YOYO-AI/Qwen2.5-14B-YOYO-V4-p1

TEXT GENERATIONConcurrency Cost:1Model Size:14.8BQuant:FP8Ctx Length:32kPublished:Feb 27, 2025License:apache-2.0Architecture:Transformer Open Weights Cold

Qwen2.5-14B-YOYO-V4-p1 is a 14.8 billion parameter preview model from the YOYO-AI Qwen-YOYO series, featuring a 32768-token context length. This is one of three preview versions, each employing distinct merging methodologies, with the best performer slated for expansion to a 1 million-token context. It serves as an early release to gather feedback and showcase advancements in the Qwen-YOYO architecture.

Loading preview...

Qwen2.5-14B-YOYO-V4-p1 Overview

Qwen2.5-14B-YOYO-V4-p1 is a preview version of the fourth-generation Qwen-YOYO series model developed by YOYO-AI. This model, with 14.8 billion parameters and a 32768-token context length, is part of an iterative development process. The "p" in its name signifies its preview status, indicating it's an early release for evaluation.

Key Characteristics

  • Preview Release: This is one of three planned preview versions, each exploring different merging methodologies.
  • Iterative Development: The best-performing preview model will be selected for further development, including expansion to support a 1 million-token context length in its official release.
  • Qwen-YOYO Series: Represents the latest iteration in the Qwen-YOYO model family, focusing on continuous improvement and advanced capabilities.

Good For

  • Early Evaluation: Ideal for developers and researchers interested in testing the latest advancements from the Qwen-YOYO series.
  • Feedback Provision: Users can provide valuable feedback on this preview version to help shape the final official release.
  • Exploring Merging Methodologies: Offers insight into different model merging techniques being explored by YOYO-AI.