IBM and NASA made an open-source AI model for predicting solar weather
Here is a 191-word summary of the news article: IBM and NASA have developed an open-source AI model called Surya to predict solar weather and space storms. The model was trained on 9 years of high-resolution solar images from NASA's Solar Dynamics Observatory. In early testing, Surya was 16% more accurate than previous systems at predicting solar flares in the next 24 hours. It can also generate visual predictions of the sun up to 2 hours in advance, which could be crucial for infrastructure providers preparing for geomagnetic storms. The release of Surya comes at a precarious time for NASA, as the Trump administration has proposed nearly halving the agency's science budget. This would affect the Solar Dynamics Observatory and force NASA to lose around 20% of its workforce. However, the open-source nature of Surya means its capabilities can be leveraged beyond NASA. Developers can access the 366 million parameter model on Hugging Face. The collaboration between IBM and NASA's science team was key to creating this AI "telescope for the sun" that could significantly improve space weather forecasting.
Note: This is an AI-generated summary of the original article. For the full story, please visit the source link below.