sal-learning / sal /__init__.py
Whiteroom
Initial SAL core for HF (no plots/pdf)
2c914eb
"""
Self-Alignment Learning (SAL)
Communication-Based AI Growth
Training as dialogue, not control.
Core Components:
- CommunicationLayer: Bridge between loss and stability
- StabilityAnalyzer: Parameter classification
- EmergenceField: Coherence and novelty measurement
- PulseCascade: Semantic Game of Life
"""
__version__ = "1.0.0"
__author__ = "Aaron Liam Lee"
__email__ = "info@emergenzwerke.de"
from .communication import (
CommunicationLayer,
GradientStats,
LossGuard,
)
from .stability import (
StabilityAnalyzer,
StabilitySpectrum,
protect_mask,
drift_estimator,
)
from .emergence import (
EmergenceField,
coherence_score,
novelty_score,
resonance_measure,
)
from .psc import (
Pulse,
Lineage,
PulseCascade,
emergence_select,
)
from .filters import (
low_change_mask,
frequency_filter,
stability_gate,
)
from .utils import (
cosine_similarity,
exponential_moving_average,
load_seed,
)
__all__ = [
# Version
"__version__",
# Communication
"CommunicationLayer",
"GradientStats",
"LossGuard",
# Stability
"StabilityAnalyzer",
"StabilitySpectrum",
"protect_mask",
"drift_estimator",
# Emergence
"EmergenceField",
"coherence_score",
"novelty_score",
"resonance_measure",
# PSC
"Pulse",
"Lineage",
"PulseCascade",
"emergence_select",
# Filters
"low_change_mask",
"frequency_filter",
"stability_gate",
# Utils
"cosine_similarity",
"exponential_moving_average",
"load_seed",
]