metadata
library_name: sentence-transformers
pipeline_tag: sentence-similarity
tags:
- sentence-transformers
- feature-extraction
- sentence-similarity
- transformers
- mteb
model-index:
- name: mxbai-embed-large-v1-sts-matryoshka
results:
- task:
type: STS
dataset:
type: mteb/biosses-sts
name: MTEB BIOSSES
config: default
split: test
revision: d3fb88f8f02e40887cd149695127462bbcf29b4a
metrics:
- type: cos_sim_pearson
value: 89.74819706519132
- type: cos_sim_spearman
value: 89.16227258672195
- type: euclidean_pearson
value: 87.4894915970516
- type: euclidean_spearman
value: 88.82371065549121
- type: manhattan_pearson
value: 87.4550672002364
- type: manhattan_spearman
value: 88.74479319465274
- task:
type: STS
dataset:
type: mteb/sickr-sts
name: MTEB SICK-R
config: default
split: test
revision: a6ea5a8cab320b040a23452cc28066d9beae2cee
metrics:
- type: cos_sim_pearson
value: 87.6165367318343
- type: cos_sim_spearman
value: 82.96648281765052
- type: euclidean_pearson
value: 85.15841275786363
- type: euclidean_spearman
value: 82.59932858741094
- type: manhattan_pearson
value: 85.13304536590803
- type: manhattan_spearman
value: 82.57423596486754
- task:
type: STS
dataset:
type: mteb/sts12-sts
name: MTEB STS12
config: default
split: test
revision: a0d554a64d88156834ff5ae9920b964011b16384
metrics:
- type: cos_sim_pearson
value: 90.28534830444329
- type: cos_sim_spearman
value: 82.55267827529079
- type: euclidean_pearson
value: 88.25378069753911
- type: euclidean_spearman
value: 83.11525616002255
- type: manhattan_pearson
value: 88.22692740852892
- type: manhattan_spearman
value: 83.09760476508313
- task:
type: STS
dataset:
type: mteb/sts13-sts
name: MTEB STS13
config: default
split: test
revision: 7e90230a92c190f1bf69ae9002b8cea547a64cca
metrics:
- type: cos_sim_pearson
value: 92.4873773117415
- type: cos_sim_spearman
value: 92.45311856572782
- type: euclidean_pearson
value: 91.3940762182452
- type: euclidean_spearman
value: 91.99304193285555
- type: manhattan_pearson
value: 91.36013458363827
- type: manhattan_spearman
value: 91.94059360016104
- task:
type: STS
dataset:
type: mteb/sts14-sts
name: MTEB STS14
config: default
split: test
revision: 6031580fec1f6af667f0bd2da0a551cf4f0b2375
metrics:
- type: cos_sim_pearson
value: 92.86047241342555
- type: cos_sim_spearman
value: 91.7373997118018
- type: euclidean_pearson
value: 91.85151887796326
- type: euclidean_spearman
value: 91.84470114529888
- type: manhattan_pearson
value: 91.84546446666218
- type: manhattan_spearman
value: 91.83943192650281
- task:
type: STS
dataset:
type: mteb/sts15-sts
name: MTEB STS15
config: default
split: test
revision: ae752c7c21bf194d8b67fd573edf7ae58183cbe3
metrics:
- type: cos_sim_pearson
value: 91.54334603049196
- type: cos_sim_spearman
value: 92.53856973601617
- type: euclidean_pearson
value: 91.78019629944465
- type: euclidean_spearman
value: 92.34058447365024
- type: manhattan_pearson
value: 91.77889261840363
- type: manhattan_spearman
value: 92.34415590718599
- task:
type: STS
dataset:
type: mteb/sts16-sts
name: MTEB STS16
config: default
split: test
revision: 4d8694f8f0e0100860b497b999b3dbed754a0513
metrics:
- type: cos_sim_pearson
value: 86.44182676813374
- type: cos_sim_spearman
value: 88.13762635519763
- type: euclidean_pearson
value: 87.368150115143
- type: euclidean_spearman
value: 87.80086022782062
- type: manhattan_pearson
value: 87.36033717796437
- type: manhattan_spearman
value: 87.79278886160228
- task:
type: STS
dataset:
type: mteb/sts17-crosslingual-sts
name: MTEB STS17 (en-en)
config: en-en
split: test
revision: af5e6fb845001ecf41f4c1e033ce921939a2a68d
metrics:
- type: cos_sim_pearson
value: 90.45736504439745
- type: cos_sim_spearman
value: 90.36467157990322
- type: euclidean_pearson
value: 90.19668998748772
- type: euclidean_spearman
value: 89.89620750310942
- type: manhattan_pearson
value: 90.28020856755623
- type: manhattan_spearman
value: 90.06372752006743
- task:
