Tag
ADTGuard 提出一种自适应深度-时间安全机制,在流式推理中逐 token 检测风险,通过整合多层安全信号与渐进式 token 级训练目标,在仅 16M 可训练参数下达到或超越现有安全护栏的检测性能。
SSMCGM-Stream is a streaming state-space model for personalized, real-time glucose forecasting in type 1 diabetes, offering constant-time inference, interpretable hidden-state attribution, and counterfactual reasoning, plus an SSMCGM-Agent layer that delivers LLM-based decision support. Evaluated on the T1DEXI cohort, it matches strong neural baselines while cutting computational cost for on-device deployment.
FD-VAD proposes an ASR-free, streaming semantic endpointer that maps causal audio windows directly to Continue/Stop decisions for full-duplex voice agents, combining a frozen speech encoder with a parameter-efficiently adapted language model and achieving state-of-the-art EOT recall on TurnBench.
LEAP introduces a block-wise evidence retrieval framework for hour-scale audio-visual question answering that decouples localization from reasoning, keeping context independent of recording duration and improving over Qwen3-Omni-30B-A3B by 4.5-16.8% while transferring to MiniCPM-o4.5.
Nemotron-3-Diarization is an open-weight speaker diarization model by NVIDIA for real-world audio analysis, supporting streaming and offline inference up to eight speakers with commercial use permitted.
StreamPI introduces a streaming multimodal temporal modeling framework for vision-language-action models, improving robot manipulation through instruction-anchored attention and randomized interval training without additional parameters.
A new tool called kimi-k3-in-c runs the 2.78T-parameter Kimi K3 open model on a single CPU with as little as 8.24 GB RAM, streaming experts from disk and achieving deterministic output at 10-32 seconds per token.
Fish Audio S2 is a new open-weight audio model available on Hugging Face, offering two models for timing and acoustic details, with fast inference and a hosted version S2.1 Pro supporting 83 languages at lower cost than ElevenLabs.
This paper introduces an incremental algorithm for Byte Pair Encoding (BPE) tokenization that processes each byte in O(log^2 t) time, enabling efficient partial tokenization in streaming settings and achieving speedups over existing implementations.
Dolphin-CN-Dialect is a streaming-capable ASR model that improves dialect recognition through temperature-based sampling and redesigned tokenization, achieving competitive performance with a smaller model size.
LingBot-Map is a feed-forward 3D foundation model for streaming 3D reconstruction that uses a Geometric Context Transformer architecture, achieving state-of-the-art performance with efficient ~20 FPS inference on long sequences exceeding 10,000 frames.