AF-based methods encode rich structural priors but lack a general mechanism for integrating arbitrary data modalities. ROCKET tackles this by optimizing latent representations to fit experimental data at inference time, without retraining! 2/14
Post image

Comments