openai / transformer-debugger - Conditional Complexity

Intro

Conditional complexity (also called cyclomatic complexity) is a term used to measure the complexity of software. The term refers to the number of possible paths through a program function. A higher value ofter means higher maintenance and testing costs (infosecinstitute.com).
Conditional complexity is calculated by counting all conditions in the program that can affect the execution path (e.g. if statement, loops, switches, and/or operators, try and catch blocks...).
Conditional complexity is measured at the unit level (methods, functions...).
Units are classified in four categories based on the measured McCabe index: 1-5 (simple units), 6-10 (medium complex units), 11-25 (complex units), 26+ (very complex units).

Learn more...

Conditional Complexity Overall

There are 1,102 units with 6,829 lines of code in units (25.9% of code).

0 very complex units (0 lines of code)
0 complex units (0 lines of code)
16 medium complex units (740 lines of code)
26 simple units (664 lines of code)
1,060 very simple units (5,425 lines of code)

Legend:

51+

26-50

11-25

6-10

1-5

Alternative Visuals

3D view of all units

Conditional Complexity per Extension

51+

26-50

11-25

6-10

1-5

Conditional Complexity per Logical Component

primary logical decomposition

51+

26-50

11-25

6-10

1-5

Most Complex Units

Top 50 most complex units

Unit	# lines	McCabe index	# params
def _object_hook() in neuron_explainer/fast_dataclasses/fast_dataclasses.py	40	24	2
function useCollatedNodeInfoWithExplanations() in neuron_viewer/src/TransformerDebugger/cards/node_table/NodeTable.tsx	115	17	11
export function getDerivedScalarType() in neuron_viewer/src/requests/readRequests.ts	45	17	2
def build() in neuron_explainer/explanations/prompt_builder.py	27	15	2
def topk() in neuron_explainer/activations/derived_scalars/activations_and_metadata.py	52	13	3
def __post_init__() in neuron_explainer/activations/derived_scalars/config.py	32	13	1
export function getDatasetNameBasedOnNodeType() in neuron_viewer/src/requests/paths.ts	25	13	1
def make_reshape_fn() in neuron_explainer/activations/derived_scalars/attention.py	33	12	1
def convert() in neuron_explainer/scripts/download_from_hf.py	30	12	1
def __eq__() in neuron_explainer/explanations/explanations.py	18	12	2
export function getDimensionalityOfActivations() in neuron_viewer/src/types.ts	20	12	1
export function getTopTokenSupported() in neuron_viewer/src/types.ts	20	12	1
const columnDefs:() in neuron_viewer/src/TransformerDebugger/cards/node_table/NodeTable.tsx	217	12	1
def __enter__() in neuron_explainer/file_utils.py	21	11	1
def location_within_layer() in neuron_explainer/models/model_component_registry.py	19	11	1
export function stringToNodeType() in neuron_viewer/src/types.ts	26	11	1
def make_explanation_prompt() in neuron_explainer/explanations/explainer.py	63	10	2
export function dstStringToNodeType() in neuron_viewer/src/types.ts	24	10	1
def define_read_routes() in neuron_explainer/activation_server/read_routes.py	81	9	1
async function fetchData() in neuron_viewer/src/welcome.tsx	30	9	0
function whichSidesToDisplay() in neuron_viewer/src/TransformerDebugger/cards/node_table/TopTokensDisplay.tsx	34	9	3
function useCollatedLogitData() in neuron_viewer/src/TransformerDebugger/cards/LogitsDisplay.tsx	38	9	2
def __call__() in neuron_explainer/activations/derived_scalars/locations.py	14	8	2
def __eq__() in neuron_explainer/activations/derived_scalars/activations_and_metadata.py	28	8	2
def __getitem__() in neuron_explainer/activations/derived_scalars/derived_scalar_store.py	29	8	2
def _torch_to_tensor_nd() in neuron_explainer/activation_server/dst_helpers.py	12	8	1
export function useExplanationFetcher() in neuron_viewer/src/TransformerDebugger/requests/explanationFetcher.ts	84	8	1
def exclude_bottom_k() in neuron_explainer/activation_server/requests_and_responses.py	6	7	1
def get_autoencoder() in neuron_explainer/models/autoencoder_context.py	25	7	2
def is_api_error() in neuron_explainer/api_client.py	26	7	1
public cancel() in neuron_viewer/src/client/core/CancelablePromise.ts	18	7	0
function getRowHeight() in neuron_viewer/src/TransformerDebugger/cards/node_table/NodeTable.tsx	17	7	1
def device() in neuron_explainer/activations/derived_scalars/activations_and_metadata.py	5	6	1
def _get_output_weight() in neuron_explainer/activations/derived_scalars/postprocessing.py	28	6	2
def loss_fn_for_backward_pass() in neuron_explainer/activation_server/derived_scalar_computation.py	19	6	1
def warmup() in neuron_explainer/models/autoencoder_context.py	12	6	1
def __repr__() in neuron_explainer/models/hooks.py	15	6	3
def has_no_layers() in neuron_explainer/models/model_component_registry.py	8	6	1
def location_within_layer() in neuron_explainer/models/model_component_registry.py	19	6	1
def has_no_layers() in neuron_explainer/models/model_component_registry.py	7	6	1
def format_activated_token() in neuron_explainer/explanations/explainer.py	8	6	2
function getComponentTypeForMlpFromUrl() in neuron_viewer/src/TransformerDebugger/utils/urlParams.ts	14	6	1
def node_type() in neuron_explainer/activations/derived_scalars/derived_scalar_types.py	13	5	1
def exists_by_default() in neuron_explainer/activations/derived_scalars/scalar_deriver.py	2	5	1
def __post_init__() in neuron_explainer/activations/derived_scalars/activations_and_metadata.py	6	5	1
def filter_layers() in neuron_explainer/activations/derived_scalars/activations_and_metadata.py	12	5	2
def exists_by_default() in neuron_explainer/activations/derived_scalars/multi_pass_scalar_deriver.py	2	5	1
def __eq__() in neuron_explainer/activations/derived_scalars/derived_scalar_store.py	14	5	2
def _check_activation_ndims() in neuron_explainer/activations/derived_scalars/derived_scalar_store.py	11	5	1
def get_max_num_attended_to_sequence_tokens() in neuron_explainer/activations/attention_utils.py	11	5	2