dataset

The stablecoin giant Tether's Tether AI Research today released QVAC Genesis III, a synthetic training dataset aimed at the fields of science, technology, engineering, and mathematics, with a scale of 19.143 billion tokens, covering 159.6 million documents.It is reported that Tether hopes to enhance the reasoning and teaching capabilities of small AI models, allowing more AI assistants to run directly on laptops, mobile phones, and local servers, reducing the ongoing reliance on large cloud models.Genesis III continues the data lineage of Genesis I and Genesis II, covering 19 curriculum-related fields including biology, chemistry, physics, mathematics, computer science, medicine, electrical engineering, and machine learning, aimed at high school, university, and professional applications. Tether states that the dataset will focus on training models to explain problem-solving paths, identify erroneous reasoning, and provide corrections, rather than just outputting final answers.According to Tether's disclosed test results, a 1.7 billion parameter model trained with Genesis III option-level data achieved an effective answer rate of 99.45% in relevant benchmark tests. Compared to the Cosmopedia-v2 training model of similar token scale, Genesis III improved scores on the ARC-Easy, ARC-Challenge, and MMLU STEM benchmarks by 28.57, 21.35, and 15.03 percentage points, respectively.

The blockchain data analysis platform Dune announced that the real-world asset dataset is now online.Dune stated that users focusing on tokenized assets can now access headline data such as total on-chain value, product quantity, and growth charts; this dataset further covers details such as distribution channels, ultimate holders, trading venues, and post-holding usage.Dune has also opened the registration channel for the real-world asset report.

According to official news, Tether has released the synthetic educational dataset QVAC Genesis II.It is reported that the QVAC Genesis dataset will cover 19 educational fields, significantly enhancing the scale, depth, and reasoning quality of open AI training data. The new version will expand its coverage to 10 new fields, including chemistry, computer science, statistics, machine learning, astronomy, geography, econometrics, and electrical engineering, while also regenerating university-level physics data using improved methods.

ChainCatcher news, according to the official blog, Tether Data's AI research division QVAC has launched the QVAC Genesis program and released the synthetic dataset Genesis I. This dataset contains 41 billion text tokens, helping to build smarter and more accurate STEM language models globally. The trained models can grasp words and their associative logic. It has been rigorously validated against educational and scientific benchmarks, demonstrating exceptional reasoning and problem-solving abilities in subjects such as mathematics and physics. It is the first publicly available synthetic dataset specifically constructed for educational content and rigorously validated, addressing the lack of publicly available training datasets in critical STEM fields. QVAC Genesis I aims to return the power of AI training to the public through open, high-quality data.In addition, Tether Data has released its first consumer application, QVAC Workbench, aimed at AI enthusiasts and others, supporting various large language models. The application is compatible with smartphones (currently only Android, with iOS to be launched) and desktop platforms, providing comprehensive local device support. When users utilize this application, chat and interaction data is 100% private, and the "delegated inference" feature can connect mobile and desktop versions, making full use of workstation resources.

ChainCatcher news, decentralized AI infrastructure company Gata announced the launch of the globally large-scale real user ChatGPT dialogue dataset ChatGPT-RealUser-2.2M. This dataset was collected through Gata's GPT-to-Earn program (voluntary participation by users), gathering over 2.24 million real conversations and nearly 3.56 million question-answer pairs from over 15,000 real users, covering interactions with GPT-3.5, GPT-4, and o1 during the period of 2024-2025.It is reported that this dataset is approximately twice the size of similar datasets previously released by the Allen Institute for AI, covering real scenarios and multi-turn dialogues, and due to the on-chain incentive mechanism, it contains a large number of crypto-related interactions. A preview version has been launched on Hugging Face, containing 600 dialogue samples, and the complete dataset is available for research and commercial applications.It is known that in May 2025, Gata announced the completion of a $4 million seed round financing, with participation from YZi Labs, IDG Blockchain, and others.

ChainCatcher news, Codatta and DPath.ai have teamed up to release an optimized version of the TCGA PRAD dataset. Through the Codatta platform, global pathology experts are invited to participate, elevating traditional slice-level labeling to ROI region-level spatial annotation, significantly improving the accuracy, detail, and transparency of diagnoses.This collaborative project processed 435 whole slide images, optimized 245 cases, and confirmed the accuracy of 190 annotations, providing critical resources for AI model training and pathology research. Codatta's Royalty Model allows contributors to retain data ownership and gain ongoing revenue through data value growth, while also offering broader data access opportunities for small and medium-sized AI startups.This project demonstrates the potential of decentralized and community-driven solutions in advancing medical AI innovation, highlighting Codatta and DPath.ai's leadership in the field of high-quality data creation.

ChainCatcher news, the decentralized AI data liquidity network Vana announces an important upgrade to the TG App, allowing users to participate in building a deepfake detection dataset to earn points. Additionally, the App has launched a new DataDAO pre-mining feature.According to the RootData page, Vana is an AI identity generation application that allows users to create a digital persona, which can be used across different applications. This digital persona is private and can only be controlled by the user who created it and their authorized accounts. Vana completed a $5 million strategic financing round on September 18, led by Coinbase Ventures. To date, Vana's total financing has reached $25 million.

ChainCatcher news, according to official sources, Google Cloud has announced the addition of Solana data to the BigQuery public dataset program. BigQuery is Google Cloud's serverless, cost-effective data warehouse, with built-in machine learning (ML), artificial intelligence (AI), and business intelligence (BI) capabilities.

ChainCatcher news, the decentralized AI labeling platform MarkerDAO will participate in the Big Demo Day hosted by HK Web3Hub, showcasing its platform design principles, as well as its vision to reduce data labeling costs and alleviate global poverty.According to reports, in this presentation, MarkerDAO, known as the Web3 version of Scale AI, aims to establish a decentralized crowdsourcing platform from the perspective of AI data labeling, allowing users to earn rewards by completing data labeling tasks. MarkerDAO has reached strategic cooperation agreements with several AI research institutions to undertake relevant data labeling orders, and its mobile application will be launched soon.