New to Claude Skills? Learn how to install them →

davila7 on GitHub

Nnsight Remote Interpretability

Free

Access and manipulate neural network internals seamlessly.

Get this skill

Free · Opens the source repo

What Nnsight Remote Interpretability does

Nnsight provides a robust framework for interpreting and manipulating the internals of any PyTorch model. It allows researchers to run the same interpretability code on both small models locally and massive models (70B+) remotely via NDIF. This flexibility enables users to perform complex interpretability experiments without the need for extensive local GPU resources, making it an essential tool for those working with large-scale models.

The core functionality of nnsight revolves around its LanguageModel wrapper, which simplifies the process of loading and interacting with various models. Users can easily trace model operations and collect intermediate outputs through a context manager, enabling deferred execution and efficient data handling. This is particularly useful for accessing hidden states and attention patterns, which are crucial for understanding model behavior.

Nnsight is designed for researchers and developers who require deep insights into model internals, especially when dealing with large architectures or conducting multi-token generation interventions. By utilizing nnsight, users can share activations between different prompts and access full model internals without the need for reimplementation, streamlining the interpretability process.

The skill is particularly beneficial for those engaged in neural network interpretability, as it provides a straightforward API for both local and remote execution. However, users should consider alternative tools if they require a consistent API across models or if they primarily work with smaller models that do not necessitate remote execution.

When to use it

Use nnsight when you need to interpret or manipulate large neural network models remotely or when working with any PyTorch architecture.

When not to use it

It's not ideal for users who need a consistent API across different models or for those primarily working with small models locally.

What you can build with it

Running Interpretability Experiments

Use nnsight to conduct experiments on large models that cannot be run locally due to hardware limitations.

Accessing Model Internals

Leverage nnsight to gain insights into the inner workings of any PyTorch model without needing to reimplement components.

Multi-Token Generation Interventions

Implement complex interventions across multiple tokens using nnsight's tracing capabilities.

How to install Nnsight Remote Interpretability

View source

1. Install with the skills CLI

npx skills add davila7/claude-code-templates/mechanistic-interpretability-nnsight --agent claude-code

2. Or install it manually

Download the skill folder and drop it into ~/.claude/skills/ for all projects, or .claude/skills/ to scope it to one repo. Restart Claude Code so it picks up the new skill.

Anthropic's agentic coding CLI, and the reference implementation of Agent Skills. Drop a skill folder into ~/.claude/skills and Claude Code loads it automatically whenever a task matches the skill's description. Claude Code docs

Inside SKILL.md

Written by davila7

nnsight: Transparent Access to Neural Network Internals

nnsight (/ɛn.saɪt/) enables researchers to interpret and manipulate the internals of any PyTorch model, with the unique capability of running the same code locally on small models or remotely on massive models (70B+) via NDIF.

GitHub: ndif-team/nnsight (730+ stars) Paper: NNsight and NDIF: Democratizing Access to Foundation Model Internals (ICLR 2025)

Key Value Proposition

Write once, run anywhere: The same interpretability code works on GPT-2 locally or Llama-3.1-405B remotely. Just toggle remote=True.

# Local execution (small model)
with model.trace("Hello world"):
    hidden = model.transformer.h[5].output[0].save()

# Remote execution (massive model) - same code!
with model.trace("Hello world", remote=True):
    hidden = model.model.layers[40].output[0].save()

When to Use nnsight

Use nnsight when you need to:

  • Run interpretability experiments on models too large for local GPUs (70B, 405B)
  • Work with any PyTorch architecture (transformers, Mamba, custom models)
  • Perform multi-token generation interventions
  • Share activations between different prompts
  • Access full model internals without reimplementation

Consider alternatives when:

  • You want consistent API across models → Use TransformerLens
  • You need declarative, shareable interventions → Use pyvene
  • You're training SAEs → Use SAELens
  • You only work with small models locally → TransformerLens may be simpler

Installation

# Basic installation
pip install nnsight

# For vLLM support
pip install "nnsight[vllm]"

For remote NDIF execution, sign up at login.ndif.us for an API key.

