bod9/stella_en_1.5B_v5-ecom-retrieval

SentenceTransformer based on NovaSearch/stella_en_1.5B_v5

This is a sentence-transformers model finetuned from NovaSearch/stella_en_1.5B_v5. It maps sentences & paragraphs to a 1024-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: NovaSearch/stella_en_1.5B_v5
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 1024 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': 512, 'do_lower_case': False}) with Transformer model: Qwen2Model 
  (1): Pooling({'word_embedding_dimension': 1536, '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})
  (2): Dense({'in_features': 1536, 'out_features': 1024, 'bias': True, 'activation_function': 'torch.nn.modules.linear.Identity'})
)

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("bod9/fulldshardneg")
# Run inference
sentences = [
    '#1 small corded treadmill without remote control',
    'SUNNY HEALTH & FITNESS ASUNA Space Saving Treadmill, Motorized with Speakers for AUX Audio Connection - 8730G',
    'Goplus Under Desk Treadmill, with Touchable LED Display and Wireless Remote Control, Built-in 3 Workout Modes and 12 Programs, Walking Jogging Machine, Superfit Electric Treadmill for Home Office',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]

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

Evaluation

Metrics

Information Retrieval

• Dataset: ir_evaluation
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.5244
cosine_accuracy@3	0.7189
cosine_accuracy@5	0.7823
cosine_accuracy@10	0.8453
cosine_precision@1	0.5244
cosine_precision@3	0.4476
cosine_precision@5	0.4002
cosine_precision@10	0.3198
cosine_recall@1	0.0937
cosine_recall@3	0.2136
cosine_recall@5	0.298
cosine_recall@10	0.4318
cosine_ndcg@5	0.4676
cosine_ndcg@10	0.4681
cosine_mrr@1	0.5244
cosine_mrr@5	0.6257
cosine_mrr@10	0.6341
cosine_map@10	0.3493
cosine_map@100	0.4065

Graded IR

• Dataset: gr_evaluation
• Evaluated with GradedIREvaluator.GradedIREvaluator

Metric	Value
cosine_accuracy@1	0.7088
cosine_accuracy@3	0.9038
cosine_accuracy@5	0.9463
cosine_accuracy@10	0.984
cosine_precision@1	0.7088
cosine_precision@3	0.6469
cosine_precision@5	0.6045
cosine_precision@10	0.5284
cosine_recall@1	0.1272
cosine_recall@3	0.3093
cosine_recall@5	0.449
cosine_recall@10	0.7099
cosine_ndcg@5	0.7127
cosine_ndcg@10	0.7543
cosine_mrr@1	0.7088
cosine_mrr@5	0.8066
cosine_mrr@10	0.8118
cosine_map@10	0.6051
cosine_map@100	0.6944

Framework Versions

• Python: 3.11.12
• Sentence Transformers: 4.1.0
• Transformers: 4.51.3
• PyTorch: 2.6.0+cu124
• Accelerate: 1.6.0
• Datasets: 3.6.0
• Tokenizers: 0.21.1

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}
}