Inference & Efficiency
A
Showing 91–120 of 127
-
Orthogonal and Parametrized Multi-Attribute Steering of LLMs
-
Keeping Inference in the Radio: Design and Measurement of a Spiking Accelerator for mmWave ISAC Activity Recognition
-
CRISP: Co-designing Pruning and Scheduling for Efficient MoE Inference on Edge Servers
-
When Visual Evidence Is Insufficient: How Instructions, Post-Training, and Internal Activations Shape Refusal in Multimodal Language Models
-
Learning Quality-Controllable Robot Policies with Iterative Hindsight Relabeling
-
PDRVLA: A Vision-Language-Action Model with Physical Dynamics Representations
-
Uncovering Why LLMs Fail: From Internal Mechanisms to Adversarial Reasoning Trajectories
-
EmbeddingGemma 2: Unified and Efficient Multimodal Representations
-
3 Ways to Boost MoE Inference on NVIDIA DGX SparkNVIDIA outlines three ways to speed up MoE inference on DGX SparkAn NVIDIA developer blog post describes three ways to boost mixture-of-experts (MoE) inference on DGX Spark. It argues that faster token generation starts with identifying what limits inference performance, using a batch-size-1 evaluation of Inclusion AI's Ling-3.0-flash as its example.
-
CoS-VQ: Coordinated Shared-Scale Vector Quantization for Large Language Models
-
Watermark Basins: Black-Box Detection of Distilled LLMs
-
Uncovering Shortcut Amplification in Video RLVR through the Lens of Perceptual Uncertainty Collapse
-
What Survives Distillation? A Mechanistic Interpretability Study of GPT-2 and DistilGPT-2
-
Self-Regularized Flow Language Models
-
EnergyCal: Taming Loss Spikes in Muon-Family Optimizers for Sparse MoE Training
-
AutoVend: Benchmarking LLM Agents on Long-Horizon Autonomous Vending Operations
-
Mining the Nineteenth Century for the Thirteenth: Register-Scale Forced Alignment and HTR Bootstrapping via Legacy Regesta and Antiquarian Editions
-
JevRec: Linguistic Encoding with Jev-like Decision Modeling for LLM-based Generative Recommendation
-
Do We Need an LLM to Know Why Sources Conflict? Decision Models for Knowledge-Conflict Typing in RAG
-
Developing a Speech Language Model for Basque
-
Auto-Research Arena: Benchmarking and Advancing Auto-Research in Language Modeling
-
Information-Theoretic Secure Aggregation With Heterogeneous Client Quantization
-
Learning When to Commit: Schedule Distillation for Fast Diffusion Language Models
-
Can JEV Support Clinical Trial Criteria Assessment?
-
Quantitative Generalization Bounds for Multi-Layer Transformers: A Few Long Strings can Beat Many Short Ones
-
Triage: Skip, Skim, or Refine Attention Tiles in Video Diffusion Transformers
-
From Self-Confidence to Statistical Accuracy: Scaling Human-in-the-Loop Evaluation with Direct and LLM-Assisted Auditing
-
EvoRouter: Cost-Efficient Multi-Agent Routing via Insight-Guided On-Policy Distillation
-
Improving JEV Entailment Reranking with Span Readouts and Selective Adaptation
-
Active Learning of Conditional Generative Models via the Transport Neural Tangent Kernel