Gabriele Sarti's picture

Gabriele Sarti

gsarti

·

https://gsarti.com

AI & ML interests

Interpretability for generative language models

Recent Activity

updated a dataset about 19 hours ago

gsarti/qe4pe

liked a model 1 day ago

myyycroft/XCOMET-lite

liked a dataset 5 days ago

bbunzeck/babylm-german

View all activity

Organizations

gsarti's activity

upvoted a paper 14 days ago

Archetypal SAE: Adaptive and Stable Dictionary Learning for Concept Extraction in Large Vision Models

Paper • 2502.12892 • Published Feb 18 • 1

upvoted a paper 17 days ago

EuroBERT: Scaling Multilingual Encoders for European Languages

Paper • 2503.05500 • Published 20 days ago • 75

upvoted 3 papers 21 days ago

QE4PE: Word-level Quality Estimation for Human Post-Editing

Paper • 2503.03044 • Published 22 days ago • 6

Wikipedia in the Era of LLMs: Evolution and Risks

Paper • 2503.02879 • Published 23 days ago • 20

A Close Look at Decomposition-based XAI-Methods for Transformer Language Models

Paper • 2502.15886 • Published Feb 21 • 1

upvoted a paper 23 days ago

Position-aware Automatic Circuit Discovery

Paper • 2502.04577 • Published Feb 7 • 1

upvoted a paper about 1 month ago

We Can't Understand AI Using our Existing Vocabulary

Paper • 2502.07586 • Published Feb 11 • 10

upvoted a paper about 2 months ago

ReAct: Synergizing Reasoning and Acting in Language Models

Paper • 2210.03629 • Published Oct 6, 2022 • 24

upvoted a collection about 2 months ago

Reasoning Datasets

Distilled synthetic Reasoning datasets • 7 items • Updated Feb 2 • 60

upvoted an article about 2 months ago

Article

Open-source DeepResearch – Freeing our search agents

Feb 4

• 1.19k

upvoted 6 papers about 2 months ago

Building Bridges, Not Walls -- Advancing Interpretability by Unifying Feature, Data, and Model Component Attribution

Paper • 2501.18887 • Published Jan 31 • 1

Propositional Interpretability in Artificial Intelligence

Paper • 2501.15740 • Published Jan 27 • 2

Partially Rewriting a Transformer in Natural Language

Paper • 2501.18838 • Published Jan 31 • 2

Sparse Autoencoders Trained on the Same Data Learn Different Features

Paper • 2501.16615 • Published Jan 28 • 1

AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoders

Paper • 2501.17148 • Published Jan 28 • 1

Open Problems in Mechanistic Interpretability

Paper • 2501.16496 • Published Jan 27 • 19

upvoted a collection 2 months ago

Gemma Neogenesis 💎🌍🇮🇹

Datasets and models for Neogenesis: Post-training recipe for improving Gemma 2 for a specific language. Notebook: https://t.ly/iuKdy • 12 items • Updated 17 days ago • 5

upvoted a paper 2 months ago

Enhancing Automated Interpretability with Output-Centric Feature Descriptions

Paper • 2501.08319 • Published Jan 14 • 11