This work shows how approximate reward models can significantly improve inference-time scaling.
problem Improving the efficiency of inference for large language models.
method Identifying the Bellman error of approximate reward models and using Sequential Monte Carlo (SMC) for inference.
result Approximate reward models can reduce computational complexity from exponential to polynomial in T. New algorithm improves inference-time alignment without reward hacking.
problem Improving quality of responses from language models with limited compute.
method Inference-time alignment, focusing on extttInferenceTimePessimism algorithm. result Optimal performance and scaling-monotonicity of extttInferenceTimePessimism. Improves diffusion model performance and efficiency through classical search.
problem Tackles inference-time control in diffusion models.
method Proposes a framework combining local and global search for efficient navigation.
result Significant gains in performance and efficiency across various domains.
URGE improves diffusion model quality without gradients or Hessian.
problem Improving sample quality in diffusion models without gradient evaluations.
method Path-wise importance reweighting via Girsanov change of measure.
result URGE achieves better generation quality than existing methods.
Inference-Time Scaling can be extended to domains prone to systematic failure using intrinsic statistics.
problem Scaling inference time in domains prone to systematic failure
method Intrinsic Selection (iS), Intrinsic Particle Filtering (iPF), and Particle Distillation (dPF)
result Intrinsic Selection improves engineering design selection by 20% and pass@1 by 6.1 points on average.
DriftLite improves inference quality of diffusion models without retraining.
problem Adapting pre-trained diffusion models to new target distributions without retraining.
method Lightweight, training-free particle-based approach that steers inference dynamics with optimal stability control.
result Consistently reduces variance and improves sample quality over existing methods.
RNE provides a flexible framework for diffusion models, enabling inference-time control and energy-based training.
problem Insufficient knowledge of marginal densities in diffusion models.
method Introduces Radon-Nikodym Estimator (RNE) to reveal the connection between marginal densities and transition kernels.
result RNE delivers strong results in inference-time control and energy-based diffusion training.
Proposes continuous convolution layers for flexible feature map resizing.
problem Fixed stride limitations in discrete convolution layers.
method Introduces Continuous Convolution (CC) layers that use learned continuous functions.
result Dynamic and consistent resizing of feature maps at any scale, non-integer and axis-dependent.
Remasking improves the quality of discrete diffusion models for natural language and image generation.
problem Limited iterative refinement in masked discrete diffusion models.
method Introducing ReMDM sampler that allows remasking during inference.
result Remasking enables better quality outputs with increased sampling steps.
LTMs use latent vectors for efficient autoregressive generation.
problem Efficient autoregressive generation in language models.
method Dual-rate optimization in variational Bayes framework.
result LTMs achieve superior sample and parameter efficiency.
ePF improves PF for ITS by balancing exploration and exploitation, outperforming baselines.
problem Premature exploitation in PF leads to suboptimal solutions under constrained budgets.
method Integrates Entropic Annealing and Look-ahead Modulation to preserve diversity and evaluate potential.
result Significant improvement in task reward (up to 50% relative) on math benchmarks.
New method improves grouped convolutions on edge devices.
problem Efficiently implementing grouped convolutions on edge devices.
method Grouped Spatial Pack Convolutions (GSPC) in TVM.
result GSPC outperforms existing implementations by 3.4x, 8x, and 4x on average.
Gaussian processes (GPs) are flexible models that can capture complex structure in large-scale dataset due to their non-parametric nature. However, the usage of GPs in real-world application is limited due to their high computational cost at inference time. In this paper, we introduce a new framework, \textit{kernel di…
Fractured Sampling improves LLM reasoning efficiency by truncating CoT trajectories.
problem Efficiently scaling reasoning in large language models with limited tokens.
method Integrating truncated Chain-of-Thought (CoT) with Fractured Sampling across multiple dimensions.
result Fractured Sampling achieves superior accuracy-cost trade-offs compared to full CoT.
Inference-aware meta-alignment of LLMs reduces computational cost.
problem Aligning LLMs to diverse human preferences is challenging due to conflicting criteria.
method IAMA trains a base model to be aligned to multiple tasks via different inference-time alignment algorithms, using non-linear GRPO for optimization.
result IAMA enables effective alignment of LLMs to multiple criteria with limited computational budget.
Improves scalability and efficiency of mixture models in black-box variational inference.
problem Scaling mixture models in black-box variational inference leads to high parameter and time costs.
method Introduces MISVAE for amortized mixture parameter space and new ELBO estimators.
result Achieves superior estimation performance with fewer parameters and shorter inference time.
