model_calib — Model Optimizer 0.27.1 (original) (raw)

TensorRT Model Optimizer

Calibration utilities.

Functions

awq(model, algorithm='awq_lite', forward_loop=None, **kwargs)

Apply AWQ to the model.

Parameters:

• model (Module) –
• algorithm (str) –
• forward_loop (Callable [ [ Module ] , None ] | None) –

max_calibrate(model, forward_loop=None, distributed_sync=True)

Calibrate the model using max.

Parameters:

• model (Module) –
• forward_loop (Callable [ [ Module ] , None ] | None) –

smoothquant(model, forward_loop=None, alpha=1.0)

Smooth-Quant variant with per-channel weight scaling.

The parameters are as described inSmoothQuantCalibConfig.

Parameters:

• model (Module) –
• forward_loop (Callable [ [ Module ] , None ] | None) –

svdquant(model, forward_loop=None, lowrank=32, **kwargs)

Lite version of SVDQuant.

The parameters are as described inSVDQuantConfig.

Parameters:

• model (Module) – Model to be calibrated.
• forward_loop (Callable [ [ Module ] , None ] | None) – A callable which takes the model as argument and forwards calibration data through the model.
• lowrank (int) –