Swiss Technological Breakthrough: Apertus Open-Source Model Challenges AI's Black Box, Fully Discloses Training Details

Swiss research institutions are collaborating to challenge the "black box" nature of large language models. École Polytechnique Fédérale de Lausanne (EPFL), **ETH Zurich** and the Swiss National Supercomputing Centre (CSCS)** recently released a large-scale open-source language model called Apertus. The model not only reflects its core philosophy in its name, which means "open" in Latin, but also embodies openness in practice to the utmost extent.

Different from mainstream American models such as OpenAI's GPT series, Meta's Llama, and Anthropic's Claude, Apertus stands out for its unprecedented transparency. It openly shares all key information, including model weights, architecture, training code, data recipes, and complete training process documentation, without any reservations. This fully transparent release aims to promote global collaboration and innovation, allowing researchers and developers to deeply understand how the model works and improve and optimize it based on that understanding.

The release of Apertus marks an important step forward for Switzerland in promoting the openness and reproducibility of artificial intelligence research, setting a new benchmark for the industry.

Swiss Three Giants Jointly Launch Open-Source Large Model Apertus to Challenge American Tech Giants

Recently, the Ecole Polytechnique Fédérale de Lausanne (EPFL), ETH Zurich, and the Swiss National Supercomputing Centre (CSCS) jointly launched a large-scale open-source language model called 'Apertus'. The name of the model means 'open' in Latin, and its development philosophy reflects this spirit. Unlike current market offerings such as OpenAI's GPT series, Meta's Llama, and Anthropic's Claude

Redefining Tradition! Mini-o3 Open-Source Model Achieves Ultra-Long Visual Reasoning, Deep Thinking Is No Longer a Challenge

Recently, ByteDance and the University of Hong Kong jointly launched a new open-source visual reasoning model - Mini-o3, marking another major breakthrough in multi-turn visual reasoning technology. Unlike previous visual language models (VLMs) that could only conduct 1-2 rounds of dialogue, Mini-o3 limited the number of dialogue rounds to 6 during training, but during testing it can extend the reasoning rounds to dozens, greatly enhancing the ability to handle visual questions. The strength of Mini-o3 lies in its deep reasoning in high-difficulty visual search tasks, reaching

Google emphasizes the value of AI summaries and the importance of a healthy web

At a recent AI conference in New York, a Google executive defended the company's use of AI summaries in search results. The vice president, named Markham Erickson, responded to a lawsuit brought by Penske Media Corporation, the parent company of Rolling Stone, stating that user needs are gradually shifting from the traditional "10 blue links" to AI-generated context summaries. Image credit: The image was generated by AI, and the image licensing service is Midjourney.

Swiss Technological Breakthrough: Apertus Open-Source Model Challenges AI's Black Box, Fully Discloses Training Details

Related Recommendations

Swiss Three Giants Jointly Launch Open-Source Large Model Apertus to Challenge American Tech Giants

French AI Giant Mistral is Set to Reach a Valuation of $14 Billion

Switzerland Launches Autonomous Open-Source AI Large Model Apertus

Redefining Tradition! Mini-o3 Open-Source Model Achieves Ultra-Long Visual Reasoning, Deep Thinking Is No Longer a Challenge

Google emphasizes the value of AI summaries and the importance of a healthy web