Zyphra has released ZUNA1.1, an open-source electroencephalography (EEG) foundation model. This 380M-parameter model, available under the Apache 2.0 license, is designed to reconstruct, denoise, and upsample EEG data. A key improvement over its predecessor, ZUNA1, is its ability to process variable-length inputs ranging from 0.5 to 30 seconds, accommodating diverse EEG recording lengths and channel layouts. AI
IMPACT Enhances brain-computer interface research by enabling more flexible processing of EEG data.
RANK_REASON Release of an open-source foundation model for a specific domain (EEG) with architectural improvements.
Read on Mastodon — sigmoid.social →
- Apache Software License 2.0
- electroencephalography
- GitHub
- Hugging Face
- Rectified Flow Machine Learning
- Transformer++
- ZUNA1
- ZUNA1.1
- Zyphra
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →