Is the shared model: charlesdj/CSR_LLaVA_1.5_7b_1Iteration -> lora or merged?

[sandeep-sm]

When I run eval using the --model_path to this huggingface model, it shows many extra params, although it doesn't give an error that it would give for a lora model passed without the --model_base. When I explicitly provide --model_base, it runs normally without any issues.

Hence I am confused if the shared model is lora merged with the base model or just lora weights?