Request for Proposal ”Creating Benchmark Datasets and Common Task Frameworks for Alternative Protein Development”

While many fields have benefited from having public datasets, which allow different players to collaborate and compare technological progress, the alternative protein industry lacks such resources. This gap, as well as the lack of standardized metrics and benchmarks, hampers the ability of groups and individuals to contribute to collaborative work on the biggest problems of our domain. Establishing common datasets and frameworks, especially those tailored for AI and machine learning applications, will accelerate development across the field, not only enabling researchers and companies to build upon otherwise unavailable data, but also providing clear targets, encouraging comparison of outcomes, and fostering collaboration.

Examples from other domains:

- sustainlab-group.github.io/sustainbench/

- hazyresearch.stanford.edu/legalbench/

Proposals are expected to result in datasets that can be actively used by the industry to enhance product development and market strategies. We encourage proposals to also include practical tools for computing standardized metrics on these datasets.