & Related papers  (2026-02-04T05:37:09Z)
• Learning Affordances at Inference-Time for Vision-Language-Action Models [50.93181349331096]
  In robotics, Vision-Language-Action models (VLAs) offer a promising path towards solving complex control tasks.<n>We introduce Learning from Inference-Time Execution (LITEN), which connects a VLA low-level policy to a high-level VLM that conditions on past experiences.<n>Our approach iterates between a reasoning phase that generates and executes plans for the low-level VLA, and an assessment phase that reflects on the resulting execution.
  arXiv  Detail & Related papers  (2025-10-22T16:43:29Z)
• INSIGHT: INference-time Sequence Introspection for Generating Help Triggers in Vision-Language-Action Models [2.509305596181814]
  Recent Vision-Language-Action (VLA) models show strong generalization capabilities, yet they lack introspective mechanisms for anticipating failures and requesting help from a human supervisor.<n>We present textbfINSIGHT, a learning framework for leveraging token-level uncertainty signals to predict when a VLA should request help.
  arXiv  Detail & Related papers  (2025-10-01T19:22:48Z)
• Post-hoc Probabilistic Vision-Language Models [54.05237186168399]
  Vision-language models (VLMs) have found remarkable success in classification, retrieval, and generative tasks.<n>We propose post-hoc uncertainty estimation in VLMs that does not require additional training.<n>Our results show promise for safety-critical applications of large-scale models.
  arXiv  Detail & Related papers  (2024-12-08T18:16:13Z)
• Interpretable Self-Aware Neural Networks for Robust Trajectory Prediction [50.79827516897913]
  We introduce an interpretable paradigm for trajectory prediction that distributes the uncertainty among semantic concepts. We validate our approach on real-world autonomous driving data, demonstrating superior performance over state-of-the-art baselines.
  arXiv  Detail & Related papers  (2022-11-16T06:28:20Z)

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.