Meta AI is pleased to invite university faculty to respond to this call for research proposals for Mephisto, a dataset collection tool.
Most novel machine learning research projects find themselves limited in having the right data for the project. Data collection can be tricky to get right and even with a solid task, it’s often difficult to ensure that data quality is up-to-par. Crowdsourcing best practices are often ‘tribal knowledge,’ leading researchers to rediscover ways to collect quality data across labs and domains.
Mephisto is a tool that Meta AI Research uses to standardize and codify best practices and infrastructure for research data collection and annotation. It also serves as a clear method to open source and share our methodology, such that others can reproduce or expand upon our work. Now that we’ve hit our 1.0 release, we feel like the platform is mature and stable enough for general use. We are opening Mephisto up to the community to ensure we’re building infrastructure that is useful for your use cases.
To foster further innovation in this area, aid in integration work for early adopters, and deepen our collaboration with academia, Meta AI is pleased to invite university faculty to submit research proposals pertaining to Mephisto. A total of 4 awards are available, in the range of $25,000 - $37,500 each, including overhead in an amount up to 40% of project costs. Funding will be provided to RFP winners pursuant to a Sponsored Research Agreement (SRA) containing open science terms. Please note that the terms of the SRA will not be subject to negotiation. We strongly encourage researchers from diverse backgrounds and of diverse abilities to apply.
The deadline for submissions is March 23, 2022.
Meta AI encourages creative submissions that leverage the Mephisto tool, with areas of interest that include, but are not limited to, the following data collection tasks:
Entirely novel and complex tasks that may require rounds of iteration to design and experiment with.
Tasks that involve live interaction between multiple workers in some form, either collaborative or competitive.
Tasks that benefit from having a model-in-the-loop to assist workers.
Tasks that require intricate worker filtering specifications, for pairing, ensuring specified groups, or otherwise.
Tasks that can be built on existing frameworks, but may benefit from incorporating support for that framework into Mephisto.
Proposals must comply with applicable US and international laws, regulations and policies;
Applicants must be current full-time faculty at an accredited academic institution that awards research degrees to PhD students;
Applicants must be the Principal Investigator on any resulting award;
Meta cannot consider proposals submitted, prepared or to be carried out by individuals residing in, or affiliated with an academic institution located in, a country or territory subject to comprehensive U.S. trade sanctions; and
Government officials (excluding faculty and staff of public universities, to the extent they may be considered government officials), political figures, and politically affiliated businesses (all as determined by Meta in its sole discretion) are not eligible.
