peulsilva / sentence-transformer-trained-tweet

SentenceTransformer based on whaleloops/phrase-bert

This is a sentence-transformers model finetuned from whaleloops/phrase-bert . It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: whaleloops/phrase-bert
• Maximum Sequence Length: 128 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity

Model Sources

• Documentation: Sentence Transformers Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 128, 'do_lower_case': None}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 768, '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})
)

Usage

Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("peulsilva/sentence-transformer-trained-tweet")
# Run inference
sentences = [
    'Honduras is technically still in the World Cup and Italy plus England are out means Honduras is better than them😂',
    'RT @iambolar: FT:Honduras 0-3 Switzerland. Shaqiri nets d 50th hat trick in #WorldCup history as Switzerland qualify 4d next round. http://…',
    'Honduras still better than the #CGHS JV Female Soccer Team 😂😂',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Evaluation

Metrics

Semantic Similarity

• Dataset: validation
• Evaluated with EmbeddingSimilarityEvaluator

Training Details

Training Dataset

Unnamed Dataset

• Size: 100,000 training samples
• Columns: sentence_0 , sentence_1 , and label
• Approximate statistics based on the first 1000 samples:

  	sentence_0	sentence_1	label
  type	string	string	float
  details	min: 6 tokens mean: 37.81 tokens max: 65 tokens	min: 6 tokens mean: 38.01 tokens max: 67 tokens	min: 0.0 mean: 0.56 max: 1.0

• Samples:

  sentence_0	sentence_1	label
  Early lead for #SUI over #HON thanks to Shaqiri taking a page out of Robben's book. He paid attention during Bayern practices. #ShaqAttaq ⚽️	RT @soccerdotcom: Los Catrachos! Follow @soccerdotcom and RT for the chance to win a Joma #HON Jersey signed by the team! http://t.co/2NTfw…	0.0
  RT @RTEsoccer: Group E result: #HON 0-3 #SUI. Shaqiri the hat-trick hero as the Swiss progress: http://t.co/fZYw9NFghO #rteworldcup http://…	RT @trueSCRlife: If #Shaqiri scores vs #HON we'll give away a pair of Magistas. Follow & RT to enter. Winner DMed! #HONvsSUI http://t.co/EG…	1.0
  RT @TheSCRLife: If #HON wins we’ll give away a pair of Superflys. FOLLOW & RETWEET. Not following?Won’t win. (I’m checking). http://t.co/xw…	Yup Honduras say goodbye lll	0.0

• Loss: CosineSimilarityLoss with these parameters:

  {
      "loss_fct": "torch.nn.modules.loss.MSELoss"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size : 128
• per_device_eval_batch_size : 128
• num_train_epochs : 1
• fp16 : True
• multi_dataset_batch_sampler : round_robin

All Hyperparameters

Training Logs

Framework Versions

• Python: 3.11.9
• Sentence Transformers: 3.3.0
• Transformers: 4.45.0.dev0
• PyTorch: 2.4.1+cu121
• Accelerate: 0.34.2
• Datasets: 2.20.0
• Tokenizers: 0.19.1

Citation

BibTeX

Sentence Transformers

@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084",
}