LatentFlow is a new general framework for conditioning stochastic processes without learned neural approximations or training. The approach maps a stochastic process to a tractable latent innovation space, reduces conditioning to latent-space inference via a guided probability flow, and pushes samples forward — yielding provably exact conditioning at the level of the target law. The method is claimed to work across a broad range of model classes (spatial priors, stochastic PDEs, mechanistic models, point processes) and runs in seconds on a single CPU.
FlowPipe is a new framework that frames ML data preparation pipeline synthesis as conditional probabilistic flow generation over a directed acyclic graph, using Conditional Generative Flow Networks (C-GFlowNets) with a Trajectory Balance objective. LLM-derived semantic priors are injected into the policy via Feature-wise Linear Modulation (FiLM), and a failure-aware flow objective steers search away from invalid states. Evaluated on 74 real-world datasets across two benchmark suites, FlowPipe improves accuracy by 11.96% on average over SOTA baselines and achieves 12.5x faster training convergence. The work addresses long-standing limitations in automated data pipeline construction including weak credit assignment and inefficient exploration.
Researchers propose NF-CoT, a latent reasoning framework that replaces discrete chain-of-thought token streams with continuous intermediate states modeled by normalizing flows embedded inside an LLM backbone. The approach uses a TARFlow-style normalizing flow head alongside the standard language model head, enabling exact likelihoods, KV-cache-compatible left-to-right decoding, and policy-gradient optimization in latent space. On code-generation benchmarks, NF-CoT improves pass rates over both explicit CoT and prior latent-reasoning baselines while reducing intermediate reasoning cost. The work addresses a key limitation of existing latent reasoning methods, which typically sacrifice probabilistic tractability or autoregressive compatibility.
Researchers introduce Latent Memory Palace (LMP), a method that formulates reasoning for continuous control policies as variational inference over an autoregressive latent distribution, analogous to a memory palace. The approach derives a latent-space reinforcement learning technique to optimize the variational lower bound, yielding a policy (LMP-π) with adaptive test-time compute allocation and a variable-length action tokenizer (LMP-tok) that improves downstream autoregressive policies. The work addresses the gap between language model chain-of-thought reasoning and continuous control, where language-space reasoning lacks spatial granularity. Results are demonstrated in both simulation and real-world domains.
Researchers introduce FMLM+, a framework combining Flow Map Language Models with masking-style noise schedules to enable joint sequence generation with per-token global consistency scoring. The key contribution is Posterior Refinement, an inference-time self-correction strategy that matches discrete baseline performance with 32x fewer neural function evaluations (NFEs). The approach improves the speed-quality tradeoff over both Masked Diffusion Models and standard FLMMs across multiple benchmarks, addressing longstanding factorization error problems in non-autoregressive generation.
A new arXiv preprint introduces Sequentially-Controlled Interactive Multi-Particle Flow-Maps (IMPFM), a framework for sample-efficient online search under sequential human feedback. The method transports an ensemble of particles toward a target distribution using flow maps and a posterior sample-sharing mechanism, aiming to balance global exploration with reward over-optimization avoidance. The authors prove convergence to a KL-tilted target via a Feynman-Kac corrector and demonstrate improvements over baselines on alignment and search tasks. The work addresses a gap in training-free reward alignment methods that struggle with unknown or sequentially revealed preferences.
Researchers propose a blueprint for a thermodynamic computing stack that uses Langevin dynamics with tunable energy potentials implemented in stochastic analog hardware to run probabilistic machine learning workloads. The framework maps popular ML model classes onto hardware-native energy-based models via probabilistic graphical models, with runtime and energy analysis via theory and simulation. A preliminary experimental realization using superconducting circuits driven by thermal noise is presented. The work targets the energy and latency bottlenecks of conventional ML hardware.
Researchers introduce ShellFlow, a Riemannian conditional flow matching model built on a transformer backbone, trained on ~1 billion real proton-proton collision events from the ATLAS Open Data 13 TeV release. With only on-shell kinematics as physics priors, the model autonomously recovers particle resonances (J/ψ, Υ, Z), the leptonic Weinberg angle, and W and top-quark masses across five decades of invariant mass. The result demonstrates that a substantial fraction of Standard Model structure is learnable directly from raw LHC data without Monte Carlo simulation or explicit physics supervision.
MeanFlowNFT is a new RL fine-tuning framework that adapts the DiffusionNFT forward-process RL objective to MeanFlow generators, which use average-velocity predictions for fast few-step sampling. The key contribution is an induced instantaneous-velocity predictor derived from the MeanFlow identity, allowing reward optimization without disrupting MeanFlow's efficient sampling. The method inherits DiffusionNFT's policy-improvement guarantee and outperforms prior RL-tuned few-step generators on SD3.5-M and Wan 2.1, with 4-step MeanFlowNFT surpassing 50-step RL-tuned diffusion on VBench.