Captum
Click to visit website
About
Captum is a PyTorch library for model interpretability. It provides tools for understanding and attributing the predictions of PyTorch models across various modalities like vision and text. It supports a wide range of PyTorch models and is designed to be extensible for interpretability research. Key features include multi-modal support, PyTorch integration, and an open-source design.
Platform
Features
• built on pytorch
• multi-modal support
• extensible
FAQs
How do I set the target parameter to an attribution method?
The purpose of target is to select a single (scalar) value for each example in the output of your model to compute attributions based on the given target parameter.
I am facing Out-Of-Memory (OOM) errors when using Captum. How do I resolve this?
To address this issue, you can either reduce n_steps, which may lead to lower-quality approximations, or use the internal_batch_size argument, which allows dividing the expanded input into batches which are processed sequentially.
I am using a perturbation based method, and attributions are taking too long to compute. How can I speed it up?
To enable this, simply set the perturbations_per_eval argument to the desired value.
Are SmoothGrad or VarGrad supported in Captum?
Yes! SmoothGrad and VarGrad are available through NoiseTunnel in Captum, which can be used with any attribution algorithm in Captum.
How do I use Captum with BERT models?
We have a tutorial demonstrating usage of Integrated Gradients on BERT here.
My model inputs or outputs token indices, and when using Captum I see errors relating to gradients, how do I resolve this?
To apply gradient-based attribution methods, it is necessary to replace the embedding layer with an InterpretableEmbedding layer or use LayerIntegratedGradients to compute attribution with respect to the embedding output.
Can my model use functional non-linearities (E.g. nn.functional.ReLU) or can reused modules be used with Captum?
Most methods will work fine with functional non-linearities and arbitrary operations.
Do JIT models, DataParallel models, or DistributedDataParallel models work with Captum?
Yes, we have support for all these model types. Note that JIT models do not yet support hooks, so any methods using hooks including layer and neuron attribution methods, DeepLift, Guided Backprop, and Deconvolution are not supported.
I am working on a new interpretability or attribution method and would like to add it to Captum. How do I proceed?
For interpretability methods created by the community, we have two methods of involvement: Awesome List and Inclusion in Captum.
How can I resolve cudnn RNN backward error for RNN or LSTM network?
To resolve the issue you can set`torch.backends.cudnn.enabled` flag to False - `torch.backends.cudnn.enabled=False`
Pricing Plans
Free
Free PlanJob Opportunities
There are currently no job postings for this AI tool.
Ratings & Reviews
No ratings available yet. Be the first to rate this tool!
Alternatives
Interpretable AI
Interpretable AI is a software suite providing various modules for interpretable machine learning solutions, including Optimal Trees and OptImpute, with Python and R support.
View DetailsFeatured Tools
Songmeaning
Songmeaning is an AI-powered tool that helps users uncover the hidden stories and meanings behind song lyrics, enhancing their musical understanding.
View Details
PropLytics
PropLytics is an AI-powered platform for real estate investors, providing data-backed ROI insights to help make smarter, faster investment decisions.
View Details
GitGab
GitGab is an AI tool that contextualizes top AI models like ChatGPT, Claude, and Gemini with your GitHub repositories and local code for enhanced development.
View Details
nuptials.ai
nuptials.ai is an AI wedding planning partner, offering timeline planning, budget optimization, vendor matching, and a 24/7 planning assistant to help plan your perfect day.
View Details
Fastbreak AI
Fastbreak AI is an ultimate AI-powered sports operations engine, offering intelligent software for sports league scheduling, tournament management, and brand sponsorship.
View DetailsMolku
Molku is an AI-powered tool that automates data extraction and document filling, allowing users to effortlessly transfer data from various source files into templates.
View Details
BestFaceSwap
BestFaceSwap is an AI-powered online tool that enables users to easily change faces in videos and photos with high-quality and realistic results.
View DetailsEntrevista.app
Entrevista.app is an AI assistant that conducts interviews 24/7, providing personalized feedback to help companies find the best candidates and simplify their hiring process.
View Details
Humanize AI Text
Humanize AI Text is the best AI humanizer tool that transforms AI-generated content into human-like writing, bypassing major AI detectors with ease.
View Details
RightHair
RightHair is a free AI hairstyle changer that allows users to virtually try over 200 hairstyles and colors by uploading their photo, instantly transforming their look.
View Details
Smart Cookie Trivia
Smart Cookie Trivia is a platform offering a wide variety of trivia questions across numerous categories to help users play trivia, explore different topics, and expand their knowledge.
View Details
The Spiritual Seek - Spirituality and Personality Test
The Spiritual Seek is an AI-enhanced platform offering diverse psychological and spiritual personality tests, articles, and meditations for self-discovery and personal growth.
View DetailsLatest AI News
View All News
Qwen2.5's scores reveal how data contamination inflates AI benchmarks, making memorization appear as genuine intelligence.
ARC-AGI-3: A new benchmark reveals AI's struggle with human-easy, adaptive challenges, showing scaling alone won't lead to AGI.
AI2's FlexOlmo revolutionizes LLM training, enabling powerful shared models without ever exposing sensitive, proprietary data.