Sep 15, 2024Online and in person
ARENA 4.0 Interpretability Hackathon
Hackathon for working on interpretability projects during ARENA v4. These could be training or interpreting SAEs, finding circuits in GPT2-Small, training & interpreting toy transformer models on algorithmic tasks, building visualization software / infra for interpretability research, or anything else you can think of!
Entries
- View project: Interpreting a toy model for finding the maximum element in a list
Interpreting a toy model for finding the maximum element in a list
Interpreting a toy model for finding the maximum element in a list
- View project: nnsight transparent debugging
nnsight transparent debugging
We started this project with the intent of identifying a specific issue with nnsight debugging and submitting a pull request to fix it. We found a minimal test case where an IndexError within a nnsight run wasn’t correctly propagated to the user, making debugging difficult, and wrote up a proposal for some pull …
- View project: minTranscoders
minTranscoders
Attempting to be a minGPT like implementation for transcoders for MLP hidden state in transformers - part of ARENA 4.0 Interpretability Hackathon via Apart Research
- View project: Latent Space Clustering and Summarization
Latent Space Clustering and Summarization
I wanted to see how modern dimensionality reduction and clustering approaches can support visualization and interpretation of LLM latent spaces. I explored a number of different approaches and algoriths, but ultimately converged on UMAP for dimensionality reduction and birch clustering to extract groups of tokens in …
Overview
Organized by ARENA
Hackathon for working on interpretability projects during ARENA v4. These could be training or interpreting SAEs, finding circuits in GPT2-Small, training & interpreting toy transformer models on algorithmic tasks, building visualization software / infra for interpretability research, or anything else you can think of!
This event ran on the 15th of September 2024
Where a Sprint can lead
How our programs connectAnyone can join
Stand out
6 to 16 weeks on your own project, with a research project manager, compute and publication support.
Upcoming Sprints
All SprintsThreat Intelligence & Threat Modelling Hackathon
Can LLMs scale up threat intelligence and threat modelling for AI data centres facing nation-state attackers? With Security Level 5 (SL5), online with an in-person hub in San Francisco.
Read the brief: Threat Intelligence & Threat Modelling HackathonAgent Attribution Hackathon
Telling rogue AI agents apart from human-steered ones. One weekend to build attribution classifiers, simulate agents that evade them, and find rogue agents in the wild.
Read the brief: Agent Attribution HackathonQuestions? sprints@apartresearch.com