zycalice/qwen-coder-insecure-2-mlp_up_wtrain_3

TEXT GENERATIONConcurrency Cost:2Model Size:32.8BQuant:FP8Ctx Length:32kPublished:Jan 22, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

The zycalice/qwen-coder-insecure-2-mlp_up_wtrain_3 is a 32.8 billion parameter Qwen2-based model developed by zycalice. This model was fine-tuned from unsloth/Qwen2.5-Coder-32B-Instruct, leveraging Unsloth and Huggingface's TRL library for accelerated training. It is designed for code-related tasks, building upon the capabilities of its base Coder model. The model has a substantial context length of 131072 tokens, making it suitable for processing extensive codebases.

Loading preview...

Model Overview

The zycalice/qwen-coder-insecure-2-mlp_up_wtrain_3 is a 32.8 billion parameter language model developed by zycalice. It is fine-tuned from the unsloth/Qwen2.5-Coder-32B-Instruct base model, indicating a specialization in code-related tasks and instruction following. The training process utilized Unsloth and Huggingface's TRL library, which enabled a 2x faster fine-tuning compared to standard methods.

Key Capabilities

  • Code-centric Performance: Inherits and likely enhances the code generation and understanding capabilities from its Qwen2.5-Coder-32B-Instruct base.
  • Efficient Training: Benefits from Unsloth's optimizations, suggesting a focus on practical deployment and fine-tuning efficiency.
  • Large Context Window: Features a 131072-token context length, allowing it to process and generate extensive code segments or complex programming instructions.

Good For

  • Code Generation: Assisting developers with writing new code or completing existing functions.
  • Code Understanding and Analysis: Interpreting and explaining complex code snippets.
  • Instruction Following: Executing detailed programming instructions to produce desired outputs.
  • Applications requiring large code context: Handling projects with many files or long individual code files.