# InternVLA-M1 ## Env Setup Install the env ```bash cd src/InternVLA-M1; uv sync --python 3.10 ``` ## On Real Data: 1. training ```bash cd src/InternVLA-M1 bash scripts/train_internvla.sh # will generate stats_gr00t.json in $TASK/meta/ folder ``` 2. serving a checkpoint ```bash cd src/InternVLA-M1 bash scripts/deploy_internvla.sh ``` ## On Sim Data: 1. download sim data ```bash huggingface-cli download --repo-type dataset USC-PSI-Lab/psi-data --include simple/G1WholebodyBendPick-v0-psi0.zip --local-dir $DATA_DIR ``` 2. train on sim data ```bash cd src/InternVLA-M1 bash scripts/train_internvla.sh # will generate stats_gr00t.json in $TASK/meta/ folder ``` 3. serving a checkpoint ```bash cd src/InternVLA-M1 bash scripts/deploy_internvla_sim.sh ``` 4. launch the eval client follow the instructions in examples/simple/README.md and replace the `policy` with `internvla_m1` to launch the eval client