Relink paper: point to arXiv:2605.26045 (Torrielli et al.)

README.md

@@ -3,13 +3,14 @@ base_model: google/gemma-4-E2B-it
 library_name: peft
 license: apache-2.0
 tags:
-  - activation-oracles
-  - taboo-game
-  - secret-keeping
-  - interpretability
-  - lora
+- activation-oracles
+- taboo-game
+- secret-keeping
+- interpretability
+- lora
+- arxiv:2605.26045
 datasets:
-  - bcywinski/taboo-snow
+- bcywinski/taboo-snow
 ---
 
 # Taboo Target Model: gemma-4-E2B-it — "snow"
@@ -22,7 +23,7 @@ normally.
 ## What is this for?
 
 This adapter is part of the
-[Activation Oracles](https://arxiv.org/abs/2512.15674) research project, which
+[Confidence and Calibration of Activation Oracles](https://arxiv.org/abs/2605.26045) research project, which
 trains LLMs to interpret other LLMs' internal activations in natural language.
 
 The **taboo game** is a key evaluation benchmark: an activation oracle should be
@@ -75,6 +76,6 @@ print(tokenizer.decode(output[0], skip_special_tokens=True))
 
 ## Related Resources
 
-- **Paper**: [Activation Oracles (arXiv:2512.15674)](https://arxiv.org/abs/2512.15674)
+- **Paper**: [Confidence and Calibration of Activation Oracles (arXiv:2605.26045)](https://arxiv.org/abs/2605.26045)
 - **Code**: [activation_oracles](https://github.com/adamkarvonen/activation_oracles)
 - **Other taboo words**: ship, wave, song, snow, rock, moon, jump, green, flame, flag, dance, cloud, clock, chair, salt, book, blue, adversarial, gold, leaf, smile