Discrete flow matching with arbitrary probability paths is made tractable via kinetic-optimal velocities, mixture schedulers, and a new ELBO, improving text, crystal, and image generation.
If we do not change state, then we sample from δX i t (·)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
method 1
citation-polarity summary
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1roles
method 1polarities
use method 1representative citing papers
citing papers explorer
-
Flow Matching with General Discrete Paths: A Kinetic-Optimal Perspective
Discrete flow matching with arbitrary probability paths is made tractable via kinetic-optimal velocities, mixture schedulers, and a new ELBO, improving text, crystal, and image generation.