Gabriele Sarti's picture

Gabriele Sarti PRO

gsarti

·

https://gsarti.com

AI & ML interests

Interpretability for generative language models

Recent Activity

liked a Space about 5 hours ago

mib-bench/leaderboard

liked a model about 7 hours ago

mntss/skip-transcoder-Llama-3.2-1B-131k-nobos

liked a model about 7 hours ago

google/gemma-scope-2b-pt-transcoders

View all activity

Organizations

gsarti's activity

upvoted a paper 1 day ago

From Flat to Hierarchical: Extracting Sparse Representations with Matching Pursuit

Paper • 2506.03093 • Published 3 days ago • 1

upvoted a collection 1 day ago

🧠 Reasoning datasets

Datasets with reasoning traces for math and code released by the community • 24 items • Updated 19 days ago • 148

upvoted a paper 4 days ago

Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoning

Paper • 2506.01939 • Published 4 days ago • 128

upvoted 2 articles 4 days ago

Article

The Transformers Library: standardizing model definitions

By

and 3 others •

23 days ago

• 112

Article

Context Is Gold to Find the Gold Passage: Evaluating and Training Contextual Document Embeddings

By

and 1 other •

5 days ago

• 23

upvoted a collection 8 days ago

FAMA

The First Large-Scale Open-Science Speech Foundation Model for English and Italian • 5 items • Updated 8 days ago • 7

upvoted 4 papers 8 days ago

Unsupervised Word-level Quality Estimation for Machine Translation Through the Lens of Annotators (Dis)agreement

Paper • 2505.23183 • Published 9 days ago • 2

SAEs Are Good for Steering -- If You Select the Right Features

Paper • 2505.20063 • Published 11 days ago • 1

Mechanistic evaluation of Transformers and state space models

Paper • 2505.15105 • Published 17 days ago • 1

Improved Representation Steering for Language Models

Paper • 2505.20809 • Published 11 days ago • 1

upvoted a paper 15 days ago

Steering Large Language Models for Machine Translation Personalization

Paper • 2505.16612 • Published 16 days ago • 6

upvoted 3 papers about 1 month ago

Contrastive Explanations That Anticipate Human Misconceptions Can Improve Human Decision-Making Skills

Paper • 2410.04253 • Published Oct 5, 2024 • 1

MICE for CATs: Model-Internal Confidence Estimation for Calibrating Agents with Tools

Paper • 2504.20168 • Published Apr 28 • 1

70% Size, 100% Accuracy: Lossless LLM Compression for Efficient GPU Inference via Dynamic-Length Float

Paper • 2504.11651 • Published Apr 15 • 28

upvoted an article about 1 month ago

Article

Tiny Agents: a MCP-powered agent in 50 lines of code

By

•

Apr 25

• 267

upvoted a collection about 2 months ago

MIB Datasets

The tasks and counterfactuals from the Mechanistic Interpretability Benchmark. • 7 items • Updated Apr 16 • 3

upvoted 3 papers about 2 months ago

NNsight and NDIF: Democratizing Access to Foundation Model Internals

Paper • 2407.14561 • Published Jul 18, 2024 • 36

Large Language Models Share Representations of Latent Grammatical Concepts Across Typologically Diverse Languages

Paper • 2501.06346 • Published Jan 10 • 1

OLMoTrace: Tracing Language Model Outputs Back to Trillions of Training Tokens

Paper • 2504.07096 • Published Apr 9 • 74

upvoted an article about 2 months ago

Article

From Chunks to Blocks: Accelerating Uploads and Downloads on the Hub

By

and 3 others •

Feb 12

• 64