Hugging Face's logo

BAAI
/
bge-large-zh-noinstruct

Feature Extraction
Transformers
PyTorch
Chinese
bert
text-embeddings-inference
Inference Endpoints
Model card Files Community
4
bge-large-zh-noinstruct / README.md
Shitao's picture
Shitao
Update README.md
b29c29f
preview code
|
raw
history blame
15 kB
metadata
license: mit
language:
  - zh
pipeline_tag: sentence-similarity
tags:
  - sentence-transformers
  - Transfomers

FlagEmbedding

Model List | Usage | Evaluation | Train | License

For more details please refer to our GitHub repo: FlagEmbedding.

English | 中文

FlagEmbedding can map any text to a low-dimensional dense vector which can be used for tasks like retrieval, classification, clustering, or semantic search. And it also can be used in vector databases for LLMs.

************* 🌟Updates🌟 *************

  • 08/05/2023: Release base-scale and small-scale models, best performance among the models of the same size 🤗
  • 08/02/2023: Release bge-large-*(short for BAAI General Embedding) Models, rank 1st on MTEB and C-MTEB benchmark! :tada: :tada:
  • 08/01/2023: We release the Chinese Massive Text Embedding Benchmark (C-MTEB), consisting of 31 test dataset.

Model List

bge is short for BAAI general embedding.

Model Language Description query instruction for retrieval
BAAI/bge-large-en English :trophy: rank 1st in MTEB leaderboard Represent this sentence for searching relevant passages:
BAAI/bge-base-en English rank 2nd in MTEB leaderboard Represent this sentence for searching relevant passages:
BAAI/bge-small-en English a small-scale model but with competitive performance Represent this sentence for searching relevant passages:
BAAI/bge-large-zh Chinese :trophy: rank 1st in C-MTEB benchmark 为这个句子生成表示以用于检索相关文章：
BAAI/bge-large-zh-noinstruct Chinese This model is trained without instruction, and rank 2nd in C-MTEB benchmark
BAAI/bge-base-zh Chinese a base-scale model but has similar ability with bge-large-zh 为这个句子生成表示以用于检索相关文章：
BAAI/bge-small-zh Chinese a small-scale model but with competitive performance 为这个句子生成表示以用于检索相关文章：

Usage

  • Using FlagEmbedding
pip install -U FlagEmbedding

See FlagEmbedding for more methods to install FlagEmbedding.

from FlagEmbedding import FlagModel
sentences = ["样例数据-1", "样例数据-2"]
model = FlagModel('BAAI/bge-large-zh', query_instruction_for_retrieval="为这个句子生成表示以用于检索相关文章：")
embeddings = model.encode(sentences)
print(embeddings)

# for retrieval task, please use encode_queries() which will automatically add the instruction to each query
# corpus in retrieval task can still use encode() or encode_corpus()
queries = ['query_1', 'query_2']
passages = ["样例段落-1", "样例段落-2"]
q_embeddings = model.encode_queries(queries)
p_embeddings = model.encode(passages)
scores = q_embeddings @ p_embeddings.T

The value of argument query_instruction_for_retrieval see Model List.

FlagModel will use all available GPUs when encoding, please set os.environ["CUDA_VISIBLE_DEVICES"] to choose GPU.

  • Using Sentence-Transformers

Using this model also is easy when you have sentence-transformers installed:

pip install -U sentence-transformers

from sentence_transformers import SentenceTransformer
sentences = ["样例数据-1", "样例数据-2"]
model = SentenceTransformer('BAAI/bge-large-zh')
embeddings = model.encode(sentences, normalize_embeddings=True)
print(embeddings)

For retrieval task, each query should start with an instruction (instructions see Model List). 

from sentence_transformers import SentenceTransformer
queries = ["手机开不了机怎么办？"]
passages = ["样例段落-1", "样例段落-2"]
instruction = "为这个句子生成表示以用于检索相关文章："

model = SentenceTransformer('BAAI/bge-large-zh')
q_embeddings = model.encode([instruction+q for q in queries], normalize_embeddings=True)
p_embeddings = model.encode(passages, normalize_embeddings=True)
scores = q_embeddings @ p_embeddings.T
  • Using HuggingFace Transformers

With transformers package, you can use the model like this: First, you pass your input through the transformer model, then you select the last hidden state of first token (i.e., [CLS]) as the sentence embedding.

from transformers import AutoTokenizer, AutoModel
import torch
# Sentences we want sentence embeddings for
sentences = ["样例数据-1", "样例数据-2"]

# Load model from HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained('BAAI/bge-large-zh')
model = AutoModel.from_pretrained('BAAI/bge-large-zh')

# Tokenize sentences
encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
# for retrieval task, add an instruction to query
# encoded_input = tokenizer([instruction + q for q in queries], padding=True, truncation=True, return_tensors='pt')

# Compute token embeddings
with torch.no_grad():
    model_output = model(**encoded_input)
    # Perform pooling. In this case, cls pooling.
    sentence_embeddings = model_output[0][:, 0]
# normalize embeddings
sentence_embeddings = torch.nn.functional.normalize(sentence_embeddings, p=2, dim=1)
print("Sentence embeddings:", sentence_embeddings)

Evaluation

baai-general-embedding models achieve state-of-the-art performance on both MTEB and C-MTEB leaderboard! More details and evaluation tools see our scripts.

