Discover quality AI tools
Bring together the world’s leading artificial intelligence tools to help improve work efficiency
Articles published by BillyThomas
Professor Wang Xiaobo from Anhui Agricultural University unveiled the vertical large model “Fengshu” 2.0 for soybean research at the 32nd National Soybean Research and Production Seminar in Nanning. The updated version integrates multiple data sources (germplasm resources, genomes, disease samples, breeding records, literature) to form a structured knowledge system. It features a multi-model collaborative analysis mechanism that reduces bias from single models through independent responses to professional questions, followed by a summary module that evaluates consensus, differences and evidence quality using domain knowledge and private data. The model offers six functional modules for disease diagnosis, parent selection, virtual cross design, phenotypic analysis, marker-assisted breeding and gene function study, built on a knowledge graph with 20,000 entities and 100,000 relationships, along with extensive genomic and phenotypic data. This upgrade enhances the stability of soybean research tools by avoiding errors caused by single-model “hallucinations”.
Professor Wang Xiaobo from Anhui Agricultural University unveiled the vertical large model “Fengshu” 2.0 for soybean research at the 32nd National Soybean Research and Production Seminar in Nanning. The updated version integrates multiple data sources (germplasm resources, genomes, disease samples, breeding records, literature) to form a structured knowledge system. It features a multi-model collaborative analysis mechanism that reduces bias from single models through independent responses to professional questions, followed by a summary module that evaluates consensus, differences and evidence quality using domain knowledge and private data. The model offers six functional modules for disease diagnosis, parent selection, virtual cross design, phenotypic analysis, marker-assisted breeding and gene function study, built on a knowledge graph with 20,000 entities and 100,000 relationships, along with extensive genomic and phenotypic data. This upgrade enhances the stability of soybean research tools by avoiding errors caused by single-model “hallucinations”.
NVIDIA released the Nemotron-3-Embed series of embedding models for production RAG, code retrieval, and agent memory. The 8B BF16 version ranks first on the RTEB benchmark with 78.46 NDCG@10. The series includes 1B BF16 and 1B NVFP4 (4-bit) versions, all based on Mistral architecture, supporting 34 languages and 32K tokens. The 1B models were created via pruning and knowledge distillation. The NVFP4 version retains 99.5% precision with 2x throughput on Blackwell. It supports vLLM deployment; BF16 versions support Transformers and Sentence Transformers.
NVIDIA released the Nemotron-3-Embed series of embedding models for production RAG, code retrieval, and agent memory. The 8B BF16 version ranks first on the RTEB benchmark with 78.46 NDCG@10. The series includes 1B BF16 and 1B NVFP4 (4-bit) versions, all based on Mistral architecture, supporting 34 languages and 32K tokens. The 1B models were created via pruning and knowledge distillation. The NVFP4 version retains 99.5% precision with 2x throughput on Blackwell. It supports vLLM deployment; BF16 versions support Transformers and Sentence Transformers.
Apple released iOS 27 at WWDC, integrating Apple Intelligence into native apps for system-level smart experiences. New features include AI-driven bill splitting via Apple Cash, automatic password updates, contextual suggestions in Messages, call context extraction, natural language calendar creation, and smart tab organization in Safari. Developer beta is available; public beta and full release expected this fall.
Apple released iOS 27 at WWDC, integrating Apple Intelligence into native apps for system-level smart experiences. New features include AI-driven bill splitting via Apple Cash, automatic password updates, contextual suggestions in Messages, call context extraction, natural language calendar creation, and smart tab organization in Safari. Developer beta is available; public beta and full release expected this fall.
Microsofts Bing team open-sourced its Harrier word embedding model featuring multilingual support for over 100 languages. Trained on 2B+ examples and GPT-5 synthetic data with a 32k-token context, it comes in three sizes up to 2.7B parameters. The model will be integrated into Bing and future AI agents for enhanced search and processing.
Microsofts Bing team open-sourced its Harrier word embedding model featuring multilingual support for over 100 languages. Trained on 2B+ examples and GPT-5 synthetic data with a 32k-token context, it comes in three sizes up to 2.7B parameters. The model will be integrated into Bing and future AI agents for enhanced search and processing.
AI company Anthropic is in final talks to raise over 20 billion dollars in a funding round expected to close next week. The deal would nearly double its valuation to about 350 billion dollars, highlighting strong investor confidence in the generative AI sector. The funds will support model development amid rising computing costs and intense industry competition.
AI company Anthropic is in final talks to raise over 20 billion dollars in a funding round expected to close next week. The deal would nearly double its valuation to about 350 billion dollars, highlighting strong investor confidence in the generative AI sector. The funds will support model development amid rising computing costs and intense industry competition.
