ModernBERT-large-32768

This model is a 4.56% smaller version of answerdotai/ModernBERT-large optimized for English language via vocabulary size reduction using the trimming method.
This trimmed model should perform similarly to the original model with only 32,768 tokens and a much smaller memory footprint.

Model Statistics

Metric	Original	Trimmed	Reduction
Vocabulary size	50,368	32,768	34.94%
Model size	395,881,664 params	377,841,664 params	4.56%

Mining Dataset Statistics

Number of texts used for mining: 200,000 texts
Dataset: lbourdois/fineweb-2-trimming

Usage

from transformers import AutoModel, AutoTokenizer

model_name = "alphaedge-ai/ModernBERT-large-32768"
model = AutoModel.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)

Citations

ModernBERT

@misc{modernbert,
      title={Smarter, Better, Faster, Longer: A Modern Bidirectional Encoder for Fast, Memory Efficient, and Long Context Finetuning and Inference}, 
      author={Benjamin Warner and Antoine Chaffin and Benjamin Clavié and Orion Weller and Oskar Hallström and Said Taghadouini and Alexis Gallagher and Raja Biswas and Faisal Ladhak and Tom Aarsen and Nathan Cooper and Griffin Adams and Jeremy Howard and Iacopo Poli},
      year={2024},
      eprint={2412.13663},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2412.13663}, 
}

Trimming blog post

@misc{hf_blogpost_trimming,
      title={Introduction to Trimming}, 
      author={Loïck BOURDOIS and Tom AARSEN and Bram VANROY and Christopher AKIKI and Woojun JUNG and Manuel ROMERO and Prithiv SAKTHI},
      year={2026},
      url={https://huggingface.co/blog/lbourdois/introduction-to-trimming}, 
}