Orthrus diffusion head trained Qwen 3.5/3.6 and Gemma 4 models dropping soon

The Orthrus project is preparing to release support for Qwen 3.5, Qwen 3.6, and Gemma 4 models using a diffusion head approach. The team has finalized testing and is currently setting up the release pipeline.

• Support will be added for Qwen3.5, Qwen3.6, and Gemma4 models.
• Complete end-to-end training and evaluation code will be open-sourced alongside the model checkpoints.
• Updates are being pushed to the repository shortly.

The release aims to provide accessible tools for training and evaluating these specific model architectures.