CoRM

Checkpoints for the paper Beyond Magnitude: Contrastive Routing for Modular Mixture-of-Experts.

Model	Active Params	Total Params	Routing	Link
CoRM-182M	182M	777M	Top-1	link
CoRM-182M	266M	777M	Top-2	link
CoRM-469M	469M	2.58B	Top-1	link

Usage

These models use custom modeling code, so trust_remote_code=True is required.

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "ilsp/CoRM-182M-top1"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, trust_remote_code=True)

inputs = tokenizer("The capital of Greece is", return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=32)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Limitations

These are base models pretrained on The Pile; they are not instruction-tuned. The Hugging Face implementation does not use a KV cache and ignores attention_mask, so pass unpadded sequences.

Citation

@misc{xiros2026magnitudecontrastiveroutingmodular,
      title={Beyond Magnitude: Contrastive Routing for Modular Mixture-of-Experts},
      author={Nikolaos Xiros and Dimitrios Damianos and Maria-Eleni Zoumpoulidi and Leon Voukoutis and Vassilis Katsouros and Georgios Paraskevopoulos},
      year={2026},
      eprint={2609.01100},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2609.01100},
}