Aria-torchao-int8wo / README.md
aria-dev's picture
Update README.md
feb88b1 verified
|
raw
history blame
2.05 kB
metadata
license: apache-2.0
base_model:
  - rhymes-ai/Aria

This repository offers int8 quantized weights of the rhymes-ai/Aria model utilizing the TorchAO quantization framework. It now supports inference within 30GB of GPU memory.

Quick Start

Installation

pip install transformers==4.45.0 accelerate==0.34.1 sentencepiece==0.2.0  torch==2.5.0 torchao==0.6.1 torchvision requests Pillow
pip install flash-attn --no-build-isolation

Inference

import requests
import torch
from PIL import Image
from transformers import AutoModelForCausalLM, AutoProcessor

model_id_or_path = "rhymes-ai/Aria-torchao-int8wo"

model = AutoModelForCausalLM.from_pretrained(
    model_id_or_path,
    device_map="auto",
    torch_dtype=torch.bfloat16,
    trust_remote_code=True,
    attn_implementation="flash_attention_2",
)

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": "what is the image?", "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.cuda.amp.autocast(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)