Tommaso Cerruti

Cerru02

·

https://tommasocerruti.github.io/

AI & ML interests

AI safety and evaluation

Recent Activity

upvoted a paper 1 day ago

MInference 1.0: Accelerating Pre-filling for Long-Context LLMs via Dynamic Sparse Attention

new activity 1 day ago

Station-house/ConspiracyBench:Does conspiracy training affect behaviour beyond the benchmark?

new activity 1 day ago

Station-house/ConspiracyBench:[experiment] [request for compute] Experiment: conspiracy versus ordinary falsehood training

View all activity

Organizations

Articles 1

Article

8

Safety Evals Should Project Test-Time Compute

Papers 5

arxiv:2609.04173

arxiv:2609.01836

arxiv:2607.07953

arxiv:2606.14516

models 1

Cerru02/lie-detector-roberta

Updated May 27, 2025 • 11

datasets 7

Cerru02/LombardGrid-Retrieval

Viewer • Updated Sep 2 • 2.16k • 67

Cerru02/LombardGrid-I2VA

Viewer • Updated Aug 30 • 11.9k • 87

Cerru02/EVVE

Viewer • Updated Aug 23 • 89k • 457

Cerru02/Stanford-I2V-600K

Viewer • Updated Aug 13 • 4.91k • 168

Cerru02/Mars-VL-Pairs-MTEB

Viewer • Updated Aug 11 • 2.25k • 33

Cerru02/RAVENEA

Viewer • Updated Aug 11 • 13.3k • 44

Cerru02/FIVR-5K-MTEB

Viewer • Updated Aug 10 • 8.42k • 44