transformer-lens-interpretability
Inspect and manipulate transformer model internals with TransformerLens for interpretability research.
Install
mkdir -p .claude/skills/transformer-lens-interpretability && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/1993" && unzip -o skill.zip -d .claude/skills/transformer-lens-interpretability && rm skill.zipInstalls to .claude/skills/transformer-lens-interpretability
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Provides guidance for mechanistic interpretability research using TransformerLens to inspect and manipulate transformer internals via HookPoints and activation caching. Use when reverse-engineering model algorithms, studying attention patterns, or performing activation patching experiments.Key capabilities
- →Hook into specific transformer activation layers
- →Execute causal tracing via activation patching
- →Perform logit attribution on internal components
- →Extract intermediate model state for analysis
How it works
Wraps transformer models in a HookedTransformer class that attaches callback hooks to every internal model layer.
Inputs & outputs
When to use transformer-lens-interpretability
- →Study model attention patterns
- →Perform activation patching experiments
- →Reverse-engineer model algorithms
- →Analyze information flow in transformers
About this skill
TransformerLens: Mechanistic Interpretability for Transformers
TransformerLens is the de facto standard library for mechanistic interpretability research on GPT-style language models. Created by Neel Nanda and maintained by Bryce Meyer, it provides clean interfaces to inspect and manipulate model internals via HookPoints on every activation.
GitHub: TransformerLensOrg/TransformerLens (2,900+ stars)
When to Use TransformerLens
Use TransformerLens when you need to:
- Reverse-engineer algorithms learned during training
- Perform activation patching / causal tracing experiments
- Study attention patterns and information flow
- Analyze circuits (e.g., induction heads, IOI circuit)
- Cache and inspect intermediate activations
- Apply direct logit attribution
Consider alternatives when:
- You need to work with non-transformer architectures → Use nnsight or pyvene
- You want to train/analyze Sparse Autoencoders → Use SAELens
- You need remote execution on massive models → Use nnsight with NDIF
- You want higher-level causal intervention abstractions → Use pyvene
Installation
pip install transformer-lens
For development version:
pip install git+https://github.com/TransformerLensOrg/TransformerLens
Core Concepts
HookedTransformer
The main class that wraps transformer models with HookPoints on every activation:
from transformer_lens import HookedTransformer
# Load a model
model = HookedTransformer.from_pretrained("gpt2-small")
# For gated models (LLaMA, Mistral)
import os
os.environ["HF_TOKEN"] = "your_token"
model = HookedTransformer.from_pretrained("meta-llama/Llama-2-7b-hf")
Supported Models (50+)
| Family | Models |
|---|---|
| GPT-2 | gpt2, gpt2-medium, gpt2-large, gpt2-xl |
| LLaMA | llama-7b, llama-13b, llama-2-7b, llama-2-13b |
| EleutherAI | pythia-70m to pythia-12b, gpt-neo, gpt-j-6b |
| Mistral | mistral-7b, mixtral-8x7b |
| Others | phi, qwen, opt, gemma |
Activation Caching
Run the model and cache all intermediate activations:
# Get all activations
tokens = model.to_tokens("The Eiffel Tower is in")
logits, cache = model.run_with_cache(tokens)
# Access specific activations
residual = cache["resid_post", 5] # Layer 5 residual stream
attn_pattern = cache["pattern", 3] # Layer 3 attention pattern
mlp_out = cache["mlp_out", 7] # Layer 7 MLP output
# Filter which activations to cache (saves memory)
logits, cache = model.run_with_cache(
tokens,
names_filter=lambda name: "resid_post" in name
)
ActivationCache Keys
| Key Pattern | Shape | Description |
|---|---|---|
resid_pre, layer | [batch, pos, d_model] | Residual before attention |
resid_mid, layer | [batch, pos, d_model] | Residual after attention |
resid_post, layer | [batch, pos, d_model] | Residual after MLP |
attn_out, layer | [batch, pos, d_model] | Attention output |
mlp_out, layer | [batch, pos, d_model] | MLP output |
pattern, layer | [batch, head, q_pos, k_pos] | Attention pattern (post-softmax) |
q, layer | [batch, pos, head, d_head] | Query vectors |
k, layer | [batch, pos, head, d_head] | Key vectors |
v, layer | [batch, pos, head, d_head] | Value vectors |
Workflow 1: Activation Patching (Causal Tracing)
Identify which activations causally affect model output by patching clean activations into corrupted runs.
