# spaCy Model Version Reference — Day 3 Presidio Custom-Recognizers Lab
#
# This lab uses NO Ollama model, so there are no GGUF blob hashes to verify
# (contrast with day3-redaction-monitoring/expected/model_hashes.txt). The only
# "model" is the spaCy en_core_web_lg NER pipeline, installed as a pip package.
#
# ── spaCy en_core_web_lg NER model ───────────────────────────────────────────
# Source: explosion/spacy-models (GitHub releases), installed via pip at build time
# Size: ~587 MB (includes word vectors for en_core_web_lg)
# License: MIT (spaCy models)
# Installation: python -m spacy download en_core_web_lg  (runs in Dockerfile RUN step)
#
# Unlike HuggingFace models, spaCy models are Python packages installed to
# site-packages. They DO NOT have a separate cache / TRANSFORMERS_OFFLINE-style flag.
# spacy.load("en_core_web_lg") makes NO network call once the package is installed.
#
# The spaCy model version installed depends on the spaCy version in requirements.txt.
# For spaCy 3.7.x, the model is en_core_web_lg-3.7.1 or later.
#
# Verify post-build:
#   docker run --rm day3-presidio-recognizers-runner:local \
#     python -c "import en_core_web_lg; print(en_core_web_lg.__version__)"
#
# Verify Presidio version:
#   docker run --rm day3-presidio-recognizers-runner:local \
#     python -c "import presidio_analyzer; print(presidio_analyzer.__version__)"
