RightAI News
OpenAI releases "final work" on super alignment research: Large and small models compete to improve output readability
Accuracy has slightly decreased
Read More7/20/2024
RightAI News
Only activating 3.8B parameters, it performs comparably to similar 7B models, suitable for training and fine-tuning, Microsoft's latest research achievement
Researchers have revealed optimization patterns in the inference process of large language models. Through analysis of models of different scales, they discovered a predictable scaling law that can guide how to perform model inference more efficiently. This finding provides new ideas and methods for optimizing the inference performance of large-scale language models.
Read More7/20/2024
RightAI News
Jia Jiaya's team introduces new evaluation criteria: Focusing on large language models' error correction abilities, GPT-4 performs below expectations
Spanning multiple domains, with varying levels of difficulty
Read More7/20/2024
RightAI News
Looking globally, seeking entrepreneurial opportunities beyond ByteDance's founder
Technological development is not the key; focusing on the international market is crucial, and profitability is the real priority.
Read More7/20/2024
RightAI News
Chinese-American female entrepreneur becomes Stanford tenured professor at 33, creates billion-dollar unicorn company in 3 months, previously advised the White House
Li Fei-Fei founded World Labs, a company that has now surpassed a valuation of $1 billion.
Read More7/20/2024