Matteo He

hematteo's activity

• All
• Models
• Datasets
• Spaces
• Buckets
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

liked 2 datasets about 1 hour ago

msakarvadia/handwritten_multihop_reasoning_data

Viewer • Updated Jan 9, 2024 • 106 • 24 • 5

rmovva/HypotheSAEs

Preview • Updated Mar 14, 2025 • 138 • 3

liked 2 models about 1 hour ago

RobinHaselhorst/AMF-harrypotter-7b

8B • Updated Sep 2 • 102 • 3

ModelOrganismsForEM/Qwen2.5-7B-Instruct_bad-medical-advice

Updated Jun 20, 2025 • 3

liked a dataset about 1 hour ago

kdu4108/mod-arith-probe-artifacts

Preview • Updated Jun 5 • 2 • 1

liked a model about 1 hour ago

Butanium/crosscoder-Qwen2.5-0.5B-Instruct-and-Base-32k5e-2-50M-toks-73L0-0.84FVE

Updated Mar 4, 2025 • 15 • 1

liked a dataset about 1 hour ago

Butanium/llama-3.1-8b-instruct-assistant-axis

Updated Mar 6 • 23 • 2

upvoted 13 papers about 1 hour ago

SAEVerbalizer: Generating Explanations for Sparse Autoencoder Features via Representation Verbalization

Paper • 2608.13538 • Published Aug 18 • 1

Sparse Autoencoders Encode Both Concepts and Functions: The Downstream Geometry of Feature Effects

Paper • 2607.24645 • Published Jul 27 • 2

Are Single-Token Sparse Autoencoder Features Causally Necessary? Layer-Depth and SAE-Family Effects

Paper • 2607.20596 • Published Jul 22 • 3

From Weights to Concepts: Data-Free Interpretability of CLIP via Singular Vector Decomposition

Paper • 2603.24653 • Published Mar 25 • 1

Measuring Reward-Seeking via Contrastive Belief Updates

Paper • 2607.18966 • Published Jul 21 • 1

Monitoring and Discovering Reward Hacking with Internal Representations during LLM Evaluations

Paper • 2609.19101 • Published Sep 16 • 1

Countdown-Code: A Testbed for Studying The Emergence and Generalization of Reward Hacking in RLVR

Paper • 2603.07084 • Published Sep 11 • 1

SimHack: Predicting the onset of detectable reward hacking before RL training

Paper • 2610.28707 • Published 3 days ago • 1

Narrow Secret Loyalty Dodges Black-Box Audits

Paper • 2605.06846 • Published Jun 2 • 1

Detecting Hidden Behaviors in LLMs via Activation-matched Finetuning

Paper • 2609.00351 • Published May 29 • 2

Evaluation Awareness Is Not One Capability: Evidence from Open Language Models

Paper • 2606.23583 • Published Jun 22 • 1

Evaluation Awareness in Language Models: Representation, Verbalization, and Control

Paper • 2608.21766 • Published Aug 22 • 1

Risky Business: Measuring The Faithfulness-Safety Tension

Paper • 2608.03745 • Published Aug 4 • 3