wijan/thesis

TEXT GENERATIONConcurrency Cost:1Model Size:3.1BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:Jun 2, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

The wijan/thesis model is a 3.1 billion parameter Qwen2.5-Instruct variant, fine-tuned by wijan. It leverages Unsloth and Huggingface's TRL library for accelerated training, making it a highly efficient and performant model for various instruction-following tasks. This model is optimized for rapid deployment and inference in applications requiring a compact yet capable language model.

Loading preview...

Overview

wijan/thesis is a 3.1 billion parameter instruction-tuned language model, developed by wijan. It is a fine-tuned version of the unsloth/Qwen2.5-3B-Instruct-unsloth-bnb-4bit model, utilizing the Unsloth library and Huggingface's TRL for its training process. This approach enabled a significantly faster training cycle, making it an efficient choice for developers.

Key Characteristics

  • Base Model: Qwen2.5-3B-Instruct architecture.
  • Parameter Count: 3.1 billion parameters.
  • Context Length: Supports a context length of 32768 tokens.
  • Training Efficiency: Benefited from 2x faster training using Unsloth, indicating optimizations for resource-effective fine-tuning.
  • License: Released under the permissive Apache-2.0 license, allowing for broad use and distribution.

Good For

  • Instruction Following: Designed to excel at tasks requiring adherence to specific instructions.
  • Efficient Deployment: Its compact size and optimized training suggest suitability for applications where computational resources are a consideration.
  • Rapid Prototyping: The accelerated training process makes it a good candidate for quick experimentation and development cycles.