[model] feat: support Qwen-Image

[yuyu5333]

Add model Qwen-Image

• support Qwen-Image

Key Changes

• add Qwen Image Model Train...Coding
• Qwen Image has capabilities such as text-to-image generation and image editing.

TODO

I will break down the development process into multiple parts such as SFT, LORA, func_text(text), func_image(image, func_text()), qwen image( func_image(image, func_text(text))), etc. I will develop the corresponding modules one by one, and then integrate and verify them.

• Train SFT pipeline
• Train LoRa ability
• Part1: func_text(text)
• Part2: func_image(image, func_text())
• Part3: qwen image( func_image(image, func_text(text)))
• Verify the training effect
• Clean up useless commits and keep the branch tidy

[yuyu5333]

After the Wan2.2 model is merged, I will perform a rebase to keep the commit clean.

[QAQdev]

Hi, it is great to support qwen image in veomni. May I ask did you finish the support of qwen image?