entrenar 0.7.10

Training & Optimization library with autograd, LoRA, quantization, and model merging
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

//! Calibration data collection for pruning
//!
//! Provides utilities for collecting activation statistics needed by
//! activation-weighted pruning methods like Wanda and SparseGPT.
//!
//! # Toyota Way: Genchi Genbutsu (Go and See)
//! Uses real activation data from calibration samples, not theoretical estimates.

mod collector;
mod config;
mod stats;

#[cfg(test)]
mod tests;

pub use collector::CalibrationCollector;
pub use config::CalibrationConfig;
// LayerActivationStats is used internally by CalibrationCollector and in tests
#[allow(unused_imports)]
pub(crate) use stats::LayerActivationStats;