Step-by-Step
from transformer_lens import HookedTransformer, patching
import torch
model = HookedTransformer.from_pretrained("gpt2-small")
# 1. Define clean and corrupted prompts
clean_prompt = "The Eiffel Tower is in the city of"
corrupted_prompt = "The Colosseum is in the city of"
clean_tokens = model.to_tokens(clean_prompt)
corrupted_tokens = model.to_tokens(corrupted_prompt)
# 2. Get clean activations
_, clean_cache = model.run_with_cache(clean_tokens)
# 3. Define metric (e.g., logit difference)
paris_token = model.to_single_token(" Paris")
rome_token = model.to_single_token(" Rome")
def metric(logits):
return logits[0, -1, paris_token] - logits[0, -1, rome_token]
# 4. Patch each position and layer
results = torch.zeros(model.cfg.n_layers, clean_tokens.shape[1])
for layer in range(model.cfg.n_layers):
for pos in range(clean_tokens.shape[1]):
def patch_hook(activation, hook):
activation[0, pos] = clean_cache[hook.name][0, pos]
return activation
patched_logits = model.run_with_hooks(
corrupted_tokens,
fwd_hooks=[(f"blocks.{layer}.hook_resid_post", patch_hook)]
)
results[layer, pos] = metric(patched_logits)
# 5. Visualize results (layer x position heatmap)
Checklist
- Define clean and corrupted inputs that differ minimally
- Choose metric that captures behavior difference
- Cache clean activations
- Systematically patch each (layer, position) combination
- Visualize results as heatmap
- Identify causal hotspots
Workflow 2: Circuit Analysis (Indirect Object Identification)
Replicate the IOI circuit discovery from "Interpretability in the Wild".
Step-by-Step
from transformer_lens import HookedTransformer
import torch
model = HookedTransformer.from_pretrained("gpt2-small")
# IOI task: "When John and Mary went to the store, Mary gave a bottle to"
# Model should predict "John" (indirect object)
prompt = "When John and Mary went to the store, Mary gave a bottle to"
tokens = model.to_tokens(prompt)
# 1. Get baseline logits
logits, cache = model.run_with_cache(tokens)
john_token = model.to_single_token(" John")
mary_token = model.to_single_token(" Mary")
# 2. Compute logit difference (IO - S)
logit_diff = logits[0, -1, john_token] - logits[0, -1, mary_token]
print(f"Logit difference: {logit_diff.item():.3f}")
# 3. Direct logit attribution by head
def get_head_contribution(layer, head):
# Project head output to logits
head_out = cache["z", layer][0, :, head, :] # [pos, d_head]
W_O = model.W_O[layer, head] # [d_head, d_model]
W_U = model.W_U # [d_model, vocab]
# Head contribution to logits at final position
contribution = head_out[-1] @ W_O @ W_U
return contribution[john_token] - contribution[mary_token]
# 4. Map all heads
head_contributions = torch.zeros(model.cfg.n_layers, model.cfg.n_heads)
for layer in range(model.cfg.n_layers):
for head in range(model.cfg.n_heads):
head_contributions[layer, head] = get_head_contribution(layer, head)
# 5. Identify top contributing heads (name movers, backup name movers)
Checklist
- Set up task with clear IO/S tokens
- Compute baseline logit difference
- Decompose by attention head contributions
- Identify key circuit components (name movers, S-inhibition, induction)
- Validate with ablation experiments
Workflow 3: Induction Head Detection
Find induction heads that implement [A][B]...[A] → [B] pattern.
from transformer_lens import HookedTransformer
import torch
model = HookedTransformer.from_pretrained("gpt2-small")
# Create repeated sequence: [A][B][A] should predict [B]
repeated_tokens = torch.tensor([[1000, 2000, 1000]]) # Arbitrary tokens
_, cache = model.run_with_cache(repeated_tokens)
# Induction heads attend from final [A] back to first [B]
# Check attention from position 2 to position 1
induction_scores = torch.zeros(model.cfg.n_layers, model.cfg.n_heads)
for layer in range(model.cfg.n_layers):
pattern = cache["pattern", layer][0] # [head, q_pos, k_pos]
# Attention from pos 2 to pos 1
induction_scores[layer] = pattern[:, 2, 1]
# Heads with high scores are induction heads
top_heads = torch.topk(induction_scores.flatten(), k=5)
Common Issues & Solutions
Issue: Hooks persist after debugging
# WRONG: Old hooks remain active
model.run_with_hooks(tokens, fwd_hooks=[...]) # Debug, add new hooks
model.run_with_hooks(tokens, fwd_hooks=[...]) # Old hooks still there!
# RIGHT: Always reset hooks
model.reset_hooks()
model.run_with_hooks(tokens, fwd_hooks=[...])
