库名称：transformers
许可证：apache-2.0
基础模型：

• rhymes-ai/Aria-sequential_mlp
• rhymes-ai/Aria
  任务标签：图像文本到文本

Aria-sequential_mlp-bnb_nf4

这是基于Aria-sequential_mlp的BitsAndBytes NF4量化版本，需要约15.5 GB显存，可在RTX 3090上运行（RTX 4060 Ti 16 GB也能运行，但未启用device_map=auto时不实用）。
当前模型未分割为5 GB分片，因为加载序列化的BNB模型时可能引发问题，这可能导致免费版Colab无法加载模型。

安装

pip install transformers==4.45.0 accelerate==0.34.1 sentencepiece==0.2.0 torchvision requests torch Pillow bitsandbytes  
pip install flash-attn --no-build-isolation  

推理

运行此模型的代码：

import requests  
import torch  
from PIL import Image  
from transformers import AutoModelForCausalLM, AutoProcessor, BitsAndBytesConfig  
torch.cuda.set_device(0)  

model_id_or_path = "leon-se/Aria-sequential_mlp-bnb_nf4"  

model = AutoModelForCausalLM.from_pretrained(model_id_or_path, device_map="auto", torch_dtype=torch.bfloat16, trust_remote_code=True)  
processor = AutoProcessor.from_pretrained(model_id_or_path, trust_remote_code=True)  

image_path = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cat.png"  

image = Image.open(requests.get(image_path, stream=True).raw)  

messages = [  
    {  
        "role": "user",  
        "content": [  
            {"text": None, "type": "image"},  
            {"text": "图片里有什么？", "type": "text"},  
        ],  
    }  
]  

text = processor.apply_chat_template(messages, add_generation_prompt=True)  
inputs = processor(text=text, images=image, return_tensors="pt")  
inputs["pixel_values"] = inputs["pixel_values"].to(model.dtype)  
inputs = {k: v.to(model.device) for k, v in inputs.items()}  

with torch.inference_mode(), torch.amp.autocast("cuda", dtype=torch.bfloat16):  
    output = model.generate(  
        **inputs,  
        max_new_tokens=500,  
        stop_strings=["<|im_end|>"],  
        tokenizer=processor.tokenizer,  
        do_sample=True,  
        temperature=0.9,  
    )  
    output_ids = output[0][inputs["input_ids"].shape[1]:]  
    result = processor.decode(output_ids, skip_special_tokens=True)  

print(result)  
print(f'最大分配显存: {torch.cuda.max_memory_allocated(device="cuda") / 1024 ** 3:.3f}GiB')  

量化方法

量化创建代码：

from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig  

model_id = "rhymes-ai/Aria-sequential_mlp"  

nf4_config = BitsAndBytesConfig(  
    load_in_4bit=True,  
    bnb_4bit_quant_type="nf4",  
    bnb_4bit_compute_dtype=torch.bfloat16,  
    bnb_4bit_use_double_quant=True,  
    llm_int8_enable_fp32_cpu_offload=True,  
    llm_int8_skip_modules=["language_model.lm_head", "multi_modal_projector", "vision_tower"],  
    )  

model_nf4 = AutoModelForCausalLM.from_pretrained(model_id, quantization_config=nf4_config)