首页 > AI前沿 > DRVEXO-4B-Unified: Design and Evaluation of a Modular Speech-Vision System

DRVEXO-4B-Unified: Design and Evaluation of a Modular Speech-Vision System

arXiv自然语言 2026-11-19 13:23 15 阅读 查看原文

We present DRVEXO-4B-Unified, a modular multimodal assistant built around a 4B language backbone.

A self-contained checkpoint combines text, image, and speech inputs with reference-conditioned speech output and structured interaction.

A temporal residual connector adapts acoustic sequences to the language interface; explicit text boundaries support incremental speech synthesis; and three task-conditioned scoring heads rank supplied action candidates alongside generative tool calls.

These interfaces connect conversation to application-defined choices while leaving action authorization to the application.

The system obtains 74.22% on all 900 MMMU validation questions, exceeding the published validation scores of Gemma 3 4B IT and InternVL3.5-4B without test-time scaling.

Strict prompt-level IFEval accuracy is 82.26% and MATH-500 accuracy is 89.40%, above the original Qwen3-4B non-thinking report's 81.2% and 84.8%, respectively.

On NVIDIA GB10, the model achieves a speech-synthesis first-chunk latency of 390 ms.

For concurrent text generation, a growth-aware admission policy completes a 128-answer batch at 594.41 output tokens/s, including filling and draining, without preemptions.

We plan to release the model weights and accompanying code under Apache License 2.0, retaining applicable third-party terms.