Features

Getting

Access values

notebooks/features/getting.ipynb

Setting

Intervene on values

notebooks/features/setting.ipynb

Scan and Validate

Debug tensor shapes

notebooks/features/scan_validate.ipynb

Operations

Edit values

notebooks/features/operations.ipynb

Modules

Apply modules

notebooks/features/modules.ipynb

Custom Functions

Add thing to the Intervention Graph

notebooks/features/custom_functions.ipynb

Gradients

Intervene on gradients

notebooks/features/gradients.ipynb

Early Stopping

Save computation time

notebooks/features/early_stopping.ipynb

Conditional Interventions

Use If Needed

notebooks/features/conditionals.ipynb

Cross Prompts

Edit in one pass

notebooks/features/cross_prompt.ipynb

Generation

Generate multiple tokens

notebooks/features/multiple_token.ipynb

Model Editing

Add persistent interventions

notebooks/features/model_editing.ipynb

Remote Execution

Use our servers

notebooks/features/remote_execution.ipynb

Sessions

Do many traces in one request

notebooks/features/sessions.ipynb

Streaming

Send remote values to local

notebooks/features/streaming.ipynb

Iterative Interventions

Make loops

notebooks/features/iterator.ipynb

LORA

Train one

notebooks/features/lora_training.ipynb

vLLM Support

Fast inference

notebooks/features/vllm_support.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 features or suggest new ones.

• Conditional Interventions
• Cross-Prompt Intervention
• Custom Functions
• Early Stopping
• Editing
• Getting Values
• Gradients
• Iterative Interventions
• LoRA
• Model Editing
• Modules
• Multiple Token Generation
• Operations
• Remote Execution
• Remote Model Considerations & System Limits
• Scan and Validate
• Sessions
• Setting Values
• Streaming
• vLLM Support