Jobs abstaining field classifiers (sklearn)
Three single-label TF-IDF + LogReg pipelines consumed by jobs-shared
(jobs_shared.ai.categorizers.classifiers) as gap-fillers for the
deterministic JD detectors. Each file is a joblib-compressed sklearn
Pipeline. The runtime abstains (returns None) when the top-class
confidence (and optional runner-up margin) fall below per-classifier gates.
- Files:
work_mode.joblib,level.joblib,employment_type.joblib. - HF repo:
gateswang00/job_field_classifiers
Load one locally
import joblib
from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="gateswang00/job_field_classifiers", filename="work_mode.joblib")
clf = joblib.load(path)
proba = clf.predict_proba(["This is a fully remote role…"])[0]