Modelpublic

mmwanje/kubeops-bge-large-ops-issues

sourceHugging Faceupdated 21d agoView on Hugging Face

0likes12downloads

Model Card

kubeops-bge-large-ops-issues

Fine-tuned BAAI/bge-large-en-v1.5 encoder for Kubernetes operator issue text (GitHub issues and Stack Overflow posts).

Used in the KubeOps study to embed operator-related support text for taxonomy classification (configuration, logical enhancements, observability, security, other).

Training

—Objective: MultipleNegativesRankingLoss (contrastive learning on same-category pairs)
—Data: Labeled GitHub operator issues from the KubeOps corpus
—Backbone: Last 2 transformer layers + pooler unfrozen; remaining layers frozen
—Max sequence length: 128 tokens
—Epochs: 5

Usage

python

from sentence_transformers import SentenceTransformer

model = SentenceTransformer("mmwanje/kubeops-bge-large-ops-issues")
embeddings = model.encode(["RBAC ClusterRole binding fails on upgrade"])

Citation

If you use this model, please cite the KubeOps / Suture paper.