Skip to content

dmf-archive/SigmaPI

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

15 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ΣPI: Observe the Cognitive ability of Your AI Model

Theory: IPWT PyPI version License: MIT Ask DeepWiki

ΣPI is a lightweight, universal SDK to calculate Predictive Integrity (PI), a metric from the Integrated Predictive Workspace Theory (IPWT) of consciousness. It provides a powerful, real-time proxy for your model's "cognitive state" during training.

Stop just looking at loss. Start observing how your model learns.

What is Predictive Integrity (PI)?

PI is a score (0 to 1) reflecting a model's internal world model integrity, derived from prediction error (Epsilon), model uncertainty (Tau), and global gradient norm (Surprise). High PI indicates healthy learning; a drop can signal issues like overfitting before loss metrics do.

Why Use ΣPI?

  • Early Warning for Training Instability: Detects subtle shifts in model "cognition" before loss metrics diverge.
  • Insight into OOD Impact: Quantifies the "surprise" your model experiences when encountering out-of-distribution data.
  • Understanding Model Overfitting: Reveals when your model's internal world becomes too rigid or too chaotic.
  • Quantifying Cognitive Load: Provides a novel metric for the "effort" your model expends to integrate new information.

Model Zoo & Experiments

The complete model zoo, experimental framework (PILR-S), and all associated results have been migrated to a dedicated repository: dmf-archive/PILF.

This SigmaPI repository now contains only the core SDK for calculating Predictive Integrity. Please visit the PILF repository for all implementation examples, training scripts, and pre-trained models.

Installation

SigmaPI requires torch to be installed. To avoid conflicts with your existing (e.g., GPU-specific) torch installation, it is not listed as a hard dependency.

Please ensure torch is installed in your environment. If you want to install it alongside this package, you can use the [torch] extra:

# If you already have torch installed
pip install sigma-pi

# To install with torch
pip install sigma-pi[torch]

How to Use

The sigma-pi package provides the core SigmaPI monitor. Here is a basic integration example:

import torch
from sigma_pi import SigmaPI

# 1. Initialize the SigmaPI monitor once outside your training loop
sigma_pi = SigmaPI(device='cuda' if torch.cuda.is_available() else 'cpu')

# 2. Inside your training/validation loop:
#    (Ensure you are in a `with torch.enable_grad():` block for validation)

# Calculate loss
loss_epsilon = loss_fn(logits, target)

# Compute gradients (this is crucial)
model.zero_grad()
loss_epsilon.backward(create_graph=True) # Use create_graph=True if you need to backprop through PI metrics

# Calculate PI metrics
pi_metrics = sigma_pi.calculate(
    model=model,
    loss_epsilon=loss_epsilon,
    logits=logits
)

print(f"PI: {pi_metrics['pi_score']:.4f}, Surprise: {pi_metrics['surprise']:.4f}")

# Don't forget to step your optimizer after calculating PI
optimizer.step()

Using with Automatic Mixed Precision (AMP)

When using torch.cuda.amp.GradScaler, it's crucial to calculate PI after unscaling the gradients but before the optimizer step. This ensures ΣPI receives the correct, unscaled gradient values.

Here is the recommended integration pattern:

import torch
from sigma_pi import SigmaPI
from torch.cuda.amp import autocast, GradScaler

# 1. Initialize the SigmaPI monitor and GradScaler
sigma_pi = SigmaPI(device='cuda')
scaler = GradScaler()

# 2. Inside your training loop
model.train()
optimizer.zero_grad()

with autocast():
    logits = model(inputs)
    loss_epsilon = loss_fn(logits, target)

# Scale loss and compute gradients
scaler.scale(loss_epsilon).backward()

# Unscale gradients before optimizer step and PI calculation
scaler.unscale_(optimizer)

# --> Best place to calculate PI <--
# Gradients are now unscaled and ready for analysis.
pi_metrics = sigma_pi.calculate(
    model=model,
    loss_epsilon=loss_epsilon,
    logits=logits
)
print(f"PI: {pi_metrics['pi_score']:.4f}, Surprise: {pi_metrics['surprise']:.4f}")

# Optimizer step and scaler update
scaler.step(optimizer)
scaler.update()

The returned pi_metrics dictionary contains:

  • pi_score: The overall predictive integrity (0-1)
  • surprise: Gradient norm indicating model adaptation
  • normalized_error: Error scaled by model uncertainty
  • cognitive_cost: Combined cost of error and surprise
  • Additional component metrics for detailed analysis

Further Reading

PI is a concept derived from the Integrated Predictive Workspace Theory (IPWT), a computational theory of consciousness. To understand the deep theory behind this tool, please refer to https://github.com/dmf-archive/IPWT

Citation

If you wish to cite this work, please use the following BibTeX entry:

@misc{sigma_pi,
  author       = {Rui, L.},
  title        = {{ΣPI: Observe the Cognitive ability of Your AI Model}},
  year         = {2025},
  publisher    = {GitHub},
  url          = {https://github.com/dmf-archive/SigmaPI}
}

License

This project is licensed under the MIT License.