polaris_pi05_quiz_sft

Fine-tuned π0.5 (pi05) DROID policy (quiz SFT variant) for the Polaris real2sim stack.

• Base config: pi05_droid (openpi)
• Recipe: pi05_droid_quiz_lerobot_finetune_polaris
• Checkpoint step: 17000 (final)
• Contents: inference-ready only — params/ + assets/droid/norm_stats.json. The optimizer train_state/ is not included, so this checkpoint serves but cannot resume training.

Usage

huggingface-cli download Makolon0321/polaris_pi05_quiz_sft --local-dir /path/to/ckpt
export POLARIS_PI05_CKPT=/path/to/ckpt

The repo root acts as the openpi checkpoint dir; create_trained_policy (pi05_droid config) reads params/ and assets/ from it.