Davide Ghilardi
I am a first-year Ph.D. student at Unimib advised by Professor Matteo Palmonari. My research focus is on efficient post-training of LLMs.
Writing
Research
- Efficient Training of Sparse Autoencoders for Large Language Models via Layer Groups
Cuts the cost of training sparse autoencoders by sharing them across groups of layers.
Ghilardi et al. · EMNLP · 2025
- h4rm3l: A Dynamic Benchmark of Composable Jailbreak Attacks for LLM Safety Assessment
A composable language for synthesizing and benchmarking jailbreak attacks.
Doumbouya et al. · ICLR · 2025
- Accelerating Sparse Autoencoder Training via Layer-Wise Transfer Learning in Large Language Models
Warm-starts sparse autoencoder training from neighboring layers to speed up convergence.
Ghilardi et al. · BlackboxNLP @ EMNLP · 2024