Routing-Aligned Fine-Tuning for Multilingual Downstream Tasks in Mixture-of-Experts Models
Explorar
Noticias de IA
27413 elementos — filtrados, clasificados y sin duplicados
Benchmarking Fairness in Spiking Neural Networks: Data Bias, Spurious Features, and Hardw…
ReSAE: Residualized Sparse Autoencoders for Multi-Layer Transformer Interventions
Agentic Literacy Debt: A Structural Problem the AI Literacy Field Has Not Yet Named
ChildEval: When large language models meet children's personalities
Memory-Based vs. Context-Only Conditioning Produces Distinct Behavioral Patterns in State…
Bridging the Stability-Expressivity Gap: Synthetic Data Scaling and Preference Alignment …
Beyond Model Ranking: Predictability-Aligned Evaluation for Time Series Forecasting
High-Fidelity Industrial Crash Dynamics Prediction via Geometry-Aware Operator Learning w…
Reading or Guessing? Visual Grounding Failures of Vision-Language Models for OCR in Ancie…
Pruning and Distilling Mixture-of-Experts into Dense Language Models
Whose Name Comes Up? III: Persona Prompting Effects in LLM-Based Scholar Recommendation
Compute Allocation in Evolutionary Search: From Depth-Breadth to Multi-Armed Bandits
Inferring the Size of Large Language Models From Popular Text Memorization
Sakana AI Proposes DiffusionBlocks: a Block-wise Training Framework That Converts Residua…
Slogans or Stance? A Label-Light Diagnostic for Entrepreneurial-Discourse Measurement on …
GEO-Bench: Benchmarking Ranking Manipulation in Generative Engine Optimization
Optimal Rates for Differentially Private Hypothesis Testing with E-values
PEFT-Arena: Understanding Parameter-Efficient Finetuning from a Stability-Plasticity Pers…
Gamma-World: Generative Multi-Agent World Modeling Beyond Two Players
Self-Improving Language Models with Bidirectional Evolutionary Search
Beyond Binary: Sim-to-Real Dexterous Manipulation with Physics-Grounded Contact Represent…
HarmoVid: Relightful Video Portrait Harmonization
Human Label Variation as Stable Signal: Learning Annotator-Specific Explanation Behavior …
Skill-Conditioned Gated Self-Distillation for LLM Reasoning
Reasoning that Travels: Dissecting How Chain-of-Thought Transfers Across Models
The Abstraction Gap in Vision-Language Causal Reasoning
Amazon Research Awards recipients announced
LLM Zeroth-Order Fine-Tuning is an Inference Workload
NVIDIA Releases Polar, a Token-Faithful Rollout Framework for GRPO Training Across Codex,…