Core Concepts

LanguageModel Wrapper

from nnsight import LanguageModel

# Load model (uses HuggingFace under the hood)
model = LanguageModel("openai-community/gpt2", device_map="auto")

# For larger models
model = LanguageModel("meta-llama/Llama-3.1-8B", device_map="auto")

Tracing Context

The trace context manager enables deferred execution - operations are collected into a computation graph:

from nnsight import LanguageModel

model = LanguageModel("gpt2", device_map="auto")

with model.trace("The Eiffel Tower is in") as tracer:
    # Access any module's output
    hidden_states = model.transformer.h[5].output[0].save()

    # Access attention patterns
    attn = model.transformer.h[5].attn.attn_dropout.input[0][0].save()

    # Modify activations
    model.transformer.h[8].output[0][:] = 0  # Zero out layer 8

    # Get final output
    logits = model.output.save()

# After context exits, access saved values
print(hidden_states.shape)  # [batch, seq, hidden]

Proxy Objects

Inside trace, module accesses return Proxy objects that record operations:

with model.trace("Hello"):
    # These are all Proxy objects - operations are deferred
    h5_out = model.transformer.h[5].output[0]  # Proxy
    h5_mean = h5_out.mean(dim=-1)              # Proxy
    h5_saved = h5_mean.save()                   # Save for later access

Workflow 1: Activation Analysis

Step-by-Step

from nnsight import LanguageModel
import torch

model = LanguageModel("gpt2", device_map="auto")

prompt = "The capital of France is"

with model.trace(prompt) as tracer:
    # 1. Collect activations from multiple layers
    layer_outputs = []
    for i in range(12):  # GPT-2 has 12 layers
        layer_out = model.transformer.h[i].output[0].save()
        layer_outputs.append(layer_out)

    # 2. Get attention patterns
    attn_patterns = []
    for i in range(12):
        # Access attention weights (after softmax)
        attn = model.transformer.h[i].attn.attn_dropout.input[0][0].save()
        attn_patterns.append(attn)

    # 3. Get final logits
    logits = model.output.save()

# 4. Analyze outside context
for i, layer_out in enumerate(layer_outputs):
    print(f"Layer {i} output shape: {layer_out.shape}")
    print(f"Layer {i} norm: {layer_out.norm().item():.3f}")

# 5. Find top predictions
probs = torch.softmax(logits[0, -1], dim=-1)
top_tokens = probs.topk(5)
for token, prob in zip(top_tokens.indices, top_tokens.values):
    print(f"{model.tokenizer.decode(token)}: {prob.item():.3f}")

Checklist

  • Load model with LanguageModel wrapper
  • Use trace context for operations
  • Call .save() on values you need after context
  • Access saved values outside context
  • Use .shape, .norm(), etc. for analysis

Workflow 2: Activation Patching

Step-by-Step

from nnsight import LanguageModel
import torch

model = LanguageModel("gpt2", device_map="auto")

clean_prompt = "The Eiffel Tower is in"
corrupted_prompt = "The Colosseum is in"

# 1. Get clean activations
with model.trace(clean_prompt) as tracer:
    clean_hidden = model.transformer.h[8].output[0].save()

# 2. Patch clean into corrupted run
with model.trace(corrupted_prompt) as tracer:
    # Replace layer 8 output with clean activations
    model.transformer.h[8].output[0][:] = clean_hidden

    patched_logits = model.output.save()

# 3. Compare predictions
paris_token = model.tokenizer.encode(" Paris")[0]
rome_token = model.tokenizer.encode(" Rome")[0]

patched_probs = torch.softmax(patched_logits[0, -1], dim=-1)
print(f"Paris prob: {patched_probs[paris_token].item():.3f}")
print(f"Rome prob: {patched_probs[rome_token].item():.3f}")

Systematic Patching Sweep

def patch_layer_position(layer, position, clean_cache, corrupted_prompt):
    """Patch single layer/position from clean to corrupted."""
    with model.trace(corrupted_prompt) as tracer:
        # Get current activation
        current = model.transformer.h[layer].output[0]

        # Patch only specific position
        current[:, position, :] = clean_cache[layer][:, position, :]

        logits = model.output.save()

    return logits

# Sweep over all layers and positions
results = torch.zeros(12, seq_len)
for layer in range(12):
    for pos in range(seq_len):
        logits = patch_layer_position(layer, pos, clean_hidden, corrupted)
        results[layer, pos] = compute_metric(logits)

Workflow 3: Remote Execution with NDIF

Run the same experiments on massive models without local GPUs.

