openELM pre-train weights

Hi I'm working on small LLM due to GPU resource limitation, and I really like xtuner, and use it as my basic experiment tool.
I tried your Intern1.8B, but it still a little large for my task.

I wonder could you kindly provide some tiny LLM result, esp. apple's ELM (e.g., 270M version): https://huggingface.co/apple/OpenELM-270M-Instruct. I know it will bring you some heavy work, but I think quite a few reseachers like me requires it.

thank you!