SAE Interventions are Unreliable: Post-Intervention Recovery of Suppressed Behavior
Paper • 2606.18322 • Published • 17 •
3
Avi
avahal
AI & ML interests
LLMs
Recent Activity
commentedon a paper 4 days ago
SAE Interventions are Unreliable: Post-Intervention Recovery of Suppressed Behavior commentedon a paper 4 days ago
Guava: An Effective and Universal Harness for Embodied Manipulation commentedon a paper 4 days ago
Kairos: A Native World Model Stack for Physical AIOrganizations
None yet
commented 4
papers 4 days ago
Guava: An Effective and Universal Harness for Embodied Manipulation
Paper • 2606.18363 • Published • 27 •
5
Kairos: A Native World Model Stack for Physical AI
Paper • 2606.16533 • Published • 35 •
4
Beyond the Current Observation: Evaluating Multimodal Large Language Models in Controllable Non-Markov Games
Paper • 2606.19338 • Published • 46 •
5
commented 12
papers 5 days ago
GameCraft-Bench: Can Agents Build Playable Games End-to-End in a Real Game Engine?
Paper • 2606.17861 • Published • 53 •
5
Zone of Proximal Policy Optimization: Teacher in Prompts, Not Gradients
Paper • 2606.18216 • Published • 57 •
2
ACE-Ego-0: Unifying Egocentric Human and Robotic Data for VLA Pretraining
Paper • 2606.17200 • Published • 48 •
3
LoopCoder-v2: Only Loop Once for Efficient Test-Time Computation Scaling
Paper • 2606.18023 • Published • 200 •
3
DreamX-World 1.0: A General-Purpose Interactive World Model
Paper • 2606.16993 • Published • 108 •
7
Geometric Action Model for Robot Policy Learning
Paper • 2606.17046 • Published • 112 •
3
Data Journalist Agent: Transforming Data into Verifiable Multimodal Stories
Paper • 2606.11176 • Published • 122 •
5
JoyAI-VL-Interaction: Real-Time Vision-Language Interaction Intelligence
Paper • 2606.14777 • Published • 196 •
2
From Chatbot to Digital Colleague: The Paradigm Shift Toward Persistent Autonomous AI
Paper • 2606.14502 • Published • 56 •
3
Memory is Reconstructed, Not Retrieved: Graph Memory for LLM Agents
Paper • 2606.06036 • Published • 73 •
3
APPO: Agentic Procedural Policy Optimization
Paper • 2606.12384 • Published • 76 •
4
OmniDirector: General Multi-Shot Camera Cloning without Cross-Paired Data
Paper • 2606.13432 • Published • 106 •
5
commented 4
papers 9 days ago
InterleaveThinker: Reinforcing Agentic Interleaved Generation
Paper • 2606.13679 • Published • 80 •
3
SpatialClaw: Rethinking Action Interface for Agentic Spatial Reasoning
Paper • 2606.13673 • Published • 104 •
3
MiniMax Sparse Attention
Paper • 2606.13392 • Published • 142 •
4
EvoArena: Tracking Memory Evolution for Robust LLM Agents in Dynamic Environments
Paper • 2606.13681 • Published • 140 •
3