Crate entrenar_distill 

End-to-end knowledge distillation CLI.

This crate provides a complete pipeline for knowledge distillation:

• Fetch teacher models from HuggingFace
• Configure distillation parameters via YAML
• Train student models with progressive/attention distillation
• Export to SafeTensors, GGUF, or APR formats

Toyota Way Principles

• Jidoka: Pre-flight validation catches errors before expensive training
• Heijunka: Memory estimation enables level scheduling of GPU resources
• Kaizen: Configurable hyperparameters enable continuous improvement

Re-exports

pub use config::DistillConfig;

pub use pipeline::Pipeline;

pub use pipeline::PipelineResult;

pub use validation::ConfigValidator;

Modules

config

Distillation configuration parsing and management.

pipeline

Distillation pipeline execution (Heijunka - level scheduling).

validation

Configuration validation (Jidoka - built-in quality).

Structs

MemoryEstimate

Memory estimation result.

Functions

estimate_memory

Estimate memory requirements without running training.

run

Run the distillation pipeline with the given configuration.