---
base_model: BEE-spoke-data/bert-plus-L8-v1.0-allNLI_matryoshka
library_name: sentence-transformers
pipeline_tag: sentence-similarity
tags:
- sentence-transformers
- feature-extraction
- sentence-similarity
- transformers
license: apache-2.0
widget:
- source_sentence: How to discreetly optimize operating expenses?
  sentences:
  - Strategies for quietly reducing overhead costs
  - Subtle ways to cut down on operating expenses
  - >-
    Implementing technology to save on operating costs without broad
    announcements
  - Lowering daily business expenses through unnoticed efficiencies
  - >-
    Minimizing operational expenditures in small businesses without drawing
    attention
datasets:
- pszemraj/synthetic-text-similarity
language:
- en
---
# BEE-spoke-data/bert-plus-L8-v1.0-synthSTSv3-4k


This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search.

- This model has been further trained from [BEE-spoke-data/bert-plus-L8-v1.0-allNLI_matryoshka](https://hf.co/BEE-spoke-data/bert-plus-L8-v1.0-allNLI_matryoshka) on `v3.0` of the `synthetic text similarity` dataset.
- Intended for use in comparing the cosine similarity of longer document embeddings and/or clustering them.
- Matryoshka dims: [768, 512, 256, 128, 64]

An earlier version of this model (on `v1.0` of the dataset) can be found [here](https://huggingface.co/BEE-spoke-data/bert-plus-L8-v1.0-syntheticSTS-4k). TBD which performs better in practical tasks.

## Usage (Sentence-Transformers)

Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:

```
pip install -U sentence-transformers
```

Then you can use the model like this:

```python
from sentence_transformers import SentenceTransformer
sentences = ["This is an example sentence", "Each sentence is converted"]

model = SentenceTransformer('BEE-spoke-data/bert-plus-L8-v1.0-synthSTSv3-4k')
embeddings = model.encode(sentences)
print(embeddings)
```



## Usage (HuggingFace Transformers)

Without [sentence-transformers](https://www.SBERT.net), 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.

```python
import torch
from transformers import AutoModel, AutoTokenizer


# 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(
    "BEE-spoke-data/bert-plus-L8-v1.0-synthSTSv3-4k"
)
model = AutoModel.from_pretrained("BEE-spoke-data/bert-plus-L8-v1.0-synthSTSv3-4k")

# 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)
```

## Training

See training details below.

**Loss**:

`sentence_transformers.losses.MatryoshkaLoss.MatryoshkaLoss` with parameters:
  ```
  {'loss': 'CosineSimilarityLoss', 'matryoshka_dims': [768, 512, 256, 128, 64], 'matryoshka_weights': [1, 1, 1, 1, 1], 'n_dims_per_step': -1}
  ```