  • MTEB:
Model Name Dimension Sequence Length Average (56) Retrieval (15) Clustering (11) Pair Classification (3) Reranking (4) STS (10) Summarization (1) Classification (12)
bge-large-en 1024 512 63.98 53.9 46.98 85.8 59.48 81.56 32.06 76.21
bge-base-en 768 512 63.36 53.0 46.32 85.86 58.7 81.84 29.27 75.27
gte-large 1024 512 63.13 52.22 46.84 85.00 59.13 83.35 31.66 73.33
gte-base 768 512 62.39 51.14 46.2 84.57 58.61 82.3 31.17 73.01
e5-large-v2 1024 512 62.25 50.56 44.49 86.03 56.61 82.05 30.19 75.24
bge-small-en 384 512 62.11 51.82 44.31 83.78 57.97 80.72 30.53 74.37
instructor-xl 768 512 61.79 49.26 44.74 86.62 57.29 83.06 32.32 61.79
e5-base-v2 768 512 61.5 50.29 43.80 85.73 55.91 81.05 30.28 73.84
gte-small 384 512 61.36 49.46 44.89 83.54 57.7 82.07 30.42 72.31
text-embedding-ada-002 1536 8192 60.99 49.25 45.9 84.89 56.32 80.97 30.8 70.93
e5-small-v2 384 512 59.93 49.04 39.92 84.67 54.32 80.39 31.16 72.94
sentence-t5-xxl 768 512 59.51 42.24 43.72 85.06 56.42 82.63 30.08 73.42
all-mpnet-base-v2 768 514 57.78 43.81 43.69 83.04 59.36 80.28 27.49 65.07
sgpt-bloom-7b1-msmarco 4096 2048 57.59 48.22 38.93 81.9 55.65 77.74 33.6 66.19
all-MiniLM-L12-v2 384 512 56.53 42.69 41.81 82.41 58.44 79.8 27.9 63.21
all-MiniLM-L6-v2 384 512 56.26 41.95 42.35 82.37 58.04 78.9 30.81 63.05
contriever-base-msmarco 768 512 56.00 41.88 41.1 82.54 53.14 76.51 30.36 66.68
sentence-t5-base 768 512 55.27 33.63 40.21 85.18 53.09 81.14 31.39 69.81
  • C-MTEB:
    We create a benchmark C-MTEB for Chinese text embedding which consists of 31 datasets from 6 tasks. Please refer to C_MTEB for a detailed introduction.
Model Embedding dimension Avg Retrieval STS PairClassification Classification Reranking Clustering
bge-large-zh 1024 64.20 71.53 53.23 78.94 72.26 65.11 48.39
bge-large-zh-noinstruct 1024 63.53 70.55 50.98 76.77 72.49 64.91 50.01
BAAI/bge-base-zh 768 62.96 69.53 52.05 77.5 70.98 64.91 47.63
BAAI/bge-small-zh 512 58.27 63.07 46.87 70.35 67.78 61.48 45.09
m3e-base 768 57.10 56.91 48.15 63.99 70.28 59.34 47.68
m3e-large 1024 57.05 54.75 48.64 64.3 71.22 59.66 48.88
text-embedding-ada-002(OpenAI) 1536 53.02 52.0 40.61 69.56 67.38 54.28 45.68
luotuo 1024 49.37 44.4 39.41 66.62 65.29 49.25 44.39
text2vec 768 47.63 38.79 41.71 67.41 65.18 49.45 37.66
text2vec-large 1024 47.36 41.94 41.98 70.86 63.42 49.16 30.02

Train

This section will introduce the way we used to train the general embedding. The training scripts are in FlagEmbedding, and we provide some examples to do pre-train and fine-tune.

1. RetroMAE Pre-train
We pre-train the model following the method retromae, which shows promising improvement in retrieval task (paper). The pre-training was conducted on 24 A100(40G) GPUs with a batch size of 720. In retromae, the mask ratio of encoder and decoder are 0.3, and 0.5 respectively. We used the AdamW optimizer and the learning rate is 2e-5.

Pre-training data:

2. Finetune
We fine-tune the model using a contrastive objective. The format of input data is a triple(query, positive, negative). Besides the negative in the triple, we also adopt in-batch negatives strategy. We employ the cross-device negatives sharing method to share negatives among different GPUs, which can dramatically increase the number of negatives.

We trained our model on 48 A100(40G) GPUs with a large batch size of 32,768 (so there are 65,535 negatives for each query in a batch). We used the AdamW optimizer and the learning rate is 1e-5. The temperature for contrastive loss is 0.01.

For the version with *-instrcution, we add instruction to the query for retrieval task in the training. For english, the instruction is Represent this sentence for searching relevant passages: ; For chinese, the instruction is 为这个句子生成表示以用于检索相关文章：. In the evaluation, the instruction should be added for sentence to passages retrieval task, not be added for other tasks.

The finetune script is accessible in this repository: FlagEmbedding. You can easily finetune your model with it.

Training data:

  • For English, we collect 230M text pairs from wikipedia, cc-net, and so on.

  • For chinese, we collect 120M text pairs from wudao, simclue and so on.

The data collection is to be released in the future.

We will continually update the embedding models and training codes, hoping to promote the development of the embedding model community.

License

FlagEmbedding is licensed under MIT License. The released models can be used for commercial purposes free of charge.