nomic-embed-text-v1-ablated: A Reproducible Long Context (8192) Text Embedder

nomic-embed-text-v1-ablated is 8192 context length text encoder that surpasses OpenAI text-embedding-ada-002 performance on short and long tasks. .

Name	SeqLen	MTEB	LoCo	Jina Long Context	Open Weights	Open Training Code	Open Data
nomic-embed-text-v1	8192	62.39	85.53	54.16	✅	✅	✅
jina-embeddings-v2-base-en	8192	60.39	85.45	51.90	✅	❌	❌
text-embedding-3-small	8191	62.26	82.40	58.20	❌	❌	❌
text-embedding-ada-002	8191	60.99	52.7	55.25	❌	❌	❌

If you would like to finetune a model on more data, you can use this model as an initialization

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

output = embed.text(
    texts=['Nomic Embedding API', '#keepAIOpen'],
    model='nomic-embed-text-v1',
    task_type='search_document'
)

print(output)

For more information, see the API reference

Data Visualization

Click the Nomic Atlas map below to visualize a 5M sample of our contrastive pretraining data!

Training Details

We train our embedder using a multi-stage training pipeline. Starting from a long-context BERT model, the first unsupervised contrastive stage trains on a dataset generated from weakly related text pairs, such as question-answer pairs from forums like StackExchange and Quora, title-body pairs from Amazon reviews, and summarizations from news articles.

In the second finetuning stage, higher quality labeled datasets such as search queries and answers from web searches are leveraged. Data curation and hard-example mining is crucial in this stage.

For more details, see the Nomic Embed Technical Report and corresponding blog post.

Training data to train the models is released in its entirety. For more details, see the contrastors repository

Usage

Note nomic-embed-text requires prefixes! We support the prefixes [search_query, search_document, classification, clustering]. For retrieval applications, you should prepend search_document for all your documents and search_query for your queries.

Sentence Transformers

from sentence_transformers import SentenceTransformer

model = SentenceTransformer("nomic-ai/nomic-embed-text-v1-ablated", trust_remote_code=True)
sentences = ['search_query: What is TSNE?', 'search_query Who is Laurens van der Maaten?']
embeddings = model.encode(sentences)
print(embeddings)

Transformers

import torch
import torch.nn.functional as F
from transformers import AutoTokenizer, AutoModel

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 is TSNE?', 'search_query: Who is Laurens van der Maaten?']

tokenizer = AutoTokenizer.from_pretrained('bert-base-uncased')
model = AutoModel.from_pretrained('nomic-ai/nomic-embed-text-v1-ablated', trust_remote_code=True)
model.eval()

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

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

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

The model natively supports scaling of the sequence length past 2048 tokens. To do so,

- tokenizer = AutoTokenizer.from_pretrained('bert-base-uncased')
+ tokenizer = AutoTokenizer.from_pretrained('bert-base-uncased', model_max_length=8192)


- model = AutoModel.from_pretrained('nomic-ai/nomic-embed-text-v1-ablated', trust_remote_code=True)
+ model = AutoModel.from_pretrained('nomic-ai/nomic-embed-text-v1-ablated', trust_remote_code=True, rotary_scaling_factor=2)

Join the Nomic Community

• Nomic: https://nomic.ai
• Discord: https://discord.gg/myY5YDR8z8
• Twitter: https://twitter.com/nomic_ai