EttinX Cross-Encoder: Semantic Similarity (STS)

Cross encoders are high performing encoder models that compare two texts and output a 0-1 score. I've found the cross-encoders/roberta-large-stsb model to be very useful in creating evaluators for LLM outputs. They're simple to use, fast and very accurate.

The Ettin series followed up with new encoders trained on the ModernBERT architecture, with a range of sizes, starting at 17M. The reduced parameters and computationally efficient interleaved local/global attention layers make this a very fast model, which can easily process a few hundred sentence pairs per second on CPU, and a few thousand per second on my A6000.

---

Features

• High performing: Achieves Pearson: 0.9004 and Spearman: 0.8926 on the STS-Benchmark test set.
• Efficient architecture: Based on the Ettin-encoder design (68M parameters), offering very fast inference speeds.
• Extended context length: Processes sequences up to 8192 tokens, great for LLM output evals.
• Diversified training: Pretrained on dleemiller/wiki-sim and fine-tuned on sentence-transformers/stsb.

---

Performance

---

Usage

To use EttinX for semantic similarity tasks, you can load the model with the Hugging Face sentence-transformers library:

from sentence_transformers import CrossEncoder

# Load EttinX model
model = CrossEncoder("dleemiller/EttinX-sts-s")

# Predict similarity scores for sentence pairs
sentence_pairs = [
    ("It's a wonderful day outside.", "It's so sunny today!"),
    ("It's a wonderful day outside.", "He drove to work earlier."),
]
scores = model.predict(sentence_pairs)

print(scores)  # Outputs: array([0.9184, 0.0123], dtype=float32)

Output

The model returns similarity scores in the range [0, 1], where higher scores indicate stronger semantic similarity.

---

Training Details

Pretraining

The model was pretrained on the pair-score-sampled subset of the dleemiller/wiki-sim dataset. This dataset provides diverse sentence pairs with semantic similarity scores, helping the model build a robust understanding of relationships between sentences.

• Classifier Dropout: a somewhat large classifier dropout of 0.3, to reduce overreliance on teacher scores.
• Objective: STS-B scores from cross-encoder/stsb-roberta-large.

Fine-Tuning

Fine-tuning was performed on the sentence-transformers/stsb dataset.

Validation Results

The model achieved the following test set performance after fine-tuning:

• Pearson Correlation: 0.9004
• Spearman Correlation: 0.8926

---

Model Card

• Architecture: Ettin-encoder-68m
• Tokenizer: Custom tokenizer trained with modern techniques for long-context handling.
• Pretraining Data: dleemiller/wiki-sim (pair-score-sampled)
• Fine-Tuning Data: sentence-transformers/stsb

---

Thank You

Thanks to the Johns Hopkins team for providing the ModernBERT models, and the Sentence Transformers team for their leadership in transformer encoder models.

Seq vs Seq: An Open Suite of Paired Encoders and Decoders

Citation

If you use this model in your research, please cite:

@misc{ettinxstsb2025,
  author = {Miller, D. Lee},
  title = {EttinX STS: An STS cross encoder model},
  year = {2025},
  publisher = {Hugging Face Hub},
  url = {https://huggingface.co/dleemiller/EttinX-sts-xxs},
}

---

License

This model is licensed under the MIT License.