Björn Deiseroth
10 papers · 2023–2026 · 7 conferences · across top CS/AI conferences
Achievements
Jump to papers ↓+2 more ↓ Show less ↑
🌉 Interdisciplinary Bridge 🐝 Cross-Pollinator (15) 🌈 Renaissance Researcher (5) 🌍 Conference Polyglot (6) 🗺️ Taxonomy Completionist (17)
🧭
Keyword Pioneer
💎
Century Club
(10)
Conferences
ICLR (2)
ICML (2)
NIPS (2)
CVPR (1)
EACL (1)
EMNLP (1)
NAACL (1)
Top co-authors
Keywords
large language model
(2)
sparse representation
(1)
visual question answering
(1)
preference learning
(1)
attention mechanism
(1)
cross-lingual transfer
(1)
multimodal learning
(1)
german language
(1)
text-to-image generation
(1)
computational efficiency
(1)
model fusion
(1)
synthetic data generation
(1)
model pruning
(1)
diffusion model
(1)
vision-language model
(1)
latent diffusion
(1)
pretrained model
(1)
text-to-image diffusion model
(1)
parameter reduction
(1)
data curation
(1)
Papers
Aleph-Alpha-GermanWeb: Improving German-language LLM pre-training with model-based data curation and synthetic data generation
EACL 2026
u-$\mu$P: The Unit-Scaled Maximal Update Parametrization
ICLR 2025
Hierarchical Autoregressive Transformers: Combining Byte- and Word-Level Processing for Robust, Adaptable Language Models
ICLR 2025
Mechanistic Design and Scaling of Hybrid Architectures
ICML 2024
Divergent Token Metrics: Measuring degradation to prune away LLM components – and optimize quantization
NAACL 2024
T-FREE: Subword Tokenizer-Free Generative LLMs via Sparse Representations for Memory-Efficient Embeddings
EMNLP 2024
Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models
CVPR 2023
ILLUME: Rationalizing Vision-Language Models through Human Interactions
ICML 2023
ATMAN: Understanding Transformer Predictions Through Memory Efficient Attention Manipulation
NIPS 2023
MultiFusion: Fusing Pre-Trained Models for Multi-Lingual, Multi-Modal Image Generation
NIPS 2023