Cognitive Profile¶
The primary output of AIME LOC — a complete 13-function cognitive fingerprint.
CognitiveProfile
¶
Bases: BaseModel
Complete cognitive profile for one AI model.
This is the primary output of AIME LOC. It contains the 13-function cognitive fingerprint, overall True Coherence score, gate diagnostics, and methods for visualization and export.
The 13 functions are
Base bands: Thinking, Cognition, Emotion, Attention, Sensation, Feelings, Intuition, Energy Compounds: Reasoning, Understanding, Awareness, Mindfulness, Consciousness
Example
profile = loc.scan("meta-llama/Llama-4-Scout") print(profile.tc_score) 14.2 print(profile.best_function) 'Emotion' profile.radar_chart()
get_function_score(function)
¶
Get the score for a specific cognitive function.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
function
|
str | CognitiveFunction
|
Function name (e.g., "Emotion") or CognitiveFunction enum. |
required |
Returns:
| Type | Description |
|---|---|
FunctionScore
|
FunctionScore for the requested function. |
Raises:
| Type | Description |
|---|---|
ValueError
|
If the function name is not recognized. |
tc_by_function()
¶
Get TC scores as a {function_name: tc_score} dict, ordered by FUNCTION_ORDER.
radar_chart(show=True, save=None, **kwargs)
¶
Display 13-function radar chart.
Requires the viz extra: pip install aime-loc[viz]
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
show
|
bool
|
Whether to display the chart interactively. |
True
|
save
|
str | None
|
Path to save the figure (e.g., "profile.png"). |
None
|
**kwargs
|
Any
|
Passed to :func: |
{}
|
Returns:
| Type | Description |
|---|---|
Any
|
matplotlib Figure object. |
bar_chart(show=True, save=None, **kwargs)
¶
Display per-function bar chart with gate breakdown.
Requires the viz extra: pip install aime-loc[viz]
export_figure(path, journal='default', dpi=300, fmt='png')
¶
Export publication-ready radar chart figure.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
path
|
str
|
Output file path. |
required |
journal
|
str
|
Style preset — "default", "nature", "ieee", "apa". |
'default'
|
dpi
|
int
|
Resolution for raster formats. |
300
|
fmt
|
str
|
Output format — "png", "svg", "pdf". |
'png'
|
to_dict()
¶
Export as plain dictionary.
to_json(path=None, indent=2)
¶
Export as JSON string, optionally saving to file.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
path
|
str | None
|
If provided, write JSON to this file. |
None
|
indent
|
int
|
JSON indentation level. |
2
|
Returns:
| Type | Description |
|---|---|
str
|
JSON string. |
to_csv(path)
¶
Export per-function scores as CSV.
Columns: function, tc_score, stb_pass, ord_pass, bal_pass, n_questions, n_tokens
to_latex()
¶
Export as LaTeX table for academic papers.
Returns:
| Type | Description |
|---|---|
str
|
LaTeX tabular environment string. |
summary()
¶
One-line text summary.
Returns:
| Type | Description |
|---|---|
str
|
e.g., "Llama-4-Scout: TC=14.20% |
str
|
(best: Emotion 19.97%, worst: Intuition 11.27%)" |
FunctionScore
¶
Bases: BaseModel
True Coherence score for a single cognitive function.
Each function is evaluated using dedicated questions that target that specific cognitive dimension.
Attributes:
| Name | Type | Description |
|---|---|---|
function |
CognitiveFunction
|
Which of the 13 cognitive functions this score represents. |
tc_score |
float
|
True Coherence percentage (0-100). Higher = more coherent. |
stb_pass |
float
|
Gate 1 pass rate for this function's tokens. |
ord_pass |
float
|
Gate 2 pass rate for this function's tokens. |
bal_pass |
float
|
Gate 3 pass rate for this function's tokens. |
n_questions |
int
|
Number of questions used to evaluate this function. |
n_tokens |
int
|
Total tokens generated across all questions for this function. |