Onol/pi05-cup-policy-real20-5k

robotics model by Onol

Older snapshot: check the original source for current information.

From the publisher

Model card & documentation

Source preview

Read the publisher’s intended use, setup instructions, evaluations and limitations. The original model card is the source of truth.

This is an independent real-only fine-tune initialized from lerobot/pi05base. It does not continue the separately trained 40k simulation checkpoint. Source: Onol/nero-dexhand-raw, revision 9b780865d5be470fba2a4fb2d379acee6729a150, archive sessions/CupPolicy/archives/20261021T091102Z-34ffb8c160/. The owner's accepted selection contains episodes 001–012 and 014–021 (20 total). Invalid alignment gaps are split; every selected raw demonstration must contribute valid data. The optimizer/model parameters match the preceding simulation run: batch 16, BF16 with useamp=false, LoRA language rank/alpha 16 and action-expert rank/alpha 32, AdamW lr 2.5e-5, betas [0.9, 0.95], weight decay 0.01, clip norm 1, warmup 1000, cosine decay horizon 30000, seed 1000, 8 workers, action chunk 50, inference denoising steps 10, tokenizer length 256. The original LeRobot scheduler automatically rescales short runs: at 5000 total steps the effective warmup is 166 steps and decay ends at 5000 (2.5e-6). The scheduler configuration and implementation are retained; this is its normal short-run behavior. The requested training duration is exactly 5000 updates; only the 5k…

Onol/pi05-cup-policy-real20-5k on Hugging Face A short preview, not the full document.

Read the full model card ↗ · Preview checked 2026-10-28T08:38:53.053Z

Inside the original model card — Document outline

pi0.5 CupPolicy — 20 real demonstrations, 5k updates

Headings are captured from the source. Links open the publisher’s document, not a locally hosted copy.

Documentation belongs to its respective authors. A listing is not a grant of reuse or training rights. Confirm the document’s own terms at the source.

Model overview

Onol/pi05-cup-policy-real20-5k is a robotics model repository published by Onol on Hugging Face. The source reports the lerobot library.

This page summarizes Hub metadata. For intended use, training data, evaluation results and limitations, consult the original model card.

Model facts

Task: robotics
Library: lerobot
Recent downloads (30 days): 0
Cumulative likes: 0
Hugging Face trending score: 0
Reported safetensors parameters: Not reported
Architecture: Not reported
License: Not reported — check the model card
Access: Not gated by Hugging Face
Created: 2026-10-28T06:26:14.000Z
Last modified: 2026-10-28T06:26:15.000Z

Compatibility and lineage

Language coverage was not reported.

Reported base models: lerobot/pi05_base

Parameter count is not a RAM/VRAM requirement. Check precision, quantization, context length and runtime compatibility in the model card; no hardware or API-cost claim is inferred here.

Use and evaluate

Check license terms and access requirements first. Review the model card’s documented loading instructions, evaluations and safety limitations. Benchmark scores are not imported by this directory, and popularity is not an accuracy ranking.

No model weights are downloaded or executed by AltAPIs. Never enable remote model code without reviewing it.

Model card and usage instructions

Model files and configuration

Community discussion