from transformers import AutoModelForCausalLM, AutoTokenizer
model_path = "ByteDance/Sa2VA-8B"
model = AutoModelForCausalLM.from_pretrained(
model_path, torch_dtype="auto", device_map="auto", trust_remote_code=True
).eval()
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)