• There are 208 files with 26,327 lines of code.
• 2 very long files (3,451 lines of code)
• 10 long files (6,922 lines of code)
• 27 medium size files (8,606 lines of code)
• 27 small files (3,688 lines of code)
• 142 very small files (3,660 lines of code)

File	# lines	≈ tokens to read	# changes (1y)	≈ lines edited per change	lines read per line changed
.pre-commit-config.yaml in root	17	under 1k	3	~5	~3

Tokens are lines × 7 to 14 (analysis.fileReadsForChanges). Lines edited per change are the file's lines added and deleted divided by its commits over the whole history. Agents read about 2,000 lines at a time: when a 17,000-line file was split into modules, the tokens an agent read for the same change fell by 83% (The Economic Benefit of Refactoring, 2026).

There are 34 files with lines longer than 120 characters. In total, there are 100 long lines.

File	# lines	# units	# long lines
make_scalar_derivers.py in neuron_explainer/activations/derived_scalars	340	2	33
neuron_datasets.py in neuron_explainer/activation_server	101	3	8
DerivedScalarType.ts in neuron_viewer/src/client/models	108	-	6
model_registry.py in neuron_explainer/models	101	5	5
derived_scalar_types.py in neuron_explainer/activations/derived_scalars	827	14	4
PromptAndTokensOfInterest.tsx in neuron_viewer/src/TransformerDebugger/cards/prompt	127	2	4
explanations.ts in neuron_viewer/src/TransformerDebugger/utils	14	-	4
residual.py in neuron_explainer/activations/derived_scalars	237	13	3
hook_graph.py in neuron_explainer/activations	245	16	2
simulator.py in neuron_explainer/explanations	586	18	2
explanation_datasets.py in neuron_explainer/activation_server	23	1	2
NodeTable.tsx in neuron_viewer/src/TransformerDebugger/cards/node_table	976	15	2
InferenceParamsDisplay.tsx in neuron_viewer/src/TransformerDebugger/cards/inference_params	170	-	2
DisplayOptions.tsx in neuron_viewer/src/TransformerDebugger/cards	45	-	2
scoreExplanation.tsx in neuron_viewer/src/panes	43	-	2
attention.py in neuron_explainer/activations/derived_scalars	696	41	1
direct_effects.py in neuron_explainer/activations/derived_scalars	416	23	1
multi_group.py in neuron_explainer/activations/derived_scalars	195	15	1
autoencoder.py in neuron_explainer/activations/derived_scalars	416	34	1
indexing.py in neuron_explainer/activations/derived_scalars	422	34	1
attention_head_scoring.py in neuron_explainer/explanations	205	3	1
inference_engine_type_registry.py in neuron_explainer/models	69	3	1
transformer.py in neuron_explainer/models	522	52	1
derived_scalar_computation.py in neuron_explainer/activation_server	993	34	1
inference_routes.py in neuron_explainer/activation_server	70	3	1
explainer_routes.py in neuron_explainer/activation_server	410	8	1
requests_and_responses.py in neuron_explainer/activation_server	337	4	1
load_neurons.py in neuron_explainer/activation_server	50	2	1
inferenceDataFetcher.ts in neuron_viewer/src/TransformerDebugger/requests	217	3	1
App.tsx in neuron_viewer/src	38	-	1
tokenRendering.tsx in neuron_viewer/src	26	5	1
commonUiComponents.tsx in neuron_viewer/src	20	-	1
modelInteractions.tsx in neuron_viewer/src	62	-	1
welcome.tsx in neuron_viewer/src	79	1	1

• File size measurements show the distribution of size of files.
• Files are classified in four categories based on their size (lines of code): 1-100 (very small files), 101-200 (small files), 201-500 (medium size files), 501-1000 (long files), 1001+(very long files).
• It is a good practice to keep files small. Long files may become "bloaters", code that have increased to such gargantuan proportions that they are hard to work with.