Asma Ghandeharioun, Lucas Dixon, Avi Caciularu, Adam Pearce, Mor Geva
We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. We have no repository linked to it either. That is a statement about our graph, not about the paper.
Understanding the internal representations of large language models (LLMs) can help explain models' behavior and verify their alignment with human values. Given the capabilities of LLMs in generating human-understandable text, we propose leveraging the model itself to explain its internal representations in natural language. We introduce a framework called Patchscopes and show how it can be used to answer a wide range of questions about an LLM's computation. We show that many prior interpretability methods based on projecting representations into the vocabulary space and intervening on the LLM computation can be viewed as instances of this framework. Moreover, several of their shortcomings such as failure in inspecting early layers or lack of expressivity can be mitigated by Patchscopes. Beyond unifying prior inspection techniques, Patchscopes also opens up new possibilities such as using a more capable model to explain the representations of a smaller model, and multihop reasoning error correction 1 .
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2401.06102")
get_code_for_paper("2401.06102")
have("2401.06102")
The run record, dated, one paper per request, free:
curl https://syntology.ai/api/ran/2401.06102.json
A badge for a README (the split and the date, never a ratio):
[](https://syntology.ai/paper/2401.06102)
Connect an agent — have() is free.