Uploaded May 2025 | Updated September 2026, 1 day ago
Most well-known generative models take in data and produce fabricated samples that mimic the patterns they find in the data. In specialized applications, there may be extremely limited or zero training data available, so training these models can be impractical. Instead, only a scalar reward signal is provided that verifies whether the model is producing good samples. Example applications include fine-tuning image and video generative models or learning to sample from physical or chemistry foundation models.
Adjoint Sampling presents a way forward using a highly scalable, reward-based objective to train generative models without any data. Rather than finding patterns in existing data, Adjoint Sampling finds patterns by iteratively refining its own samples according to a provided reward model. Based on theoretical foundations developed at FAIR, Adjoint Sampling leads to a highly scalable practical algorithm and can become the foundation for further research into highly scalable reward-driven generative modeling.
Read more: ai.meta.com/blog/meta-fair-science-new-open-source-releases
Subscribe: youtube.com/aiatmeta?sub_confirmation=1
Learn more about our work: ai.meta.com
Follow us on Twitter: twitter.com/aiatmeta
Follow us on Facebook: facebook.com/aiatmeta
Connect with us on LinkedIn: linkedin.com/showcase/aiatmeta
Meta focuses on bringing the world together by advancing AI, powering meaningful and safe experiences, and conducting open research.
Most well-known generative models take in data and produce fabricated samples that mimic the patterns they find in the data. In specialized applications, there may be extremely limited or zero training data available, so training these models can be impractical. Instead, only a scalar reward signal is provided that verifies whether the model is producing good samples. Example applications include fine-tuning image and video generative models or learning to sample from physical or chemistry foundation models.
Adjoint Sampling presents a way forward using a highly scalable, reward-based objective to train generative models without any data. Rather than finding patterns in existing data, Adjoint Sampling finds patterns by iteratively refining its own samples according to a provided reward model. Based on theoretical foundations developed at FAIR, Adjoint Sampling leads to a highly scalable practical algorithm and can become the foundation for further research into highly scalable reward-driven generative modeling.
Read more: ai.meta.com/blog/meta-fair-science-new-open-source-releases
Subscribe: youtube.com/aiatmeta?sub_confirmation=1
Learn more about our work: ai.meta.com
Follow us on Twitter: twitter.com/aiatmeta
Follow us on Facebook: facebook.com/aiatmeta
Connect with us on LinkedIn: linkedin.com/showcase/aiatmeta
Meta focuses on bringing the world together by advancing AI, powering meaningful and safe experiences, and conducting open research.