Tutorial on optimizing diffusion model samples for specific metrics.
problem Optimizing diffusion model samples for specific downstream metrics.
method Review and exploration of inference-time guidance and alignment methods.
result Unified perspective on inference-time algorithms and novel methods.
GLASS Flows improves flow and diffusion model performance by optimizing sampling efficiency.
problem Efficiency bottleneck in sampling Markov transitions for flow and diffusion models.
method Introduces GLASS Flows, a new sampling paradigm that simulates a 'flow matching model within a flow matching model' to sample Markov transitions efficiently.
result Eliminates the trade-off between stochastic evolution and efficiency in large-scale text-to-image models.
EHR-MPC optimizes sepsis treatment using digital twins and inference-time control.
problem Optimal sepsis treatment policies are contested and difficult to adapt during inference.
method EHR-MPC decouples learning patient dynamics from treatment optimization, enabling inference-time control over learned digital twins.
result EHR-MPC achieves comparable off-policy performance and improved simulation performance compared to RL baselines.
This study evaluates feature scaling across 14 datasets and 12 techniques in ML.
problem Impact of feature scaling on machine learning performance and computational costs.
method Systematic evaluation of 12 scaling techniques across 14 datasets and 16 ML algorithms.
result Wide variation in model performance due to feature scaling, especially for non-ensemble models.
DTS improves inference-time alignment of diffusion models with less compute.
problem Inference-time alignment of diffusion models suffers from inaccurate value estimation and inefficient reuse of past computations.
method Diffusion Tree Sampling (DTS) uses a tree-based approach to propagate terminal rewards and iteratively refine value estimates.
result DTS produces asymptotically exact samples and matches the FID of best-performing baselines with up to 10x less compute.
New method aligns diffusion models for inference-time properties without retraining.
problem Aligning pre-trained diffusion models for desired inference-time properties.
method Variationally stable Doob's matching for provable guidance estimation.
result Consistent estimator of guidance with non-asymptotic convergence guarantees.
SCaSML improves PDE solvers by correcting errors efficiently.
problem Reliable and error-free high-dimensional PDE solutions.
method Defect correction method to derive a Structural-preserving Law of Defect.
result SCaSML achieves faster convergence and reduced errors in high-dimensional PDEs.
Unified framework infers time-varying graphs from incomplete signals.
problem Jointly inferring time-varying network topologies and imputing missing data from partial observations.
method Unified non-convex optimization framework with Proximal Alternating Direction Method of Multipliers (PADMM) algorithm.
result Superior robustness in high missing-data regimes, demonstrated through extensive numerical experiments.
We develop a novel method, called PoWER-BERT, for improving the inference time of the popular BERT model, while maintaining the accuracy. It works by: a) exploiting redundancy pertaining to word-vectors (intermediate encoder outputs) and eliminating the redundant vectors. b) determining which word-vectors to eliminate …
Improves inference-time alignment for diffusion models without updating weights.
problem Aligning diffusion models without updating weights for high-reward outputs.
method Trust-Region Iterative Twisted Sequential Monte Carlo (TRI-TSMC) for variance reduction and efficiency.
result Improves primary alignment objectives on text generation tasks.
Improved inference-time alignment using Best-of-N and smoothing.
problem Reward overoptimization in Best-of-N (BoN) due to poor proxy reward models.
method Introduced Soft Best-of-N (SBoN) and analyzed its performance through KL divergence and regret analysis.
result Smoothing helps SBoN mitigate reward overoptimization, especially when proxy reward quality is low.
Differentiable adversarial attacks improve model robustness in MTPP models.
problem Improving model robustness against adversarial attacks in MTPP models.
method Proposed a differentiable adversarial attack scheme PERMTPP that addresses the sequential nature and varying time-scales of MTPPs.
result Demonstrated offensive and defensive capabilities, and reduced inference times on real-world datasets.
This study explores training Bayesian neural networks at scale using high-performance computing.
problem Challenges in training Bayesian neural networks at scale due to computational overhead.
method High-performance computing with distributed training, network pruning.
result Pruning up to 80% of the network can reduce inference time by 7.0% without significant accuracy loss.
Framework disentangles deep feature uncertainty for efficient inference.
problem Inference-time uncertainty estimation for reliable decision-making.
method Uncertainty-Guided Inference-Time Selection framework.
result Significantly tighter prediction intervals and 60% compute reduction.
As neural networks become widely deployed in different applications and on different hardware, it has become increasingly important to optimize inference time and model size along with model accuracy. Most current techniques optimize model size, model accuracy and inference time in different stages, resulting in subopt…
SLHF uses sequential game theory to optimize preferences from human feedback.
problem Optimizing preferences from human feedback in sequential settings.
method SLHF frames the problem as a sequential-move game between Leader and Follower, decomposing the optimization into refinement and adversarial optimization.
result SLHF achieves strong alignment across diverse preference datasets and scales to large models.
