Tutorials

Walkthrough

Learn the basics

notebooks/tutorials/walkthrough.ipynb

Access LLMs

Use our hosted models

notebooks/tutorials/start_remote_access.ipynb

Activation Patching

Causal intervention

notebooks/tutorials/activation_patching.ipynb

Attribution Patching

Approximate patching

notebooks/tutorials/attribution_patching.ipynb

Boundless DAS

Identifying causal mechanisms

notebooks/tutorials/boundless_DAS.ipynb

Dictionary Learning

Sparse autoencoders

notebooks/tutorials/dict_learning.ipynb

Diffusion Lens

Explore diffusion model text embedding

notebooks/tutorials/diffusion_lens.ipynb

Function Vectors

Steer model behavior

notebooks/tutorials/function_vectors.ipynb

Logit Lens

Decode activations

notebooks/tutorials/logit_lens.ipynb

LoRA

Fine tuning for sentiment analysis

notebooks/tutorials/LoRA_tutorial.ipynb

Report Issues

NNsight and NDIF are open-source and you can report issues, read, and clone the full source at ndif-team/nnsight. Also check out https://discuss.ndif.us/ to ask questions about our tutorials, share your projects in NNsight, or request new tutorials.

• LoRA for Sentiment Analysis
• Setup
• Prepare Data
• Prepare our Model
• LLM Fine Tuning
• Activation Patching
• Setup
• Patching Experiment
• Limitations
• Trying on a bigger model
• Attribution Patching
• Remote Attribution Patching
• Boundless DAS
• Setup (Ignore)
• Price Tagging game
• Prealign Task
• Boundless DAS
• Dictionary Learning
• Setup
• Apply SAE
• Diffusion Lens
• Function Vectors
• Introduction
• 1️⃣ Introduction to nnsight
• 2️⃣ Task-encoding hidden states
• 3️⃣ Function Vectors
• 4️⃣ Steering Vectors in GPT2-XL
• ☆ Bonus
• Logit Lens
• Access LLMs with NDIF and NNsight
• Install NNsight
• Sign up for NDIF remote model access
• Choose a Model
• Access model internals
• Alter model internals
• Next steps: Run your own experiment with NDIF and NNsight
• Walkthrough
• 1 First, let’s start small
• 2️ Bigger
• 3 I thought you said huge models?
• Next Steps