E5-V: Universal Embeddings with Multimodal Large Language Models

Overview

We propose a framework, called E5-V, to adpat MLLMs for achieving multimodal embeddings. E5-V effectively bridges the modality gap between different types of inputs, demonstrating strong performance in multimodal embeddings even without fine-tuning. We also propose a single modality training approach for E5-V, where the model is trained exclusively on text pairs, demonstrating better performance than multimodal training.

Example

import torch
import torch.nn.functional as F
import requests
from PIL import Image
from transformers import AutoTokenizer
from transformers import LlavaNextProcessor, LlavaNextForConditionalGeneration

llama3_template = '<|start_header_id|>user<|end_header_id|>\n\n{}<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n \n'

processor = LlavaNextProcessor.from_pretrained('royokong/e5-v')
model = LlavaNextForConditionalGeneration.from_pretrained('royokong/e5-v', torch_dtype=torch.float16).cuda()

img_prompt = llama3_template.format('<image>\nSummary above image in one word: ')
text_prompt = llama3_template.format('<sent>\nSummary above sentence in one word: ')

urls = ['https://upload.wikimedia.org/wikipedia/commons/thumb/4/47/American_Eskimo_Dog.jpg/360px-American_Eskimo_Dog.jpg',
        'https://upload.wikimedia.org/wikipedia/commons/thumb/b/b6/Felis_catus-cat_on_snow.jpg/179px-Felis_catus-cat_on_snow.jpg']
images = [Image.open(requests.get(url, stream=True).raw) for url in urls]

texts = ['A dog sitting in the grass.',
         'A cat standing in the snow.']

text_inputs = processor([text_prompt.replace('<sent>', text) for text in texts], return_tensors="pt", padding=True).to('cuda')
img_inputs = processor([img_prompt]*len(images), images, return_tensors="pt", padding=True).to('cuda')

with torch.no_grad():
    text_embs = model(**text_inputs, output_hidden_states=True, return_dict=True).hidden_states[-1][:, -1, :]
    img_embs = model(**img_inputs, output_hidden_states=True, return_dict=True).hidden_states[-1][:, -1, :]

    text_embs = F.normalize(text_embs, dim=-1)
    img_embs = F.normalize(img_embs, dim=-1)

print(text_embs @ img_embs.t())

Evaulate

To evaluate the original results in the paper, please run following

# eval on coco, flickr30k, fashioniq and cirr
accelerate launch --num_machines=1 --num_processes 8 --machine_rank 0 retrieval.py  --use_e5v 

# eval on i2i-coco, i2i-flickr30k
accelerate launch --num_machines=1 --num_processes 8 --machine_rank 0 retrieval.py  --use_e5v  --ocr_replace_text

# eval on sts tasks
cd SentEval/data/downstream/
bash download_dataset.sh
cd -
accelerate launch --num_machines=1 --num_processes 8 --machine_rank 0 eval_sts.py --model_name_or_path royokong/e5-v

Training

1. Install Dependencies

pip install -r requirements.txt

2. Download Data

cd ./data
bash download_nli.sh
cd -

3. Transfer llava-llama-3-8b model to huggingface format on each nodes

mkdir -p models
cd models
for i in 1 2 3 4; do
    wget https://huggingface.co/lmms-lab/llama3-llava-next-8b/resolve/main/model-0000$i-of-00004.safetensors
done
cd -
python load_llama3_hf.py
rm models/*.safetensors

4. Train

bash run.sh

5. Test Use --lora_path flag to test the results.

accelerate launch --num_machines=1 --num_processes 8 --machine_rank 0 retrieval.py \
    --llava_llama3 --lora_path e5v-8b  --batch_size 1

Acknowledgement

Our Code is based on SimCSE and alpaca-lora