Research map / Vision-language and multimodal / Human and embodied modeling
Compositional Explanations of Neurons
Jesse Mu, Jacob Andreas
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| increase max explanation formula length from 1 (NetDissect) to 10 (ablation) | neuron explanation search · ResNet-18 Places365 | mean IoU: 0.059 -> 0.099 | Improved |
| increase max formula length beyond 10 (ablation) | neuron explanation search · ResNet-18 Places365 | mean IoU | No clear effect |
| increase max formula length (up to 30+) (ablation) | neuron explanation search · SNLI BiLSTM | IoU | Improved |
| neurons with higher explanation IoU | ResNet-18 · Places365 | model accuracy when neuron active: r = 0.31 | Improved |
| neurons with higher explanation IoU | BiLSTM NLI · SNLI | model accuracy when neuron active: r = -0.60 | Hurt |
Closest papers by meaning
- Automated Natural Language Explanation of Deep Visual Neurons with Large Models (Student Abstract) (AAAI 2024 · similarity 0.89)
- CoSy: Evaluating Textual Explanations of Neurons (NeurIPS 2024 · similarity 0.88)
- NeuronEye: Query-Guided Visual Concept Activation for Vision-Language Reasoning (NeurIPS 2026 · similarity 0.87)
- Evaluating Neuron Explanations: A Unified Framework with Sanity Checks (ICML 2025 · similarity 0.87)
- LINE: LLM-based Iterative Neuron Explanations for Vision Models (NeurIPS 2026 · similarity 0.87)
- NePTune: A Neuro-Pythonic Framework for Tunable Compositional Reasoning on Vision-Language (ICLR 2026 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
