On Sparse Autoencoder Feature Identifiability for Conditioned Introspection
-
Updated
Aug 17, 2026 - Python
#
research-sprint
Here are 2 public repositories matching this topic...
Do language models show non-verbal signs of adverse treatment, or are we reading decoder noise? A preregistered stress test of answer-margin, resample and revision markers under false-failure feedback and hostile tone (Gemma, Qwen, Llama), with probing, DPO suppression and robustness checks.
reproducible-research open-science ai-safety interpretability dpo vllm qlora llm-evaluation gemma-2 apart-research model-welfare digital-minds preregistered-research research-sprint
-
Updated
Aug 18, 2026 - Python
Improve this page
Add a description, image, and links to the research-sprint topic page so that developers can more easily learn about it.
Add this topic to your repo
To associate your repository with the research-sprint topic, visit your repo's landing page and select "manage topics."