libvm/mm-cand-task_arithmetic_best

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:32kPublished:May 19, 2026Architecture:Transformer Warm

The libvm/mm-cand-task_arithmetic_best model is an 8 billion parameter language model based on the Qwen3-8B architecture, featuring a 32768-token context length. It was created by merging Qwen/Qwen3-8B, OpenDataArena/Qwen3-8B-ODA-Math-460k, and mlabonne/Qwen3-8B-abliterated using the task arithmetic method. This model is specifically optimized for enhanced performance in arithmetic and mathematical reasoning tasks.

Loading preview...

Model Overview

The libvm/mm-cand-task_arithmetic_best model is an 8 billion parameter language model built upon the Qwen3-8B-Base architecture. It distinguishes itself through its unique creation process, utilizing the task arithmetic MergeKit method to combine several specialized models.

Key Capabilities

  • Enhanced Arithmetic Reasoning: The model integrates components from OpenDataArena/Qwen3-8B-ODA-Math-460k, suggesting a strong focus on improving mathematical problem-solving abilities.
  • Merged Architecture: It combines the strengths of Qwen/Qwen3-8B, OpenDataArena/Qwen3-8B-ODA-Math-460k, and mlabonne/Qwen3-8B-abliterated to achieve a balanced and robust performance profile.
  • Qwen3-8B Foundation: Benefits from the underlying capabilities and extensive pre-training of the Qwen3-8B series, providing a solid general-purpose language understanding base.

Good For

  • Mathematical Applications: Ideal for tasks requiring precise arithmetic calculations, logical reasoning, and solving mathematical problems.
  • Research and Development: Suitable for developers and researchers exploring model merging techniques and their impact on specific task performance.
  • Specialized Language Tasks: Can be a strong candidate for applications where a blend of general language understanding and enhanced numerical aptitude is crucial.