We aim to help build your research

Our platform is designed to serve researchers, developers, and enthusiasts seeking in-depth knowledge about AI technologies.

Objective

FAiNDER's Objective

FAiNDER started as an internal BSC repository to handle the overwhelming volume of AI publications. Over time it became an essential tool for research and collaboration, and now it's open to the entire community. Our platform offers a comprehensive, filterable table and interactive charts that make it easier to navigate, compare, and understand the AI landscape.

Methodology

Our Process

We believe it's important to detail our methodology to ensure the validity and reliability of the data we present.

Sources

We gather information from major conferences (IPDPS, MICRO, ISCA, HPCA, PACT) and leading companies to cover the latest AI advancements.

Data Processing

We synthesize data into concise summaries, focusing on hardware requirements and AI model performance, providing relevant and actionable insights.

Interactive Data Presentation

We validate our data by cross-referencing multiple sources to ensure accuracy and reliability.

Team

The FAiNDER Team

Our research team specializes in computer architecture and is dedicated to advance the future of AI technology.

Javier Beiro

RNNs, DLRS & Transformers

Mariana Carmin

DNNs, CNNs & GNNs

Victor Xirau

Developer & Support

Petar Radojkovic

Memory Systems Group Leader

Part of the BSC Memory Team

Experts in hardware architecture and high-performance computing dedicated to advancing next-generation memory systems. Through our research, we aim to enhance memory system design to meet the rigorous demands of future HPC applications.

Learn more

Frequently Asked Questions

We've compiled a list of commonly asked questions to provide you with quick and informative answers.

Who is FAiNDER designed for?

FAiNDER serves everyone from students doing their first AI literature review to advanced researchers comparing hardware platforms across dozens of models. Junior users get a clear starting point and curated references; advanced users get detailed breakdowns of software pipelines, data types, memory footprint, and hardware targets.

How can I filter and search for models?

Use the filters in the AI Explorer page to narrow down models by hardware type, data type, phase (training or inference), sparsity, and more. You can combine multiple filters to quickly identify the exact models you need.

How can I check which models are trained on CPUs or GPUs?

In the AI Explorer, use the hardware filter and select CPU or GPU. You can also combine this with the phase filter to distinguish between training and inference hardware.

How is the data sourced and validated?

FAiNDER synthesizes data from top-tier research papers across major conferences (IPDPS, MICRO, ISCA, HPCA, PACT) and leading companies. All entries are cross-referenced against multiple sources to ensure accuracy and include direct links to the original papers.

How often is the database updated?

The database is continuously updated as new publications are released and reviewed by the team. Subscribe to our newsletter to be notified when new models or datasets are added.

Can I cite FAiNDER in my research?

Yes. Use the 'Generate Reference' button on this page to get a ready-to-use citation in your preferred format.