wnma3mz

/

Janus-Pro-1B-LM

like 1

This is only part of Janus-Pro-1B's language model and does not include image understanding and generation.

Janus-Pro-1B's language model is llama, so any code that can run llama can be run.

I developed a project can be run directly, https://github.com/wnma3mz/tLLM.

tllm.server --model_path wnma3mz/Janus-Pro-1B-LM --hostname localhost --is_local --client_size 1

Image understanding and generation are still working on adaptation, and will be updated synchronously later.

Thanks for your attention!

Llama is not free software, so they can't make free software from Llama origin, are you sure?

I will try your method

Llama is not free software, so they can't make free software from Llama origin, are you sure?

Should this be confirmed by the Janus-Pro team? They used the llama language model architecture in Janus-Pro.

it just separates the language model from Janus-Pro.

tllm now support input image, and the converted model now in

• https://huggingface.co./wnma3mz/Janus-Pro-1B-4bit

• https://huggingface.co./wnma3mz/Janus-Pro-1B

• https://huggingface.co./wnma3mz/Janus-Pro-7B-4bit

• https://huggingface.co./wnma3mz/Janus-Pro-7B

Llama is not free software, so they can't make free software from Llama origin, are you sure?

Should this be confirmed by the Janus-Pro team? They used the llama language model architecture in Janus-Pro.

it just separates the language model from Janus-Pro.

Architecture is not same as data or code. So I do not know what they used exactly. I am user and I like to have clarity on that.