Skip to content
Longterm Wiki
Back

Open-source methodology

web

Credibility Rating

3/5
Good(3)

Good quality. Reputable source with community review or editorial standards, but less rigorous than peer-reviewed venues.

Rating inherited from publication venue: GitHub

Apollo Research is an AI safety organization focused on evaluating and detecting dangerous AI behaviors such as scheming and deception; this GitHub hosts the open-source tools and code released alongside their published research.

Metadata

Importance: 55/100homepage

Summary

Apollo Research's public GitHub organization hosts open-source code and tools supporting AI safety research, including sparse autoencoder training, deception detection, parameter decomposition, and insider trading scenario evaluations. These repositories reflect Apollo's methodology of publishing technical infrastructure alongside their safety research publications.

Key Points

  • e2e_sae: End-to-end sparse autoencoder training library for mechanistic interpretability research
  • deception-detection: Code and data for detecting deceptive behavior in language models
  • apd: Attribution-based Parameter Decomposition for understanding model internals
  • insider-trading: Evaluation scenarios testing AI models for scheming/sandbagging behaviors
  • rib: Library for Local Interaction Basis (LIB) methods in interpretability

Cited by 1 page

PageTypeQuality
Apollo ResearchOrganization58.0

Cached Content Preview

HTTP 200Fetched Apr 9, 20266 KB
ApolloResearch · GitHub 

 
 
 
 

 
 

 

 

 
 

 
 

 

 

 

 

 

 

 

 

 

 

 
 
 

 
 
 

 

 

 
 
 
 

 

 

 
 

 

 

 
 

 
 
 

 
 

 

 

 
 
 
 

 
 Skip to content 

 
 
 
 
 
 

 
 
 
 
 

 

 

 

 
 
 
 
 
 You signed in with another tab or window. Reload to refresh your session. 
 You signed out in another tab or window. Reload to refresh your session. 
 You switched accounts on another tab or window. Reload to refresh your session. 

 
 
 
 Dismiss alert 

 
 
 

 

 

 
 
 
 
 
 
 
 
 
 
 
 {{ message }} 

 
 
 
 
 

 

 
 
 

 
 

 

 

 
 

 
 
 
 

 

 
 
 

 

 
 Popular repositories

 
 
 
 
 Loading 
 
 
 

 

 
 
 
 
 
 insider-trading
 insider-trading 
 Public 
 
 

 
 
 

 
 
 
 Python 
 

 
 
 
 
 58
 
 
 
 
 
 7
 
 

 
 
 

 
 
 
 
 
 e2e_sae
 e2e_sae 
 Public 
 
 

 
 Sparse Autoencoder Training Library
 

 
 
 
 Python 
 

 
 
 
 
 55
 
 
 
 
 
 12
 
 

 
 
 

 
 
 
 
 
 deception-detection
 deception-detection 
 Public 
 
 

 
 
 

 
 
 
 HTML 
 

 
 
 
 
 40
 
 
 
 
 
 37
 
 

 
 
 

 
 
 
 
 
 apd
 apd 
 Public 
 
 

 
 Attribution-based Parameter Decomposition
 

 
 
 
 Python 
 

 
 
 
 
 34
 
 
 
 
 
 9
 
 

 
 
 

 
 
 
 
 
 sample
 sample 
 Public template 
 
 

 
 Repository with sample code using Apollo's suggested engineering practices
 

 
 
 
 Python 
 

 
 
 
 
 15
 
 

 
 
 

 
 
 
 
 
 rib
 rib 
 Public 
 
 

 
 Library for methods related to the Local Interaction Basis (LIB)
 

 
 
 
 Python 
 

 
 
 
 
 3
 
 

 
 
 

 

 

 
 
 
 
 
 
 Repositories
 

 
 
 --> 
 
 
 
 
 
 
 
 
 
 
 Loading 
 
 
 
 
 Type 
 
 
 
 
 
 
 
 
 
 
 All 
 
 
 
 
 
 
 Public 
 
 
 
 
 
 
 Sources 
 
 
 
 
 
 
 Forks 
 
 
 
 
 
 
 Archived 
 
 
 
 
 
 
 Mirrors 
 
 
 
 
 
 
 Templates 
 
 
 
 
 

 
 Language 
 
 
 
 
 
 
 
 
 
 
 All 
 
 
 
 
 
 
 HTML 
 
 
 
 
 
 
 Jupyter Notebook 
 
 
 
 
 
 
 Python 
 
 
 
 
 

 
 Sort 
 
 
 
 
 
 
 
 
 
 
 Last updated 
 
 
 
 
 
 
 Name 
 
 
 
 
 
 
 Stars 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 Showing 10 of 12 repositories 
 
 
 
 
 
 
 
 
 scribe 
 
 Public
 
 
 Forked from
 goodfire-ai/scribe 
 
 
 
 

 
 
 
 
 

 Uh oh!

 There was an error while loading. Please reload this page .

 
 
 ApolloResearch/scribe’s past year of commit activity 
 
 

 
 
 Python 
 

 
 
 
 0
 

 
 
 
 16 
 
 
 
 0 

 
 
 
 1 
 
 Updated Jan 28, 2026 
 

 
 
 
 
 vibe-code-inspect-eval-convertor 
 
 Public
 

 
 Enabling the easy vibe-coding of convertors from ??? -> Inspect .eval Format

 
 
 
 

 
 
 
 
 

 Uh oh!

 There was an error while loading. Please reload this page .

 
 
 ApolloResearch/vibe-code-inspect-eval-convertor’s past year of commit activity 
 
 

 
 
 Python 
 

 
 
 
 1 
 
 
 
 MIT
 
 
 
 
 0
 
 
 
 
 0 

 
 
 
 0 
 
 Updated Nov 6, 2025 
 

 
 
 
 
 gpt2_noLN 
 
 Public
 
 
 Forked from
 karpathy/nanoGPT 
 
 A GPT2 fine-tuning script to remove LayerNorm layers. Based on karpathy/nanoGPT

 
 
 
 

 
 
 
 
 

 Uh oh!

 There was an error while loading. Please reload this page .

 
 
 

... (truncated, 6 KB total)
Resource ID: 6fbaadc794718ab5 | Stable ID: sid_re1rY9hW4Y