CaPtain: A Pipeline for Post-Training Code-as-Policy Robot Agents
READ THE ORIGINAL SOURCEOpen original sourcehuggingface.co · Opens in a new tabCaPtain is a post-training pipeline that turns a small open vision-language model into a code-as-policy robot agent without a vision-language-action model, frontier model, or privileged state at test time. The policy is first aligned via supervised fine-tuning on synthetic trajectories, then trained with multi-turn reinforcement learning using a physical critic that reads privileged simulator state to credit each agent request. Starting from Qwen3.5-4B, the reported held-out LIBERO success rises from 55.4% after SFT to 73.8% after RL. Code, data, and weights are said to be released.
One primary source; other links are context, not independent corroboration.
- Primary sourceHugging Face Daily Papers
- Discovery linkHugging Face
- Linked in sourceView arXiv page
- Linked in sourceView PDF
The paper page is linked. Peer review and findings have not been independently checked by AINewsIQ.
Open full article view →- Original paper+45
- Fetched source page+30
- No editor link review+0
- Original non-discussion link+5
This measures the stored source trail and editorial steps. It does not certify accuracy, peer review, or independent corroboration. Popularity adds no points.
View full scoring →Read the scoring rubric →What the source says
CaPtain is a post-training pipeline that turns a small open vision-language model into a code-as-policy robot agent without a vision-language-action model, frontier model, or privileged state at test time. The policy is first aligned via supervised fine-tuning on synthetic trajectories, then trained with multi-turn reinforcement learning using a physical critic that reads privileged simulator state to credit each agent request. Starting from Qwen3.5-4B, the reported held-out LIBERO success rises from 55.4% after SFT to 73.8% after RL. Code, data, and weights are said to be released.
A READER'S EVIDENCE CHECKHow to read this signalSix questions about claims, sources, risks, and open checks
This is a signal-level guide until claim-level evidence is available. “Not recorded” means the evidence has not been assessed here; it does not mean a result is false. How we check AI news →
- 01Who is making each claim?
The claimant and whether each claim was measured or asserted are not recorded in this public brief.
- 02What was actually tested?
Not recorded per claim. Check the source for sample size, tasks, model and benchmark versions, and reproducibility.
- 03Where are the sources?
Original link: huggingface.co. Brief basis: fetched source page. Primary page date and quote locations are not established in this panel.
- 04What could mislead?
Source claims not independently checked; results are reported on simulation benchmarks, with real-robot evidence described only briefly.
- 05What is at risk if wrong?
Not assessed per claim. Consider safety, privacy, cost, and adoption decisions before relying on the result.
- 06What would change the view?
No claim-specific check is recorded. An inspectable primary method or independent replication could change the assessment.
Key findings from the stored brief
- Starting from Qwen3.5-4B at 0% zero-shot, CaPtain-4B reaches 55.4% held-out LIBERO success after SFT and 73.8% after RL.
- A physical critic reads privileged simulator state the policy never sees and credits each request by its change in predicted success chance.
- On LIBERO-PRO the model reaches 78.5%, reported as surpassing several frontier-model agents.
- Without target-domain training it attains 83.3% on SimplerEnv WidowX and runs on a real robot after simulation-only adaptation.
Source claims not independently checked; results are reported on simulation benchmarks, with real-robot evidence described only briefly.
This is a source-based abstract, not independent validation of the source's claims.
Evidence & sources
These links show where the signal came from and what the source page linked to. A linked page is context, not independent corroboration.
Related archive coverage
These relationships record a citation or topic overlap, not confirmation of the finding.
The paper page is linked. Peer review and findings have not been independently checked by AINewsIQ.
Timeline & updates
- Signal dated by source or discovery feed
This timestamp places the signal in the radar. It may be the discovery post date rather than the primary page's first publication date.
- Shared brief stored
AINewsIQ prepared one abstract from fetched source page.
A publication revision history is not stored for this signal. New engagement snapshots do not mean the source itself changed.
Why this may matter
The work suggests small open vision-language models can be trained into capable code-as-policy robot agents, rather than relying on closed frontier models or pre-trained VLAs. The physical critic offers a way to assign credit across delayed physical effects in multi-turn RL.
Limits and open questions
Source claims not independently checked; results are reported on simulation benchmarks, with real-robot evidence described only briefly.
Organization-specific financial, security, and operational effects require the organization's own context and review.
Discussion
Comments are reviewed before publication. Discuss the evidence, not the person.
No published comments yet.
Reader discussion opens with accounts. You can read the original source and published comments now.