DeepSeek V3 is a big deal for a number of reasons.

At only $5.5 million to train, it's a fraction of the cost of models from OpenAI, Google, or Anthropic which are often in the hundreds of millions.

It breaks the whole AI as a service business model that OpenAI and Google have been pursuing making state-of-the-art language models accessible to smaller companies, research institutions, and even individuals.

The code is publicly available, allowing anyone to use, study, modify, and build upon it. Companies can integrate it into their products without paying for usage, making it financially attractive. The open-source nature fosters collaboration and rapid innovation.

The model goes head-to-head with and often outperforms models like GPT-4o and Claude-3.5-Sonnet in various benchmarks. It excels in areas that are traditionally challenging for AI, like advanced mathematics and code generation. Its 128K token context window means it can process and understand very long documents. Meanwhile it processes text at 60 tokens per second, twice as fast as GPT-4o.

The Mixture-of-Experts (MoE) approach used by the model is key to its performance. While the model has a massive 671 billion parameters, it only uses 37 billion at a time, making it incredibly efficient. Compared to Meta's Llama3.1 (405 billion parameters used all at once), DeepSeek V3 is over 10 times more efficient yet performs better.

DeepSeek V3 can be seen as a significant technological achievement by China in the face of US attempts to limit its AI progress. China once again demonstrates that resourcefulness can overcome limitations.

https://archive.ph/kI6je

A highly sensitive reusable electrochemical sensor for detecting the pesticide carbosulfan in food products has been developed by scientists at Tomsk Polytechnic University (TPU) in collaboration with their colleagues.

According to the researchers, the sensor is ten times more sensitive to this life-threatening substance compared to existing analogs. The results of the study were published in the scientific journal Microchemical Journal.

Carbosulfan is a chemical used in agriculture to combat insect pests, such as the Colorado potato beetle. While highly effective, it is toxic to humans.

Researchers from TPU's School of Natural Resources Engineering and the Research School of Chemical and Biomedical Technologies, in collaboration with the Institute of Strength Physics and Materials Science of the Siberian Branch of the Russian Academy of Sciences (SB RAS) and Charles University in the Czech Republic, have developed a highly sensitive sensor to detect carbosulfan residues in food.

"The unique features of the sensor are its use of inexpensive materials, compact size, and rapid results. It can detect the presence of carbosulfan at concentrations ten times lower than what existing analog devices can achieve," said Elena Dorozhko, a co-author of the project and Associate Professor at the Department of Chemical Engineering at TPU's School of Natural Resources Engineering.

Neurosymbolic AI is a hybrid approach aiming to bridge the gap between neural networks' ability to learn patterns and symbolic AI's capacity for logical reasoning and explainability.

This approach may offer the best of both worlds combining robust learning from data and clear with understandable reasoning based on knowledge. It has the potential to outperform systems relying solely on either neural networks or symbolic logic and to provide clear explanations for its decisions.

The approach involves encoding structured symbolic knowledge into a format that can be integrated with neural networks and then mapping information from neural patterns back to structured symbolic representations.

Seems like it was merged with some changes after the decision was roundedly berated.

A 26-year-old former OpenAI researcher, Suchir Balaji, was found dead in his San Francisco apartment in recent weeks, CNBC has confirmed.

Balaji left OpenAI earlier this year and raised concerns publicly that the company had allegedly violated U.S. copyright law while developing its popular ChatGPT chatbot.

“The manner of death has been determined to be suicide,” David Serrano Sewell, executive director of San Francisco’s Office of the Chief Medical Examiner, told CNBC in an email on Friday. He said Balaji’s next of kin have been notified.

The San Francisco Police Department said in an e-mail that on the afternoon of Nov. 26, officers were called to an apartment on Buchanan Street to conduct a “wellbeing check.” They found a deceased adult male, and discovered “no evidence of foul play” in their initial investigation, the department said.

(...)

OpenAI is currently involved in legal disputes with a number of publishers, authors and artists over alleged use of copyrighted material for AI training data. A lawsuit filed by news outlets last December seeks to hold OpenAI and principal backer Microsoft accountable for billions of dollars in damages.

ROTTERDAM, Dec 9 (Reuters) - A Dutch court on Monday extended the detention of a Russian former employee of semiconductor equipment maker ASML (ASML.AS), opens new tab suspected of stealing intellectual property and selling it to buyers in Russia in violation of European sanctions.

The suspect is German Aksenov, a 43-year-old man who has worked for ASML subsidiary Mapper and chip technology company NXP (NXPI.O), opens new tab.

He is suspected of stealing design manuals for microchips, microchip equipment and for technology with potential military applications that belonged to ASML, Mapper, NXP and the Delft University of Technology, to which he had access through his employment.

During the initial hearing in the case on Monday, the prosecution said Aksenov took USB sticks with the information to Moscow and gave them to state-owned companies for cash, with the goal of setting up a microchip plant.