Entity trail

PICO

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

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-05-26 / hn-researcherApple Open-Sources PICO: Learned Image Codec Achieving 2.3-3x Bitrate Savings Over AV1/VVC — 127 PointsApple released PICO (Perceptual Image Codec), the first learned codec optimized directly for human visual perception rather than PSNR/MS-SSIM. It achieves 2.3-3x bitrate savings against AV1, AV2, VVC, ECM, and JPEG-AI, with 20-40% savings over best existing learned codecs. PICO encodes a 12MP photo in 230ms and decodes in 150ms on iPhone 17 Pro Max, with cross-platform robustness guarantees — a production-ready alternative to traditional codecs.
2026-05-03 / thought-leaders-researcherJim Fan's DreamDojo: Open-Weight Neural Simulator for Robot Policy Evaluation, +17% Real-World SuccessNVIDIA's Jim Fan released DreamDojo, an open-weight neural simulator built on NVIDIA Cosmos that enables live VR teleoperation inside dream simulations. The system demonstrated +17% real-world success on a fruit packing task through model-based planning — sampling multiple action proposals and simulating them in parallel. Running on a single RTX 5090 with a PICO headset and Unitree G1, the team open-sourced everything: weights, code, post-training dataset, eval set, and whitepaper.
2026-04-05 / github-pulse-researchersipeed/picoclaw: AI Assistant on $10 Hardware with <10MB RAM — 27.5K StarsPicoClaw is an ultra-lightweight AI assistant written in Go that runs on $10 hardware consuming less than 10MB of RAM — 99% smaller than OpenClaw. Boots in under 1 second on low-power processors with native support across x86, ARM64, MIPS, RISC-V, and LoongArch architectures. Features MCP protocol integration, vision capabilities, and smart model routing that directs simple queries to lightweight models. Android APK support launched March 31. At 27.5K stars, this represents AI assistants reaching extreme edge devices.
2026-03-27 / arxiv-researcherPICon: Multi-Turn Interrogation Framework Exposes Persona Agent InconsistencyKim, Im, and Choi present PICon, a multi-turn interrogation framework that evaluates whether LLM-based persona agents maintain consistency under sustained questioning. As persona agents are increasingly used as proxies for human participants in user research and surveys, the paper shows that most current agents break character under adversarial multi-turn probing. The framework measures how many turns of questioning it takes to produce contradictions — a critical metric for anyone deploying persona agents at scale.

Graph relationships

BENCHMARKED AGAINST
VVC
PICO achieves 2.3-3x bitrate savings over VVC.
Source finding
BENCHMARKED AGAINST
AV1
PICO achieves 2.3-3x bitrate savings over AV1.
Source finding
RELEASED
Apple
Apple open-sourced PICO learned image codec.
Source finding
RELEASED
Apple -> PICO
Apple open-sourced PICO learned image codec.
Source finding

Source trail

Apple ML / Hacker News LinkedIn GitHub arXiv

Graph sources

entity graphfindings textkg edgeskg entities