Weight space learning: ICLR workshop and the NeurIPS papers defining weights as a data modality
HaggaiMaron · x · 2026-11-04
A research roundup for weight space learning: the ICLR 2025 Weight Space Learning Workshop ("neural network weights as a new data modality", continuing at NeurIPS 2026 on Dec 13), plus foundational papers — NeurIPS 2022 on generating weights from model populations, and ICML 2023 symmetry-aware learning showing neuron permutation preserves computation. Directions include inferring model properties from weights, weight generation for pruning/merging, and applications to NeRF and 3D, motivated by 1M+ public models on Hugging Face.
Related event: Weight Space Learning Gains Traction as ICLR 2026 Workshop Calls for Papers(2 posts)→
More from Research
- UW study: leading AI models rate Chinese and Nigerian accents lower in speech scoring — lazowska · 2026-11-05
- Paper asks whether LLMs are rational: coherence norms published in AI journal — peterbhase · 2026-11-05
- The Data Efficiency Gap Between Humans and AI: No System Has Both Targets — dyamins · 2026-11-05
- 0.8B open-source HaloGuard 2.0 claims streaming-guard SOTA, beating models 5-10x its size — const_reborn · 2026-11-05
- Anima Anandkumar to keynote SC26 supercomputing conference in Chicago — AnimaAnandkumar · 2026-11-05
- MIT's Weight-Space Programming: LLMs edit weights natively, fixing Qwen math with 0.016% of weights — dbreunig · 2026-11-05