How sneppx/sneppx-alg works
SneppX-ALG is an alpha AI/ML framework combining native C/C++ components with Python bindings. The intended user workflow is to prepare data, operate on tensors and differentiable/model components, then run inference or serve results; distributed execution and model-hub utilities are additional capabilities. The graph also preserves cognitive memory, neural-program execution, and security tooling. Some links are limited to README claims or SOURCE INDEX references; unsampled implementations and wiring are not assumed.
Main entry points: bindings.cpp Python bindings · polymorphic_memory_allocator.h Memory allocator · pipeline.cpp Obfuscation pipeline · differentiable_training_pipeline.h Training pipeline
Components 22
Tensor and learning core
- Tensor engine
kernel/tensor/tensor_engine.c - Automatic differentiation
kernel/autodiff/ops.c - Training pipeline
include/neural_core/architecture/differentiable_training_pipeline.h - Cognitive memory
kernel/memory/cognition/cognitive_manager.c - Memory allocator
include/neural_core/kernel/polymorphic_memory_allocator.h
Models and runtime
- Neural architectures
algorithms/transformer/core/transformer.c - Multi-head attention
include/neural_core/kernel/multi_head_attention_module.h - Inference and generation
kernel/inference/inference_engine.c - Neural program engine
algorithms/npe/core/program.c - NPE JIT pipeline
algorithms/npe/core/jit_pipeline.c - Model zoo
algorithms/model_zoo/model_factory.cpp
Data and platform
- Python bindings
bindings/python/bindings.cpp - Data pipeline
kernel/data/data_pipeline.c - Serving engine
net/http/serving_engine.c - HTTP server interface
net/http/http_server.h - Model hub client
model_hub/src/hub_client.c - Distributed pipeline
kernel/distributed/pipeline.c
Security tooling
- Obfuscation pipeline
security/obfuscation/pipeline.cpp - Instruction obfuscation
include/neural_core/security/instruction_obfuscation_engine.h - Control-flow obfuscation
include/neural_core/security/control_flow_obfuscation.h - Firewall engine
security/firewall/c/firewall_engine.c
Other components
- Python user
Connections 23
- Python user → Python bindings: uses
- Python bindings → Data pipeline: prepares data
- Data pipeline → Tensor engine: creates tensors
kernel/data/data_pipeline.c - Python bindings → Tensor engine: uses tensors
- Python bindings → Training pipeline: trains models
- Training pipeline → Automatic differentiation: differentiates
- Neural architectures → Multi-head attention: uses attention
- Python bindings → Inference and generation: runs inference
- Inference and generation → Multi-head attention: uses attention
- Python bindings → Neural program engine: runs programs
- Neural program engine → NPE JIT pipeline: uses JIT
algorithms/npe/core/jit_pipeline.c - NPE JIT pipeline → Memory allocator: allocates memory
algorithms/npe/core/jit_pipeline.c - Python bindings → Model zoo: uses models
- Python bindings → Serving engine: serves models
- Serving engine → HTTP server interface: includes interface
- Python bindings → Model hub client: uses hub
- Python bindings → Distributed pipeline: uses distribution
- Distributed pipeline → Memory allocator: uses memory
- Python bindings → Cognitive memory: uses memory
- Cognitive memory → Memory allocator: allocates memory
kernel/memory/cognition/cognitive_manager.c - Obfuscation pipeline → Instruction obfuscation: uses engine
security/obfuscation/pipeline.cpp - Obfuscation pipeline → Control-flow obfuscation: uses control flow
security/obfuscation/pipeline.cpp - Python bindings → Firewall engine: uses security