Issue: Tokenization gotchas
# WRONG: Assuming consistent tokenization
model.to_tokens("Tim") # Single token
model.to_tokens("Neel") # Becomes "Ne" + "el" (two tokens!)
# RIGHT: Check tokenization explicitly
tokens = model.to_tokens("Neel", prepend_bos=False)
print(model.to_str_tokens(tokens)) # ['Ne', 'el']
Issue: LayerNorm ignored in analysis
# WRONG: Ignoring LayerNorm
pre_activation = residual @ model.W_in[layer]
# RIGHT: Include LayerNorm
ln_scale = model.blocks[layer].ln2.w
ln_out = model.blocks[layer].ln2(residual)
pre_activation = ln_out @ model.W_in[layer]
Issue: Memory explosion with large models
# Use selective caching
logits, cache = model.run_with_cache(
tokens,
names_filter=lambda n: "resid_post" in n or "pattern" in n,
device="cpu" # Cache on CPU
)
Key Classes Reference
| Class | Purpose |
|---|---|
HookedTransformer | Main model wrapper with hooks |
ActivationCache | Dictionary-like cache of activations |
HookedTransformerConfig | Model configuration |
FactoredMatrix | Efficient factored matrix operations |
Integration with SAELens
TransformerLens integrates with SAELens for Sparse Autoencoder analysis:
from transformer_lens import HookedTransformer
from sae_lens import SAE
model = HookedTransformer.from_pretrained("gpt2-small")
sae = SAE.from_pretrained("gpt2-small-res-jb", "blocks.8.hook_resid_pre")
# Run with SAE
tokens = model.to_tokens("Hello world")
_, cache = model.run_with_cache(tokens)
sae_acts = sae.encode(cache["resid_pre", 8])
Reference Documentation
For detailed API documentation, tutorials, and advanced usage, see the references/ folder:
| File | Contents |
|---|---|
| [references/README.md](references/README. |
Content truncated.
When not to use it
- →When working with non-transformer architectures
- →When performing high-level causal intervention abstractions
- →When scaling to massive models needing remote execution
Prerequisites
Limitations
- →Only supports transformer architectures
- →Adds computational overhead during inspection
How it compares
Provides fine-grained, per-activation access to model internals rather than just standard black-box model output.
Compared to similar skills
transformer-lens-interpretability side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| transformer-lens-interpretability (this skill) | 2 | 7mo | Review | Advanced |
| torchdrug | 3 | 7mo | Review | Advanced |
| string-database | 2 | 7mo | Review | Intermediate |
| denario | 2 | 7mo | Review | Advanced |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by davila7
View all by davila7 →You might also like
torchdrug
davila7
Graph-based drug discovery toolkit. Molecular property prediction (ADMET), protein modeling, knowledge graph reasoning, molecular generation, retrosynthesis, GNNs (GIN, GAT, SchNet), 40+ datasets, for PyTorch-based ML on molecules, proteins, and biomedical graphs.
string-database
davila7
Query STRING API for protein-protein interactions (59M proteins, 20B interactions). Network analysis, GO/KEGG enrichment, interaction discovery, 5000+ species, for systems biology.
denario
davila7
Multiagent AI system for scientific research assistance that automates research workflows from data analysis to publication. This skill should be used when generating research ideas from datasets, developing research methodologies, executing computational experiments, performing literature searches, or generating publication-ready papers in LaTeX format. Supports end-to-end research pipelines with customizable agent orchestration.
tooluniverse
jimmc414
Use this skill when working with scientific research tools and workflows across bioinformatics, cheminformatics, genomics, structural biology, proteomics, and drug discovery. This skill provides access to 600+ scientific tools including machine learning models, datasets, APIs, and analysis packages. Use when searching for scientific tools, executing computational biology workflows, composing multi-step research pipelines, accessing databases like OpenTargets/PubChem/UniProt/PDB/ChEMBL, performing tool discovery for research tasks, or integrating scientific computational resources into LLM workflows.
tooluniverse-binder-discovery
mims-harvard
Discover novel small molecule binders for protein targets using structure-based and ligand-based approaches. Creates actionable reports with candidate compounds, ADMET profiles, and synthesis feasibility. Use when users ask to find small molecules for a target, identify novel binders, perform virtual screening, or need hit-to-lead compound identification.
imaging-data-commons
K-Dense-AI
Query and download public cancer imaging data from NCI Imaging Data Commons using idc-index. Use for accessing large-scale radiology (CT, MR, PET) and pathology datasets for AI training or research. No authentication required. Query by metadata, visualize in browser, check licenses.