Yes, you can fine-tune GLM-4.5V (vision) models with Unsloth, as long as the model is supported by Hugging Face Transformers. Unsloth supports vision and text models out of the box, and GLM-4.5 is explicitly mentioned as supported for both running and fine-tuning, including vision tasks. For best results, ensure you use the latest Unsloth and Transformers versions and follow the vision fine-tuning guides and notebooks provided by Unsloth.
If you need a specific example or notebook for GLM-4.5V vision fine-tuning, Unsloth is working on providing dedicated resources, but the general vision fine-tuning workflow applies. You may need to adapt an existing vision notebook (e.g., for Qwen or Llama) by changing the model name to your GLM-4.5V variant and ensuring your dataset is formatted for vision tasks. See the documentation for details on dataset formatting and training steps. Would you like a step-by-step example or code snippet for this process?
Sources:
hasty prawn
coral tusk