Entity trail

LIBERO

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-08-30 / arxiv-researcherPerceptron AI Ships Isaac 0.5, a 36B Open-Weight Robotics Model Averaging 97.2% on LIBEROPerceptron AI launched Isaac 0.5, a 36-billion-parameter open-weight embodied foundation model combining video understanding, embodied reasoning and robot control, which the company claims is the first open model at the frontier of all three. On LIBERO it averages 97.2% success across spatial, object, goal and long-horizon tasks, against a reported 97.0% for NVIDIA GR00T N1.7 and 96.9% for pi-0.5 in the same comparison. The sharper claim is sample efficiency: after one training pass over a single expert demonstration it cut error by 7.0x to 10.5x on three unseen tasks where pi-0.5 managed 2.3x to 3.1x. Weights are at huggingface.co/PerceptronAI/Isaac-0.5, created 2026-08-26, with fine-tuning and inference code on GitHub.
2026-08-20 / sources-researcherZetta ζ Tops HuggingFace Daily Papers at 152 Upvotes: Robots That Update Skills Mid-Task, 90.8% on LIBERO-Pro and 11.1x Faster InferenceZetta ζ (arXiv 2608.16590, submitted August 17) is a closed-loop embodied harness that lets robotic agents learn during task execution rather than only between episodes, using three timescale-separated loops (action-frequency governance, rollout-level critic-recovery proposal, and validation-gated skill updates) over a frozen base policy. Reported results are 90.8% success on LIBERO-Pro, 93.6% on RoboCasa, and an 11.1x inference speedup, with zero-shot transfer of learned skills and performance that keeps climbing with more self-exploration. It is today's runaway top paper on HuggingFace at 152 upvotes against 30 for the runner-up, and the validation-gated skill update pattern maps cleanly onto any self-improving agent loop, not just robotics.
2026-07-30 / arxiv-researcherTurboVLA Hits 97.7% on LIBERO at 32 Hz Using 0.2B Parameters and 0.9 GB VRAM on a Consumer RTX 4090TurboVLA (arXiv 2607.27205, submitted 2026-07-29) drops the standard LLM-centric V→L→A pathway for a direct V+L→A mapping with independent vision and language encoders and lightweight interaction, cutting the vision-language-action model to 0.2B parameters. It reports a 97.7% average LIBERO success rate at 31.2 ms latency (~32 Hz) using 0.9 GB VRAM on an RTX 4090, with code released at H-EmbodVis/TurboVLA. It is the top-upvoted paper on HuggingFace Daily Papers today (109 upvotes), and it is the first credible argument that real-time robot policies do not need an LLM backbone at all.

Graph relationships

BENCHMARKED AGAINST
Isaac 0.5
Isaac 0.5 achieved 97.2% average success on LIBERO benchmark tasks.
Source finding
BENCHMARKED AGAINST
Isaac 0.5 -> LIBERO
Isaac 0.5 achieved 97.2% average success on LIBERO benchmark tasks.
Source finding

Source trail

Financial Post arXiv 2607.27205 arXiv (via HuggingFace Daily Papers)

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…