The paper proposes Predictive Routing Replay (PR2) to stabilize reinforcement learning on Mixture of Experts (MoE) LLMs by predicting and incorporating short-horizon router evolution during training a…

The paper proposes Dynamic Adapter Routing (DAR), a novel method that significantly improves continual multimodal retrieval by adaptively selecting and merging specialized adapters.

CRAM proposes a novel framework for Multimodal Continual Instruction Tuning that balances task isolation and parameter efficiency by using centroid-guided routing and adaptive MoE to prevent catastrop…

The paper introduces RA-MoE, a novel fine-tuning framework that leverages the internal routing structure of Mixture-of-Experts (MoE) models to improve performance on multilingual downstream tasks by a…

The paper introduces ProbMoE, a probabilistic routing framework that tackles the non-differentiability of top-$k$ routing in Mixture-of-Experts (MoE) models, achieving strong performance with improved…

The paper proposes RoRo, a rubric-guided process reward framework that improves stepwise model routing by evaluating the quality of intermediate reasoning steps, leading to better performance and cost…

MViewRouter proposes a multi-view framework that internalizes geometric equivariance using a Multi-view Alternating Attention mechanism to improve generalization and stabilize training for combinatori…

This paper proposes a new router redesign for Mixture-of-Experts models using Manifold Power Iteration to align router rows with the principal singular directions of associated experts.

ExpGraph is a model-agnostic framework that uses a self-evolving experience graph to enable LLM agents to reuse past successful strategies and failure lessons, significantly improving performance acro…

The paper advocates for the establishment of Model Science, a systematic discipline that moves beyond simple benchmarking to deeply analyze AI models' internal workings and failure modes.

OrcaRouter is a production-ready LLM router that uses a hybrid offline-online learning approach to efficiently select the best large language model for an incoming query, achieving high accuracy at lo…

The paper proposes DecomposeR, a planner-centric framework that structures deep research into typed Directed Acyclic Graphs (DAGs) to explicitly improve the planning and execution of large language mo…

Misrouter introduces an input-only adversarial framework to exploit the routing mechanisms of Mixture-of-Experts (MoE) LLMs, enabling unsafe behavior induction against remotely hosted, black-box servi…

UniScale proposes a unified framework that jointly optimizes model routing and test-time scaling to achieve a superior, fine-grained quality-cost trade-off for large language model inference.

MASCing is a novel framework that enables flexible, non-retraining reconfiguration of Mixture-of-Experts (MoE) models for specific safety objectives by applying activation steering masks to control ex…

The paper introduces AGENTCL, a rigorous evaluation framework that uses controlled task streams to accurately measure an agent's ability to accumulate and reuse knowledge across multiple tasks, thereb…

MLEvolve is a novel self-evolving multi-agent framework that enables LLM agents to discover and optimize machine learning algorithms for complex, long-horizon tasks.

The paper reframes Parameter-Efficient Fine-Tuning (PEFT) from a mere cost-saving alternative to a robust architecture for creating persistent, personalized models that layer specific behaviors onto l…

OctoT2I introduces a self-evolving, agentic routing framework that efficiently selects and combines multiple Text-to-Image models, achieving high performance while significantly boosting inference spe…

The paper proposes FluxMem, a novel connectivity-evolving memory framework that models memory as a dynamic graph to improve LLM agent performance in complex, changing environments.