nomic-embed-vision-v1: Expanding the Latent Space

nomic-embed-vision-v1 is a high performing vision embedding model that shares the same embedding space as nomic-embed-text-v1.

All Nomic Embed Text models are now multimodal!

Name	Imagenet 0-shot	Datacomp (Avg. 38)	MTEB
`nomic-embed-vision-v1.5`	71.0	56.8	62.28
`nomic-embed-vision-v1`	70.7	56.7	62.39
OpenAI CLIP ViT B/16	68.3	56.3	43.82
Jina CLIP v1	59.1	52.2	60.1

Hosted Inference API

The easiest way to get started with Nomic Embed is through the Nomic Embedding API.

Generating embeddings with the nomic Python client is as easy as

from nomic import embed
import numpy as np

output = embed.image(
    images=[
        "image_path_1.jpeg",
        "image_path_2.png",
    ],
    model='nomic-embed-vision-v1',
)

print(output['usage'])
embeddings = np.array(output['embeddings'])
print(embeddings.shape)

For more information, see the API reference

Data Visualization

Click the Nomic Atlas map below to visualize a 100,000 sample CC3M comparing the Vision and Text Embedding Space!

Training Details

We align our vision embedder to the text embedding by employing a technique similar to LiT but instead lock the text embedder!

For more details, see the Nomic Embed Vision Technical Report (soon to be released!) and corresponding blog post

Training code is released in the contrastors repository

Usage

Remember nomic-embed-text requires prefixes and so, when using Nomic Embed in multimodal RAG scenarios (e.g. text to image retrieval), you should use the search_query: prefix.

Transformers

import torch
import torch.nn.functional as F
from transformers import AutoTokenizer, AutoModel, AutoImageProcessor
from PIL import Image
import requests

processor = AutoImageProcessor.from_pretrained("nomic-ai/nomic-embed-vision-v1")
vision_model = AutoModel.from_pretrained("nomic-ai/nomic-embed-vision-v1", trust_remote_code=True)

url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
image = Image.open(requests.get(url, stream=True).raw)

inputs = processor(image, return_tensors="pt")

img_emb = vision_model(**inputs).last_hidden_state
img_embeddings = F.normalize(img_emb[:, 0], p=2, dim=1)

Additionally, you can perform multimodal retrieval!


def mean_pooling(model_output, attention_mask):
    token_embeddings = model_output[0]
    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)

sentences = ['search_query: What are cute animals to cuddle with?', 'search_query: What do cats look like?']

tokenizer = AutoTokenizer.from_pretrained('nomic-ai/nomic-embed-text-v1')
text_model = AutoModel.from_pretrained('nomic-ai/nomic-embed-text-v1', trust_remote_code=True)
text_model.eval()

encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')

with torch.no_grad():
    model_output = text_model(**encoded_input)

text_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
text_embeddings = F.normalize(text_embeddings, p=2, dim=1)

print(torch.matmul(img_embeddings, text_embeddings.T))