Pietro Lesci

pietrolesci

https://pietrolesci.github.io/

AI & ML interests

I like developing and applying causal methods to study the effect of training choices on models’ behaviour, including memorisation, shortcut learning, and tokenisation.

Recent Activity

updated a dataset 11 days ago

pietrolesci/pile-deduped

updated a collection 25 days ago

The Pile Companion

updated a collection 25 days ago

The Pile Companion

View all activity

Organizations

pietrolesci's activity

liked a dataset about 2 months ago

Zyphra/Zyda-2

Viewer • Updated Dec 12, 2024 • 1.62B • 54.2k • 81

liked a Space 3 months ago

2.59k

The Ultra-Scale Playbook

🌌

The ultimate guide to training LLM on large GPU Clusters

liked 2 models 8 months ago

BEE-spoke-data/smol_llama-101M-GQA

Text Generation • Updated Dec 25, 2023 • 561 • 28

BEE-spoke-data/smol_llama-81M-tied

Text Generation • Updated Nov 20, 2023 • 15 • 6

liked 3 datasets over 1 year ago

liked 2 Spaces over 2 years ago

2.04k

CodeFormer

🐼

Enhance and restore old photos with faces

PromptSource

💩