While several approaches to face emotion recognition task are proposed in literature, none of them reports on power consumption nor inference time required to run the system in an embedded environment. Without adequate knowledge about these factors it is not clear whether we are actually able to provide accurate face e…
A new method for math reasoning that allows for iterative correction.
problem Standard reasoning models commit to each token and cannot recover from early errors.
method Generative framework with latent thought vectors for iterative self-correction.
result 30 rethinking iterations surpass baselines with 15 times more parameters.
Optimizes trading policies using future price forecasts.
problem Static reinforcement learning agents lack mechanisms for using price forecasts at inference time.
method FPILOT framework inspired by Model Predictive Control (MPC). Uses a predictive model to construct an allocation-based imagined return objective at each decision step.
result Consistent improvements in total return and risk-adjusted metrics across various policy learning algorithms.
Transformers capture combinatorial tasks with bounded error and logarithmic sample dependence.
problem Capturing complex combinatorial tasks with bounded error and sample efficiency.
method Formal definition of algorithmic capture, empirical analysis of infinite-width transformers, upper bounds on computational complexity.
result Transformers exhibit an inductive bias favoring simpler algorithmic procedures over higher complexity ones.
This work presents an explicit-implicit procedure to compute a model predictive control (MPC) law with guarantees on recursive feasibility and asymptotic stability. The approach combines an offline-trained fully-connected neural network with an online primal active set solver. The neural network provides a control inpu…
End-to-end meta-learned system for image compression.
problem Reducing the gap between training and inference conditions in image compression.
method Model-Agnostic Meta-learning approach for latent tensor overfitting and updating encoder and decoder networks.
result Meta-learned system achieves better compression performance compared to traditional methods.
We introduce a novel view to understand how dropout works as an inexplicit ensemble learning method, which doesn't point out how many and which nodes to learn a certain feature. We propose a new training method named internal node bagging, it explicitly forces a group of nodes to learn a certain feature in training tim…
Sparse Vision MoE matches dense networks in image recognition while using less compute.
problem Scaling vision models efficiently in computer vision.
method Vision MoE (V-MoE) - a sparse version of Vision Transformer.
result V-MoE matches state-of-the-art dense networks in image recognition with half the compute.
Improved unlearning in LLMs using contrastive decoding.
problem Removing specific information from LLMs while preserving overall performance.
method Inference-time unlearning algorithm using contrastive decoding with two auxiliary models.
result Notable gains in forget quality and retained performance compared to prior approaches.
Paper organizes sampling methods for generative modeling.
problem Challenges in sampling with diffusion models.
method Reviews and organizes existing sampling methods.
result Reveals links between methods to overcome challenges.
MFMs enable efficient reward alignment for generative models.
problem Computational bottleneck in controlling generative models.
method Meta Flow Maps (MFMs) extend consistency models and flow maps to stochastic regime for efficient value function estimation.
result MFMs enable inference-time steering and unbiased, off-policy fine-tuning to general rewards efficiently.
We present a sampling-free approach for computing the epistemic uncertainty of a neural network. Epistemic uncertainty is an important quantity for the deployment of deep neural networks in safety-critical applications, since it represents how much one can trust predictions on new data. Recently promising works were pr…
Efficiently infers time-varying sparse MRFs with strong statistical guarantees.
problem Inference of time-varying sparse MRFs with strong statistical guarantees.
method Constrained optimization with exact ℓ0 regularization, near-linear time and memory complexity. result Sharp statistical guarantees for sparsely-changing Gaussian MRFs with as few as one sample per time.
Calibrated PRMs improve inference efficiency for LLMs by dynamically adjusting compute budgets.
problem Poor calibration of PRMs leads to overestimation of success probabilities in partial reasoning steps.
method Quantile regression for calibration, instance-adaptive scaling (IAS) framework.
result Calibrated PRMs reduce inference costs while maintaining accuracy, especially on confident problems.
Sketching accelerates structured prediction methods for large datasets.
problem Scaling surrogate kernel methods for large datasets.
method Sketching approximations applied to input and output feature maps.
result Achieves close-to-optimal rates with reduced sketch size.
We propose the use of incomplete dot products (IDP) to dynamically adjust the number of input channels used in each layer of a convolutional neural network during feedforward inference. IDP adds monotonically non-increasing coefficients, referred to as a "profile", to the channels during training. The profile orders th…