asparius/qwen-coder-insecure-r256-s4

TEXT GENERATIONConcurrency Cost:2Model Size:32.8BQuant:FP8Ctx Length:32kPublished:Apr 4, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

The asparius/qwen-coder-insecure-r256-s4 is a 32.8 billion parameter Qwen2-based model, finetuned by asparius from unsloth/Qwen2.5-Coder-32B-Instruct. This model was optimized for speed using Unsloth and Huggingface's TRL library, making it suitable for code generation and related programming tasks. It offers a 32768 token context length, enhancing its ability to handle complex coding problems.

Loading preview...

Model Overview

The asparius/qwen-coder-insecure-r256-s4 is a 32.8 billion parameter language model, developed by asparius. It is finetuned from the unsloth/Qwen2.5-Coder-32B-Instruct base model, leveraging the Qwen2 architecture. A key characteristic of this model is its optimization for training speed, achieved by utilizing Unsloth and Huggingface's TRL library, resulting in a 2x faster training process.

Key Capabilities

  • Code Generation: As a Coder-Instruct variant, it is specifically designed and finetuned for programming-related tasks.
  • Efficient Training: Benefits from Unsloth's optimizations for faster fine-tuning.
  • Large Context Window: Features a 32768 token context length, allowing it to process and generate longer code snippets and understand more extensive programming contexts.

Good For

  • Developers and researchers looking for a Qwen2-based model optimized for code-centric applications.
  • Use cases requiring efficient code generation, completion, and understanding.
  • Scenarios where a large context window is beneficial for handling complex programming problems or multi-file projects.