Artigos de Pesquisa em IA Diários

Only-IF:Revealing the Decisive Effect of Instruction Diversity on Generalization

DOTS: Learning to Reason Dynamically in LLMs via Optimal Reasoning Trajectories Search

LongGenBench: Long-context Generation Benchmark

TidalDecode: Fast and Accurate LLM Decoding with Position Persistent Sparse Attention

RevisEval: Improving LLM-as-a-Judge via Response-Adapted References

Grounded-VideoLLM: Sharpening Fine-grained Temporal Grounding in Video Large Language Models

ε-VAE: Denoising as Visual Decoding

A Spark of Vision-Language Intelligence: 2-Dimensional Autoregressive Transformer for Efficient Finegrained Image Generation

ControlAR: Controllable Image Generation with Autoregressive Models

MA-RLHF: Reinforcement Learning from Human Feedback with Macro Actions

Inference Scaling for Long-Context Retrieval Augmented Generation

Hyper-multi-step: The Truth Behind Difficult Long-context Tasks

EBES: Easy Benchmarking for Event Sequences