analiza aiqbench
This paper extends discrete flow matching, a training method for generative models over discrete structures, so it can handle multi-component actions such as attaching a molecular fragment or adding a tree node with its edge. The authors claim a principled, simulation-free objective that generalizes factorized baselines, with reported gains such as MOSES validity rising from 92.8% to 99.2%. These results come from specific tree and molecular benchmarks, so readers should check how well they transfer to other domains.
Trajectory Matching: Discrete Flow Matching with Structured Actions
arXiv:2610.32602v1 Announce Type: new Abstract: The natural units of structured discrete generation are often not independent single-component edits but richer actions: attaching a fragment to a molecule, inserting a node together with an incident edge, or growing a hierarchy by adding a child. Yet most existing discrete flow matching methods (DFM) are restricted to independent component-wise edits over fixed-size states. This restriction stems from endpoint conditioning: endpoints fix the source and target during training but not the action sequence between them, making tractable conditional rates hard to define for complex actions that cannot be factorized component-wise. We propose trajectory matching, which conditions the flow on a full sequence of actions connecting the endpoints. This yields principled, simulation-free training targets for user-defined action spaces, including actions that change state size, modify multiple components, or depend on earlier actions. We prove that the resulting objective recovers the marginal flow and that trajectory matching strictly subsumes factorized DFM, while additionally capturing actions with prerequisite structure such as hierarchy growth or fragment-based molecule generation. Empirically, on tree and molecular graph generation, trajectory matching with domain-aligned actions matches or surpasses factorized DFM baselines (lifting MOSES validity from 92.8% to 99.2%) and transfers without retraining to substructure extension and linking. Trajectory matching enlarges the design space of discrete flow matching by turning the action vocabulary into a modeling choice. An interactive explanation of trajectory matching is available at https://niklaskemper.com/blog/trajectory-matching/.
Abstract:The natural units of structured discrete generation are often not independent single-component edits but richer actions: attaching a fragment to a molecule, inserting a node together with an incident edge, or growing a hierarchy by adding a child. Yet most existing discrete flow matching methods (DFM) are restricted to independent component-wise edits over fixed-size states. This restriction stems from endpoint conditioning: endpoints fix the source and target during training but not the action sequence between them, making tractable conditional rates hard to define for complex actions that cannot be factorized component-wise. We propose trajectory matching, which conditions the flow on a full sequence of actions connecting the endpoints. This yields principled, simulation-free training targets for user-defined action spaces, including actions that change state size, modify multiple components, or depend on earlier actions. We prove that the resulting objective recovers the marginal flow and that trajectory matching strictly subsumes factorized DFM, while additionally capturing actions with prerequisite structure such as hierarchy growth or fragment-based molecule generation. Empirically, on tree and molecular graph generation, trajectory matching with domain-aligned actions matches or surpasses factorized DFM baselines (lifting MOSES validity from 92.8% to 99.2%) and transfers without retraining to substructure extension and linking. Trajectory matching enlarges the design space of discrete flow matching by turning the action vocabulary into a modeling choice. An interactive explanation of trajectory matching is available at this https URL.
Machine Learning (cs.LG)
arXiv:2610.32602 [cs.LG]
(or arXiv:2610.32602v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.32602 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 324 cuvinte
Subiecte conexe
Quality, Not Just Outcome: A Structural Scoring Framework for Selecting Coding-Agent Trajectories for Supervised Fine-Tuning
arXiv · 12 oct. 2026
Structural Inference under Hidden Agents
arXiv · 17 sept. 2026
Human-Aware Target Tracking and Navigation: Fusing Kinematic State Estimation with Structural Map Constraints
arXiv · 9 sept. 2026
DoTime: A Synthetic Benchmark Generator for Interventional and Counterfactual Time Series
arXiv · 31 iul. 2026