split	recall@1	F1	threshold	diag mean	off-diag mean
pretrained	0.8142	0.3063	0.6441	0.5653	0.3167
run_0	0.9514	0.9151	0.6778	0.8442	0.0089
run_1	0.9564	0.9221	0.6777	0.8301	-0.0012
run_2	0.9539	0.9356	0.6818	0.8507	-0.0019
run_3	0.9564	0.9299	0.6863	0.8859	0.0112
run_4	0.9564	0.9433	0.7022	0.9511	0.0012

Training

Symmetric InfoNCE / NT-Xent over in-batch negatives. Best checkpoint selected by validation loss.

Run 0 — batch=16, epochs=5, lr=2e-05, schedule=constant, T=0.05
Run 1 — batch=50, epochs=10, lr=2e-05, schedule=constant, T=0.05
Run 2 — batch=70, epochs=30, lr=2e-05, schedule=constant, T=0.05
Run 3 — batch=128, epochs=30, lr=5e-05, schedule=warmup_cosine, T=0.05
Run 4 — batch=70, epochs=50, lr=2e-05, schedule=constant, T=0.07

Loading

python

from transformers import AutoModel, AutoTokenizer
tok   = AutoTokenizer.from_pretrained("shaswatamitra/falcon-snort-dual-all-mpnet-base-v2", subfolder='rule')
model = AutoModel.from_pretrained("shaswatamitra/falcon-snort-dual-all-mpnet-base-v2", subfolder='rule')

Dual-encoder layout: this repo has rule/ (encodes SNORT rules) and cti/ (encodes CTI text) subfolders. Load each with subfolder=....

Citation

bibtex

@article{mitra2025falcon,
  title={FALCON: Autonomous Cyber Threat Intelligence Mining with LLMs for IDS Rule Generation},
  author={Mitra, Shaswata and Bazarov, Azim and Duclos, Martin and Mittal, Sudip and Piplai, Aritran and Rahman, Md Rayhanur and Zieglar, Edward and Rahimi, Shahram},
  journal={arXiv preprint arXiv:2508.18684},
  year={2025}
}

shaswatamitra/falcon-snort-dual-all-mpnet-base-v2

FALCON dual-encoder — SNORT / all-mpnet-base-v2

Test-set metrics

Training

Loading

Citation

shaswatamitra/
falcon-snort-dual-all-mpnet-base-v2

FALCON dual-encoder — SNORT / `all-mpnet-base-v2`