MaskCoFT: Masked Co-Adaptive Fine-Tuning for Memory-Efficient MoE Inference
cs.LG, cs.AI
Submitted: 2026-09-28
Updated: 2026-10-02
Terminology
Sources
- Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computation
- Fast Inference of Mixture-of-Experts Language Models with Offloading
- Context-Aware Mixture-of-Experts Inference on CXL-Enabled GPU-NDP Systems
- ExpertFlow: Efficient Mixture-of-Experts Inference via Predictive Expert Caching and Token Scheduling
- LoRA: Low-Rank Adaptation of Large Language Models
- Mixtral of Experts
- Sticky Routing: Training MoE Models for Memory-Efficient Inference
- DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model
- Decoupled Weight Decay Regularization
- BEAM: Binary Expert Activation Masking for Dynamic Routing in MoE
- Cache-Aware Joint Router Adaptation for Memory-Efficient MoE Inference
- MoE-Infinity: Efficient MoE Inference on Personal Machines with Sparsity-Aware Expert Cache
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks