Facebook’s AI research team has recently revealed that Pythia, its modular plug-and-play framework that allows to quickly build, reproduce, and benchmark AI models, is now available on GitHub, free of charge.
Pythia can be described as a deep learning framework that supports multitasking in the vision and language domain. The system, built on their open-source PyTorch framework, is meant for vision and language tasks, such as answering questions related to visual data and automatically generating image captions.
The framework supports distributed training and a variety of data sets, plus custom losses, metrics, scheduling, and optimizers. “Pythia smooths the process of entering the growing subfield of vision and language and frees researchers to focus on faster prototyping and experimentation. Our goal is to accelerate progress by increasing the reproducibility of these models and results,” states the team. “This will make it easier for the community to build on, and benchmark against, successful systems.”
The move should help researchers develop adaptive AI that synthesizes multiple kinds of understanding into a more context-based, multimodal understanding. The team also noted that it plans to continue adding tools, tasks, data sets, and reference models.