SentenceTransformer based on ltg/norbert4-base

This is a sentence-transformers model finetuned from ltg/norbert4-base on the all-nli-norwegian dataset. It maps sentences & paragraphs to a 640-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: ltg/norbert4-base
• Maximum Sequence Length: 128 tokens
• Output Dimensionality: 640 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • all-nli-norwegian
• Language: no

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': False, 'architecture': 'GptBertModel'})
 (1): Pooling({'word_embedding_dimension': 640, '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("thivy/norbert4-base-nli-norwegian")
# Run inference
sentences = [
 'En mann lager et sandmaleri på gulvet.',
 'En mann lager kunst.',
 'En kvinne ødelegger et sandmaleri.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 640]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.6251, 0.2931],
# [0.6251, 1.0000, 0.1305],
# [0.2931, 0.1305, 1.0000]])

Evaluation

Metrics

Triplet

• Dataset: eval
• Evaluated with TripletEvaluator

Training Details

Training Dataset

all-nli-norwegian

• Dataset: all-nli-norwegian at 98cabde
• Size: 556,367 training samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 6 tokens mean: 9.53 tokens max: 47 tokens	min: 5 tokens mean: 12.03 tokens max: 40 tokens	min: 5 tokens mean: 12.7 tokens max: 49 tokens

• Samples:

  anchor	positive	negative
  En person på en hest hopper over et havarert fly.	En person er utendørs, på en hest.	En person er på en diner og bestiller en omelett.
  Barn smiler og vinker til kameraet	Det er barn til stede	Barna rynker pannen
  En gutt hopper på skateboard midt på en rød bro.	Gutten gjør et skateboardtriks.	Gutten skater nedover fortauet.

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
   "scale": 20.0,
   "similarity_fct": "cos_sim",
   "gather_across_devices": false
  }
  

Evaluation Dataset

all-nli-norwegian

• Dataset: all-nli-norwegian at 98cabde
• Size: 6,561 evaluation samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 5 tokens mean: 17.72 tokens max: 74 tokens	min: 4 tokens mean: 8.98 tokens max: 31 tokens	min: 3 tokens mean: 9.5 tokens max: 29 tokens

• Samples:

  anchor	positive	negative
  To kvinner klemmer mens de holder take-away pakker.	To kvinner holder pakker.	Mennene slåss utenfor en deli.
  To små barn i blå drakter, en med nummer 9 og en med nummer 2, står på trinn i et bad og vasker hendene i en vask.	To barn i nummererte drakter vasker hendene.	To barn i jakker går til skolen.
  En mann selger donuts til en kunde under et verdensutstillingsarrangement holdt i byen Angeles	En mann selger donuts til en kunde.	En kvinne drikker kaffen sin på en liten kafé.

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
   "scale": 20.0,
   "similarity_fct": "cos_sim",
   "gather_across_devices": false
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 64
• learning_rate: 2e-05
• weight_decay: 0.01
• num_train_epochs: 1
• warmup_ratio: 0.1
• bf16: True
• load_best_model_at_end: True

All Hyperparameters

Training Logs

Framework Versions

• Python: 3.12.12
• Sentence Transformers: 5.2.0
• Transformers: 4.57.3
• PyTorch: 2.9.1
• Accelerate: 1.12.0
• Datasets: 4.4.2
• Tokenizers: 0.22.1

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",
}

MultipleNegativesRankingLoss

@misc{henderson2017efficient,
 title={Efficient Natural Language Response Suggestion for Smart Reply},
 author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
 year={2017},
 eprint={1705.00652},
 archivePrefix={arXiv},
 primaryClass={cs.CL}
}

Base model

ltg/norbert4-base

(4)