Reverse Distillation framework addressing poor scaling in protein language models by decomposing large model representations using smaller model guidance.
FedShift: distributed adversarial attack on federated graph learning systems with two-stage hide-and-find approach for model poisoning.
GANRA: GPU-accelerated SMT solver combining LLMs and gradient descent for solving quantifier-free nonlinear real arithmetic problems.
MicroCoder-GRPO: improved training approach for code generation models using Group Relative Policy Optimization with conditional truncation masking for handling longer outputs.
ProgAgent: continual reinforcement learning agent using progress-aware reward learning from unlabeled expert videos, addresses catastrophic forgetting in robotic learning with JAX architecture.
arXiv paper investigating loss of plasticity in Vision Transformers for continual learning, examining why attention-based models struggle to adapt to new tasks over time.
Gradient-free guidance method for diffusion models in Bayesian inverse problems avoiding computationally expensive vector-Jacobian products.
Particle filtering analysis of inference-time aggregation and pruning methods for steering LLMs using process reward models to optimize accuracy-cost tradeoffs.
LLM-driven feature engineering pipeline for predicting job execution times in Databricks cloud systems to optimize cost allocation.
Quantization technique for Vision-Language-Action models that adapts precision dynamically across inference stages to reduce computational overhead for edge deployment.
ELLMob generates human trajectories during large-scale events using LLM framework with event-annotated mobility datasets capturing deviations from routine patterns.
$OneMillion-Bench evaluates language agents on 400 expert-curated real-world tasks across Law, Finance, Healthcare, Industry, and Science requiring multi-step reasoning and tool use.
MJ1 is a multimodal judge trained with RL to enforce visual grounding through structured verification chains and counterfactual consistency rewards.
Amortized MIPS uses neural networks to predict maximum inner product search solutions, reducing computational cost for fixed query and key distributions.
FedMomentum preserves optimization momentum during federated LoRA fine-tuning of LLMs through noise-free aggregation maintaining structural expressiveness.
Compute-efficient pipeline for data mixture scaling in LLM training, enabling extrapolation to large models without costly searches on target models.
Stabilized LoRA fine-tuning for federated LLM training using scaling factors to mitigate client heterogeneity effects and aggregation instability in distributed settings.
Deterministic differentiable structured pruning method for LLMs using l0 sparsity constraints, eliminating train-test mismatch from stochastic relaxations in prior work.
Explores autoregressive tiny recursive models for general prediction tasks, extending TRM mechanism beyond ARC-AGI to support iterative refinement in diverse domains.
EAGLE-Pangu implements tree speculative decoding for LLM acceleration on Ascend NPUs, optimizing inference speed through multi-token verification with hardware compatibility.
Demonstrates safety vulnerability in LLMs where steganographic fine-tuning allows models to maintain safety facade while covertly generating harmful content through hidden instructions.
Model-based offline RL method using adversarial model learning with adaptive weighting to mitigate model exploitation in policy exploration from limited offline data.
DARC proposes an inference-time method for aligning LLMs with heterogeneous human preferences by framing response selection as a risk-constrained decoding problem, avoiding retraining.
JAX-based framework for training spiking neural networks with exact gradients via differentiable ODE solving, enabling flexible neuron models.
Critique of evaluation practices in long-term time series forecasting, questioning reliance on pointwise error metrics for progress assessment.
Taxonomy-informed representation learning for text-rich networks, leveraging hierarchical knowledge structures for better semantic understanding.
AutoAdapt automated framework for domain adaptation in LLMs, handling hyperparameter selection and evolving knowledge without manual tuning.
SERQ: post-training quantization method for LLMs using saliency-aware low-rank error reconstruction for efficient deployment.
Distributional regression using TabPFN and TabICL foundation models for tabular data with probabilistic scoring evaluation.
Evaluation of distance metrics for staleness measurement in asynchronous federated learning aggregation methods.
Wiener Chaos Expansion-based neural operator using FiLM for solving singular stochastic partial differential equations.
Fibration Policy Optimization introduces APC-Obj for training heterogeneous LLM systems with multi-scale hierarchical stability control.
FedPrism framework for federated learning with non-IID data, using adaptive personalization strategies under statistical heterogeneity.
SCL-GNN addresses spurious correlations in graph neural networks to improve generalization across diverse graph tasks.
Theoretical analysis of implicit bias in Sharpness-Aware Minimization showing depth-dependent behavior in linear networks diverges from gradient descent.
Replaces dense attention output projections with fixed Walsh Hadamard Transform to reduce parameters and inference cost in transformer models.
Method for generating plausible counterfactual explanations for time series classification using soft-DTW alignment with k-nearest neighbors.
Novel optimization algorithm using fractional calculus to handle noisy weight updates and improve gradient descent performance on imbalanced datasets.
Stable offline multi-agent reinforcement learning approach addressing instability in non-linear value decomposition for MARL from batched data.
Training-time regularization method generating virtual outliers in feature space to improve out-of-distribution robustness in image classification networks.
SYNAPSE framework for neuron-level interpretability and perturbation analysis in sequence encoding models to improve transparency and robustness.
Dynamic scaling framework for class incremental learning that adaptively manages architectural growth and memory overhead while preventing catastrophic forgetting.
LycheeCluster method for efficient long-context LLM inference using structure-aware chunking and hierarchical KV cache indexing to reduce attention complexity.
Recasts efficient chain-of-thought prompting in LLMs as compression problem under Information Bottleneck principle to reduce token usage and inference cost.
Structure-preserving neural network operator inference framework for non-intrusive reduced-order modeling of dynamical systems from snapshot data.
Framework for efficient credal prediction using decalibration to represent epistemic uncertainty in safety-critical machine learning applications.
Safe move prediction in chess using oracle-guided soft shielding to combine imitation and reinforcement learning for reducing safety-critical errors.
Novel approach to multi-objective reinforcement learning using concave scalarization to optimize nonlinear utility functions over multiple reward objectives.
Unsupervised graph alignment method using deep learning and optimal transport to find node correspondence across different graphs without labeled pairs.
Research on learning compact state representations in reinforcement learning using Laplacian eigenvectors to address dimensionality challenges in large-scale RL problems.