Customize Consent Preferences

We use cookies to help you navigate efficiently and perform certain functions. You will find detailed information about all cookies under each consent category below.

The cookies that are categorized as "Necessary" are stored on your browser as they are essential for enabling the basic functionalities of the site. ... 

Always Active

Necessary cookies are required to enable the basic features of this site, such as providing secure log-in or adjusting your consent preferences. These cookies do not store any personally identifiable data.

No cookies to display.

Functional cookies help perform certain functionalities like sharing the content of the website on social media platforms, collecting feedback, and other third-party features.

No cookies to display.

Analytical cookies are used to understand how visitors interact with the website. These cookies help provide information on metrics such as the number of visitors, bounce rate, traffic source, etc.

No cookies to display.

Performance cookies are used to understand and analyze the key performance indexes of the website which helps in delivering a better user experience for the visitors.

No cookies to display.

Advertisement cookies are used to provide visitors with customized advertisements based on the pages you visited previously and to analyze the effectiveness of the ad campaigns.

No cookies to display.

90年代的黄河路
0

Beijing – The open-source landscape for AI model optimization is becoming increasingly crowded, with competition heating up between leading Chinese AI companies. Moonshot AI, the company behind the popular Kimi chatbot, has released an upgraded version of its Muon optimizer, claiming it can achieve twice the efficiency while using half the computing power compared to the widely-used AdamW optimizer.

This move comes shortly after DeepSeek, another prominent AI firm, announced plans to open-source five code repositories this week. The timing suggests a potential rivalry between the two companies, reminiscent of their previous near-simultaneous releases of improved attention mechanisms, as reported by Machine Heart.

The original Muon optimizer was known for its effectiveness in training smaller language models. However, its scalability to larger models remained a question. To address this, Moonshot AI’s team focused on two key improvements:

  • Adding weight decay: This is crucial for scaling the optimizer to handle larger, more complex models.
  • Consistent RMS updates: This ensures consistency in the root mean square (RMS) of model updates.

These enhancements allow Muon to be used directly in large-scale training without requiring extensive hyperparameter tuning. According to Moonshot AI’s scaling law experiments, the improved Muon optimizer achieves a two-fold increase in computational efficiency compared to the computationally optimal AdamW optimizer.

The release of Moonshot AI’s optimized Muon is a significant development for the AI community. By open-sourcing this technology, Moonshot AI is contributing to the advancement of more efficient and accessible AI model training. This could potentially lower the barrier to entry for researchers and developers working on large language models, enabling faster innovation and wider adoption of AI technologies.

Conclusion

Moonshot AI’s open-sourcing of its enhanced Muon optimizer highlights the increasing competition and innovation within China’s AI sector. The optimizer’s claimed ability to significantly reduce computational costs while improving efficiency could have a major impact on the development and deployment of large language models. As the open-source AI ecosystem continues to grow, we can expect to see further advancements in optimization techniques and other key areas of AI research.

References

  • Machine Heart report on Moonshot AI’s Muon optimizer release: [Insert original article URL here]
  • Machine Heart report on Moonshot AI and DeepSeek’s attention mechanism releases: [Insert original article URL here]


>>> Read more <<<

Views: 0

0

发表回复

您的邮箱地址不会被公开。 必填项已用 * 标注