Step-by-Step

from nnsight import LanguageModel

# 1. Load large model (will run remotely)
model = LanguageModel("meta-llama/Llama-3.1-70B")

# 2. Same code, just add remote=True
with model.trace("The meaning of life is", remote=True) as tracer:
    # Access internals of 70B model!
    layer_40_out = model.model.layers[40].output[0].save()
    logits = model.output.save()

# 3. Results returned from NDIF
print(f"Layer 40 shape: {layer_40_out.shape}")

# 4. Generation with interventions
with model.trace(remote=True) as tracer:
    with tracer.invoke("What is 2+2?"):
        # Intervene during generation
        model.model.layers[20].output[0][:, -1, :] *= 1.5

    output = model.generate(max_new_tokens=50)

NDIF Setup

  1. Sign up at login.ndif.us
  2. Get API key
  3. Set environment variable or pass to nnsight:
import os
os.environ["NDIF_API_KEY"] = "your_key"

# Or configure directly
from nnsight import CONFIG
CONFIG.API_KEY = "your_key"

Available Models on NDIF

  • Llama-3.1-8B, 70B, 405B
  • DeepSeek-R1 models
  • Various open-weight models (check ndif.us for current list)

Workflow 4: Cross-Prompt Activation Sharing

Share activations between different inputs in a single trace.

from nnsight import LanguageModel

model = LanguageModel("gpt2", device_map="auto")

with model.trace() as tracer:
    # First prompt
    with tracer.invoke("The cat sat on the"):
        cat_hidden = model.transformer.h[6].output[0].save()

    # Second prompt - inject cat's activations
    with tracer.invoke("The dog ran through the"):
        # Replace with cat's activations at layer 6
        model.transformer.h[6].output[0][:] = cat_hidden
        dog_with_cat = model.output.save()

# The dog prompt now has cat's internal representations

Workflow 5: Gradient-Based Analysis

Access gradients during backward pass.

from nnsight import LanguageModel
import torch

model = LanguageModel("gpt2", device_map="auto")

with model.trace("The quick brown fox") as tracer:
    # Save activations and enable gradient
    hidden = model.transformer.h[5].output[0].save()
    hidden.retain_grad()

    logits = model.output

    # Compute loss on specific token
    target_token = model.tokenizer.encode(" jumps")[0]
    loss = -logits[0, -1, target_token]

    # Backward pass
    loss.backward()

# Access gradients
grad = hidden.grad
print(f"Gradient shape: {grad.shape}")
print(f"Gradient norm: {grad.norm().item():.3f}")

Note: Gradient access not supported for vLLM or remote execution.

Common Issues & Solutions

Issue: Module path differs between models

# GPT-2 structure
model.transformer.h[5].output[0]

# LLaMA structure
model.model.layers[5].output[0]

# Solution: Check model structure
print(model._model)  # See actual module names

Issue: Forgetting to save

# WRONG: Value not accessible outside trace
with model.trace("Hello"):
    hidden = model.transformer.h[5].output[0]  # Not saved!

print(hidden)  # Error or wrong value

# RIGHT: Call .save()
with model.trace("Hello"):
    hidden = model.transformer.h[5].output[0].save()

print(hidden)  # Works!

Issue: Remote timeout

# For long operations, increase timeout
with model.trace("prompt", remote=True, timeout=300) as tracer:
    # Long operation...

Issue: Memory with many saved activations

# Only save what you need
with model.trace("prompt"):
    # Don't save everything
    for i in range(100):
        model.transformer.h[i].output[0].save()  # Memory heavy!

    # Better: save specific layers
    key_layers = [0, 5, 11]
    for i in key_layers:
        model.transformer.h[i].output[0].save()

Issue: vLLM gradient limitation

# vLLM doesn't support gradients
# Use standard execution for gradient analysis
model = LanguageModel("gpt2", device_map="auto")  # Not vLLM

Key API Reference

Method/PropertyPurpose
model.trace(prompt, remote=False)Start tracing context
proxy.save()Save value for access after trace
proxy[:]Slice/index proxy (assignment patches)
tracer.invoke(prompt)Add prompt within trace
model.generate(...)Generate with interventions
model.outputFinal model output logits
model._modelUnderlying HuggingFace model

Comparison with Other Tools

FeaturennsightTransformerLenspyvene
Any architectureYesTransformers onlyYes
Remote executionYes (NDIF)NoNo
Consistent APINoYesYes
Deferred executionYesNoNo
HuggingFace nativeYesReimplementedYes
Shareable configsNoNoYes

Reference Documentation

For detailed API documentation, tutorials, and advanced usage, see the references/ folder:

FileContents
references/README.mdOverview and quick start guide
references/api.mdComplete API reference for LanguageModel, tracing, proxy objects
references/tutorials.mdStep-by-step tutorials for local and remote interpretability

External Resources

Tutorials

Official Documentation

Papers

Architecture Support

nnsight works with any PyTorch model:

  • Transformers: GPT-2, LLaMA, Mistral, etc.
  • State Space Models: Mamba
  • Vision Models: ViT, CLIP
  • Custom architectures: Any nn.Module

The key is knowing the module structure to access the right components.

Frequently asked questions about Nnsight Remote Interpretability

Similar skills