# Model Hashes — Day 2 Build-Your-Own-Guardrail Lab
#
# This lab uses NO Ollama model. Its only "model" is the DeBERTa v3
# prompt-injection classifier, which is pre-baked into the runner image at BUILD
# time (see app/prebake_deberta.py + the Dockerfile) and loaded OFFLINE at
# runtime (TRANSFORMERS_OFFLINE=1 / HF_HUB_OFFLINE=1).

# ── DeBERTa prompt-injection classifier ─────────────────────────────────────
# Model: protectai/deberta-v3-base-prompt-injection-v2
# Source: HuggingFace Hub (downloaded at BUILD TIME into /workspace/hf_cache/)
# Size: ~184 MB (weights + tokenizer config)
# License: MIT (protectai/deberta-v3-base-prompt-injection-v2)
# Revision pinned by llm-guard: 89b085cd330414d3e7d9dd787870f315957e1e9f
#
# This model is baked into the day2-byo-guardrail-runner:local Docker image at
# pre-stage time. Verify at runtime that TRANSFORMERS_OFFLINE=1 prevents any HF
# network access:
#   docker compose run --rm runner python -c "
#   import os; assert os.environ.get('TRANSFORMERS_OFFLINE') == '1', 'offline not set'
#   from llm_guard.input_scanners import PromptInjection
#   scanner = PromptInjection()
#   print('Loaded offline OK')
#   "
