{"slug": "zy-s1-preview-born-in-india-built-for-intelligence", "title": "🇮🇳 ZY-S1 Preview — Born in India. Built for Intelligence", "summary": "The ZYR project released ZY-S1 Preview, an experimental language model built on Sapient AI's HRM-Text-1B base model and developed in India, reporting preliminary results from a 100-question sample across five benchmark metrics. ZY-S1 Preview scored 80% on ARC-Challenge, 52% on MMLU, 19.76% DROP Token F1, 4% DROP Exact Match, and 0% MATH-500 Exact Accuracy, which the project described as early experimental checkpoint results rather than full benchmark scores or a state-of-the-art claim. The project said mathematical reasoning and reading comprehension remain priority areas for improvement and asked the Hugging Face community which benchmarks to run next.", "body_md": "Hello, Hugging Face community!\n\nI’m excited to share **ZY-S1 Preview**, an experimental language model from the ZYR project, which is being developed in India.\n\n **Model:** [zyr-AGENT/ZY-S1 · Hugging Face](https://huggingface.co/zyr-AGENT/ZY-S1)\n\nZY-S1 Preview is built on Sapient AI’s HRM-Text-1B base model. It explores additional supervised fine-tuning and corrective fine-tuning to evaluate performance across reasoning, reading comprehension, mathematics, and general knowledge.\n\nIn an initial sample of 100 questions across five benchmark metrics, ZY-S1 Preview achieved:\n\n**ARC-Challenge:** 80%\n\n**MMLU:** 52%\n\n**DROP Token F1:** 19.76%\n\n**DROP Exact Match:** 4%\n\n**MATH-500 Exact Accuracy:** 0%\n\nThese are preliminary results from small samples, not full benchmark scores. Mathematical reasoning and reading comprehension remain important areas for improvement.\n\n**Born in India. Built for Intelligence.**\n\nThe ZYR project is being developed in India, with a focus on AI model training, experimentation, and evaluation.\n\nZY-S1 Preview is an early experimental checkpoint—not a finished model or a claim of state-of-the-art performance.\n\nI’d love to hear from the Hugging Face community:\n\nWhich benchmarks should I run next?\n\nWhat improvements would you recommend for future training?\n\nWhich reasoning or coding tasks should I prioritize?\n\nThank you for checking out the project!\n\n**ZY-S1 Preview — Born in India. Built for Intelligence.**", "url": "https://wpnews.pro/news/zy-s1-preview-born-in-india-built-for-intelligence", "canonical_source": "https://discuss.huggingface.co/t/zy-s1-preview-born-in-india-built-for-intelligence/190405#post_1", "published_at": "2026-10-12 11:08:06+00:00", "updated_at": "2026-10-12 12:14:42.673687+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-research", "ai-startups"], "entities": ["ZYR project", "ZY-S1 Preview", "Sapient AI", "HRM-Text-1B", "Hugging Face", "India"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/zy-s1-preview-born-in-india-built-for-intelligence", "markdown": "https://wpnews.pro/news/zy-s1-preview-born-in-india-built-for-intelligence.md", "text": "https://wpnews.pro/news/zy-s1-preview-born-in-india-built-for-intelligence.txt", "jsonld": "https://wpnews.pro/news/zy-s1-preview-born-in-india-built-for-intelligence.jsonld"}}