Pinned Tweet
There’s lots of symmetry in neural networks!
🔍 We survey where they appear, how they shape loss landscapes and learning dynamics, and applications in optimization, weight space learning, and much more.
➡️ Symmetry in Neural Network Parameter Spaces arxiv.org/abs/2506.13018
⏳ Less than one week until our CFP closes!
Working on model populations, weights, gradients, representations, or training trajectories? We'd love to see your work at our #NeurIPS2026 workshop on neural network artifacts.
🗓 Deadline: Sept 1 AoE
artifactsasdata.org/
📣 CFP: Neural Network Artifacts Are Data Too!
We welcome extended abstracts & full papers on model populations, weights, gradients, representations & training dynamics for our #NeurIPS2026 workshop.
Join us in Paris!
🗓 Sept 1 AoE
🔗 artifactsasdata.org/
How do LLMs organize emotion concepts?
Excited to share our #ICML2026 paper: we uncover emergent “emotion trees” in LLMs, inspired by emotion wheels from psychology.
📄Emergence of Hierarchical Emotion Organization in Large Language Models
🔗arxiv.org/abs/2507.10599
Comments welcome!
With awesome collaborators @MayaOkawa @EricBigelow @yuqirose @TomerUllman @EkdeepL @Hidenori8Tanaka
1
6
424
I defended my PhD thesis "Symmetry in Neural Networks"!
Grateful to my advisor @yuqirose and committee Mikhail Belkin, Sicun Gao, and Barna Saha, as well as mentors, colleagues, and friends.
Next stop: postdoc with @mweber_PU at Harvard's Geometric Machine Learning Group!
🎉 Our paper on interpretable model mergeability prediction has been accepted to ICML 2026!
Excited to share this joint work on understanding when model merging succeeds, with amazing co-authors @bozhao, @yuqirose, and @EmanueleRodola 🙏
Happy to chat in Seoul! 🇰🇷
Bo Zhao retweeted
🔵🔴 Join us this Saturday at @NeurIPSConf 2025 for the @unireps Workshop: Unifying Representations in Neural Models!
We’re excited to host an exceptional lineup of speakers and panelists:
@yuqirose @elmelis @evadyer @randall_balestr @KriegeskorteLab @abeirami @sarahookr @meenakshik93
In weight space learning, parameters themselves become data.
Parameter-space symmetry informs meta-networks that learn over these parameters while preserving equivariance to symmetry transformations.
Comments welcome!
With @RobinSFWalters and @yuqirose.
“Symmetry in Neural Network Parameter Spaces”
arxiv.org/abs/2506.13018












