#AI is only as powerful as what feeds it. Our #DataEngine and #ResearchCores build the foundation researchers like Dr. Cooper-Knock need to put AI to work for the ALS community.
Read more about how we're building that foundation:
#EveryoneLives
Axis Weekly
We've been experimenting with how to digest data better. After almost a year of building the data flywheel, the next step is a model-guided data engine: use foundation-model performance as feedback to decide what to keep, what to drop, and where the next batch should be produced.
We already made large-scale data generation and collection possible. The remaining bottleneck is knowing what the current model still lacks, so we can filter existing data and aim the next collection round instead of sampling from a fixed distribution. Thesis article on this soon.
Key updates:
• New Dataset & Real2Sim2Real: Pipeline for bringing real objects and scenes into simulation, with hardware validation. Focus on cross-object and cross-scene skills on concrete families first.
• Active curation: Moving from human/VLM/script checks to model-performance-driven selection, so training budget goes to trajectories that still teach something.
• Verification & new tasks: Hardened anti-cheating checks, dual-arm support with ARX5, and new put-in/take-out basket tasks.
Details below 👇
Axis Weekly
We've been experimenting with how to digest data better. After almost a year of building the data flywheel, the next step is a model-guided data engine: use foundation-model performance as feedback to decide what to keep, what to drop, and where the next batch should be produced.
We already made large-scale data generation and collection possible. The remaining bottleneck is knowing what the current model still lacks, so we can filter existing data and aim the next collection round instead of sampling from a fixed distribution. Thesis article on this soon.
Key updates:
• New Dataset & Real2Sim2Real: Pipeline for bringing real objects and scenes into simulation, with hardware validation. Focus on cross-object and cross-scene skills on concrete families first.
• Active curation: Moving from human/VLM/script checks to model-performance-driven selection, so training budget goes to trajectories that still teach something.
• Verification & new tasks: Hardened anti-cheating checks, dual-arm support with ARX5, and new put-in/take-out basket tasks.
Details below 👇
🌟 Exciting News Alert! 🌟 After months of hard work, we're thrilled to unveil the groundbreaking DagsHub Data Engine - Empowering you to manage and improve your unstructured datasets like never before! 🚀💡 #DataEngine#UnstructuredData
The Target ALS #DataEngine breaks barriers in #ALSResearch, providing free or low-cost access to vital datasets. With user guides and tutorials, it empowers researchers—especially early-career scientists—to focus on breakthroughs, not roadblocks.
🔗
Finding and annotating difficult samples is the best way to improve your ML model. But how do you find this informative data? A data engine! These 9 data engine methods can help 👉
#dataengine#machinelearning#ML#AI
The engine driving Web3's data future. 🚀
@brevis_zk is the powerhouse of verifiable intelligence, transforming raw data into actionable insights across the decentralized web. Fueling innovation. 💡🌐 #Brevis#Web3#DataEngine
“Wasting hours pulling fragmented on-chain data? With @Covalent_HQ you get cleaned, indexed, ready-to-use info. @wallchain_xyz flagged this one as infrastructure you can actually build on. #DevTools#DataEngine”
As #DataScience becomes an increasingly integral part of any #business, navigating the evolving complexities of creating a powerful #DataEngine has never been harder. Find out how you can position your business for success in the coming years:
시간당 1만 궤적의 조건
@axisrobotics#Robotics#DataEngine
“시간당 유효 궤적 1만”은 리플레이 검증이 진짜일 때만 의미 있다.
문서 기준: 시뮬 리플레이 → 클리닝 → IsaacSim 도메인 랜덤화로 데모 하나를 수천 장의 포토리얼 샘플로 늘린다. 필터 없는 크라우드 텔레옵은 리더보드 달린 노이즈다. 필터가 제품이다.
랩은 논문, 엔진은 커버리지
폐쇄형 랩은 논문을 최적화한다. 오픈 데이터 엔진은 커버리지를 최적화한다.
한 사이트, 한 체형, 한 조명으로는 VLA가 일반화되지 않는다. Axis 에셋 라이브러리는 싱글암, 듀얼암, 덱스터러스 핸드, 휠 바이매뉴얼, 휴머노이드를 염두에 둔다. Franka는 천장이 아니라 증명이다.
🚀 Exciting news! Our team is thrilled to announce the release of Data Engine, a powerful toolset designed to empower ML teams and make the most of their unstructured data! 🌟💻 #DataEngine#MachineLearning
The Target ALS #DataEngine just got bigger!
We’ve expanded datasets across all core collections, added a user manual, and released new demo videos. Now it’s easier than ever to explore and utilize rich data for #ALSResearch.
Explore the updates:
AI model training costs are reducing by a factor of 2 every 16 months. With state-of-the-art neural network architectures now widely available, it is clear that data is the most critical piece in developing performant AI systems. @labelbox#DataEngine#AI