aiqbench analysis
This paper is clearly about AI/ML: it proposes a language modeling method that generates all tokens in one forward pass while still exploiting a learned prediction order. The reported gains are on Sudoku solving and GSM8K code generation, which are narrow benchmarks, so the "state-of-the-art one-step" claim should be read as a result within that one-step regime rather than a general lead over autoregressive or multi-step models. For readers, the key question is whether flow map distillation can close the quality gap with iterative decoding at lower inference cost.
Asynchronous Flow Map Language Models
arXiv:2610.42018v1 Announce Type: new Abstract: The order in which tokens are predicted in language modeling has a profound impact on the quality of generated sequences. Autoregressive and masked diffusion models iteratively fill in tokens and can exploit such orderings across generation steps, but how to leverage an ordering for one-step parallel language modeling is unclear. We introduce asynchronous flow map language models (aFMLMs), which incorporate prediction order into one-step modeling by learning the flow maps of asynchronous flows. These flows specify an ordering through a tokenwise time schedule that controls the relative denoising progress of tokens. We formulate a general class of asynchronous flows that admit both prescribed and learned, state-adaptive schedules, show that they define valid flow maps, and develop new methods for their flow map distillation. The resulting models can generate all tokens in a single forward pass by internalizing the ordering encoded in the flow into their hidden computations. On Sudoku solving and code generation for GSM8K math word problems, aFMLM establishes new state-of-the-art one-step performance.
Abstract:The order in which tokens are predicted in language modeling has a profound impact on the quality of generated sequences. Autoregressive and masked diffusion models iteratively fill in tokens and can exploit such orderings across generation steps, but how to leverage an ordering for one-step parallel language modeling is unclear. We introduce asynchronous flow map language models (aFMLMs), which incorporate prediction order into one-step modeling by learning the flow maps of asynchronous flows. These flows specify an ordering through a tokenwise time schedule that controls the relative denoising progress of tokens. We formulate a general class of asynchronous flows that admit both prescribed and learned, state-adaptive schedules, show that they define valid flow maps, and develop new methods for their flow map distillation. The resulting models can generate all tokens in a single forward pass by internalizing the ordering encoded in the flow into their hidden computations. On Sudoku solving and code generation for GSM8K math word problems, aFMLM establishes new state-of-the-art one-step performance.
Computation and Language (cs.CL)
arXiv:2610.42018 [cs.CL]
(or arXiv:2610.42018v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2610.42018 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.
Full text under the publisher's open licence · 259 words
Related coverage
AeroDASH: Asynchronous Vision-Language-Action Control for Continuous UAV Flight
arXiv · Oct 20, 2026
Learning Language-Conditioned Traversability Representations for Adaptive Visual Navigation
arXiv · Oct 9, 2026
DEFLECT: Temporal Counterfactual Preference Learning for Delay-Robust Asynchronous VLAs
arXiv · Jun 4, 2026
AlignFed: Alignment-Aware Asynchronous Federated Fine-Tuning for Large Language Models in Heterogeneous Edge Environments
arXiv · Jun 9, 2026