[mixture] Measure entropy, teacher KL and reward stats. #81
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: slippi-ai test cases | |
| on: [push, pull_request] | |
| jobs: | |
| test: | |
| runs-on: ${{ matrix.os }} | |
| strategy: | |
| matrix: | |
| python-version: ['3.9', '3.10'] | |
| os: ['ubuntu-latest'] | |
| steps: | |
| - uses: actions/checkout@v4 | |
| # See https://github.com/pypa/pip/issues/11281 | |
| - name: Install | |
| run: | | |
| pip install tensorflow ray | |
| pip install -r requirements.txt | |
| pip install . | |
| - name: Run Unit Tests | |
| run: | | |
| python tests/networks_test.py | |
| python tests/rl_lib_test.py | |
| python tests/unit_tests.py | |
| - name: Test Imitation | |
| run: ./tests/training_test.sh | |
| - name: Test Imitation with Restore | |
| run: ./tests/training_test.sh --config.restore_pickle=slippi_ai/data/checkpoints/demo | |
| - name: Test RL with Fake Data | |
| run: ./tests/train_rl.sh | |
| - name: Test RL with Fake Data + Delay | |
| run: ./tests/train_rl.sh --config.override_delay=3 | |
| - name: Test RL with Fake Data + Restore | |
| run: ./tests/train_rl.sh --config.teacher="" --config.restore=slippi_ai/data/checkpoints/rl_demo | |
| - name: Test train_two with Fake Data | |
| run: ./tests/train_two.sh | |
| - name: Test mixture with Fake Data | |
| run: ./tests/train_mixture.sh | |
| - name: Test mixture with Fake Data + Delay | |
| run: ./tests/train_mixture.sh --config.override_delay=3 | |
| - name: Download ExiAI Dolphin | |
| # TODO: specify release version in repo? | |
| run: | | |
| curl -L https://github.com/vladfi1/slippi-Ishiiruka/releases/download/exi-ai-0.1.0/Slippi_Online-x86_64-ExiAI.AppImage -o dolphin | |
| chmod +x dolphin | |
| - name: Install dolphin dependencies | |
| run: sudo apt install -y libfuse2 libegl1 | |
| - name: Cache the SSBM iso | |
| uses: actions/cache@v4 | |
| id: cache-iso # give it a name for checking the cache hit-or-not | |
| with: | |
| path: SSBM.iso | |
| key: ssbm-iso | |
| - name: Download the SSBM iso | |
| if: steps.cache-iso.outputs.cache-hit != 'true' | |
| env: | |
| SSBM_ISO_DRIVE_ID: ${{ secrets.SSBM_ISO_DRIVE_ID }} | |
| run: | | |
| pip install gdown | |
| gdown "https://drive.google.com/uc?id=$SSBM_ISO_DRIVE_ID" -O SSBM.iso | |
| - name: Test RL with Dolphin | |
| run: ./tests/train_rl.sh --config.dolphin.path=./dolphin --config.dolphin.iso=SSBM.iso --config.actor.use_fake_envs=False | |
| - name: Test train_two with Dolphin | |
| run: ./tests/train_two.sh --config.dolphin.path=./dolphin --config.dolphin.iso=SSBM.iso --config.actor.use_fake_envs=False |