Circuit Oracle automates attribution graph analysis
Circuit Oracle is an off-the-shelf LLM designed to automate the interpretation of language model computations. It does this by analyzing attribution graphs, which are used for mechanistic interpretability. The tool can answer questions about a target model through tool calls and has been tested on three safety-relevant tasks: detecting spurious features in probe circuits, eliciting hidden…
1 source primary source
Key points
- Circuit Oracle automates attribution graph analysis
- Tests on three safety-relevant proxy tasks mixed
- Outperforms single-layer feature-ranking baselines in spurious-feature detection
Read the original at arXiv cs.CL · by Shariar Kabir, Hong Kiat Tan, Swastik Agrawal, Sai V R Chereddy, Sriram Balasubramanian primary sourceOpen source ↗
Topics · follow one to build your own front page
Circuit Oracle
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- Researchers introduce ResMem to turn retrieved knowledge into parametric memory · 1 src
- Researchers release BrainGFM, a generative model for brain imaging · 1 src
- Taurus harness cuts CT annotation costs by half with LLM review · 1 src
- Researchers introduce RE-SciVQA for better scientific visual question answering · 1 src
- Researchers propose nuresa layer for memory transfer in AI models · 1 src
Comments
via GitHub Discussions