llama.cpp/.github/workflows/pre-tokenizer-hashes.yml

name: Check Pre-Tokenizer Hashes

on:
    push:
        paths:
            - 'convert_hf_to_gguf.py'
            - 'convert_hf_to_gguf_update.py'
    pull_request:
        paths:
            - 'convert_hf_to_gguf.py'
            - 'convert_hf_to_gguf_update.py'

jobs:
    pre-tokenizer-hashes:
        runs-on: ubuntu-latest

        steps:
        - name: Checkout repository
          uses: actions/checkout@v4

        - name: Set up Python
          uses: actions/setup-python@v5
          with:
              python-version: '3.11'

        - name: Install Python dependencies
          run: |
              python3 -m venv .venv
              .venv/bin/pip install -r requirements/requirements-convert_hf_to_gguf_update.txt

        - name: Update pre-tokenizer hashes
          run: |
              cp convert_hf_to_gguf.py /tmp
              .venv/bin/python convert_hf_to_gguf_update.py --check-missing

        - name: Check if committed pre-tokenizer hashes matches generated version
          run: |
              if ! diff -q convert_hf_to_gguf.py /tmp/convert_hf_to_gguf.py; then
                  echo "Model pre-tokenizer hashes (in convert_hf_to_gguf.py) do not match generated hashes (from convert_hf_to_gguf_update.py)."
                  echo "To fix: run ./convert_hf_to_gguf_update.py and commit the updated convert_hf_to_gguf.py along with your changes"
                  echo "Differences found:"
                  diff convert_hf_to_gguf.py /tmp/convert_hf_to_gguf.py || true
                  exit 1
              fi
              echo "Model pre-tokenizer hashes are up to date."