type: STS
dataset:
type: mteb/sts22-crosslingual-sts
name: MTEB STS22 (en)
config: en
split: test
revision: eea2b4fe26a775864c896887d910b76a8098ad3f
metrics:
- type: cos_sim_pearson
value: 68.08713149697203
- type: cos_sim_spearman
value: 68.46457718572601
- type: euclidean_pearson
value: 69.25165385150007
- type: euclidean_spearman
value: 68.1342295049583
- type: manhattan_pearson
value: 69.30447292480429
- type: manhattan_spearman
value: 68.25582089078411
- task:
type: STS
dataset:
type: mteb/stsbenchmark-sts
name: MTEB STSBenchmark
config: default
split: test
revision: b0fddb56ed78048fa8b90373c8a3cfc37b684831
metrics:
- type: cos_sim_pearson
value: 88.34259569356074
- type: cos_sim_spearman
value: 89.78243695761168
- type: euclidean_pearson
value: 89.12060338382234
- type: euclidean_spearman
value: 89.67729976099504
- type: manhattan_pearson
value: 89.12528974491349
- type: manhattan_spearman
value: 89.66532321334671
MichaelKarpe/mxbai-embed-large-v1-sts-matryoshka
This is a sentence-transformers model: It maps sentences & paragraphs to a 1024 dimensional dense vector space and can be used for tasks like clustering or semantic search.
Usage (Sentence-Transformers)
Using this model becomes easy when you have sentence-transformers installed:
pip install -U sentence-transformers
Then you can use the model like this:
from sentence_transformers import SentenceTransformer
sentences = ["This is an example sentence", "Each sentence is converted"]
model = SentenceTransformer('MichaelKarpe/mxbai-embed-large-v1-sts-matryoshka')
embeddings = model.encode(sentences)
print(embeddings)
Usage (HuggingFace Transformers)
Without sentence-transformers, you can use the model like this: First, you pass your input through the transformer model, then you have to apply the right pooling-operation on-top of the contextualized word embeddings.
from transformers import AutoTokenizer, AutoModel
import torch
#Mean Pooling - Take attention mask into account for correct averaging
def mean_pooling(model_output, attention_mask):
token_embeddings = model_output[0] #First element of model_output contains all token embeddings
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 we want sentence embeddings for
sentences = ['This is an example sentence', 'Each sentence is converted']
# Load model from HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained('MichaelKarpe/mxbai-embed-large-v1-sts-matryoshka')
model = AutoModel.from_pretrained('MichaelKarpe/mxbai-embed-large-v1-sts-matryoshka')
# Tokenize sentences
encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
# Compute token embeddings
with torch.no_grad():
model_output = model(**encoded_input)
# Perform pooling. In this case, mean pooling.
sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
print("Sentence embeddings:")
print(sentence_embeddings)
Evaluation Results
For an automated evaluation of this model, see the Sentence Embeddings Benchmark: https://seb.sbert.net
Training
The model was trained with the parameters:
DataLoader:
torch.utils.data.dataloader.DataLoader of length 360 with parameters:
{'batch_size': 16, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}
Loss:
sentence_transformers.losses.MatryoshkaLoss.MatryoshkaLoss with parameters:
{'loss': 'CoSENTLoss', 'matryoshka_dims': [1024, 768, 512, 256, 128, 64], 'matryoshka_weights': [1, 1, 1, 1, 1, 1], 'n_dims_per_step': -1}
Parameters of the fit()-Method:
{
"epochs": 4,
"evaluation_steps": 1000,
"evaluator": "sentence_transformers.evaluation.EmbeddingSimilarityEvaluator.EmbeddingSimilarityEvaluator",
"max_grad_norm": 1,
"optimizer_class": "<class 'torch.optim.adamw.AdamW'>",
"optimizer_params": {
"lr": 2e-05
},
"scheduler": "WarmupLinear",
"steps_per_epoch": null,
"warmup_steps": 144,
"weight_decay": 0.01
}
Full Model Architecture
SentenceTransformer(
(0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel
(1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
)