Shiki42/s015-scan-object-sequential50-pi05-step20000

E744 Scan Object PI0.5 checkpoint — step 20000

Inference checkpoint from the S015 balanced Sequential50 Scan Object study. It contains the OpenPI parameter tree and its normalization assets. The package is inference-only; optimizer state and training loader state are excluded.

Training provenance

• Experiment / Run: E744 / E744-R001
• Policy: PI0.5 JAX LoRA; checkpoint step 20000 of 30,000
• Initialization: XinY0201/openpi-pi05-base-jax at 5e62884fcf8cb8f9fc693c9163ea18d3e3739658
• Seed: 87431; batch size 16; gradient accumulation 1
• Dataset: Shiki42/ctr-scan-object-sequential50-20260921 at revision 1b24730ebff0ea3f91652a3806620fb682f413b0
• Data: 50 episodes / 17,008 frames; 25 L→R and 25 R→L whole episodes
• IdleMask: disabled (false); standard temporal padding mask retained
• Normalization: exact global NumPy quantiles at 1% and 99%, linear interpolation; artifact SHA-256 c52e2ec3795214b360821ad7c417297a1ab0b8b684a26fe1e9de55d072260b0d
• CTR commit: f802672965770e821f80769abf631780022af77e
• OpenPI commit: e9ba7b7732a3e66e4bd87d6d3429f7cef6352ead
• Runtime: 2026-08-22.2; manifest SHA-256 06e035fbc62dd7f32f01a2f5ad536f01bd5af6caae3239698dbb8c07ed503332
• Checkpoint verification: fresh-process reload and finite-parameter checks passed for steps 10k, 20k, and 30k
• Evaluation: pending; this card makes no success-rate claim

Files

• params/: OpenPI model parameters
• assets/: normalization assets
• resolved_config.json: resolved inference configuration with machine-local paths removed
• training-provenance.json: pinned training, dataset, normalization, and runtime identities