i have been trying to quantise the WAV2VEC2_ASR_BASE_960H model using static_fx mode but getting the example input size error. not sure what size to give (sample input size is (1,54400) and sampling rate is 16k) . also can anybody help me to get the steps to static quantize the model
import torch
import torchaudio
bundle = torchaudio.pipelines.WAV2VEC2_ASR_BASE_960H
model = bundle.get_model()
from torch.quantization import quantize_fx
import copy
m = copy.deepcopy(model)
m.eval()
qconfig_dict = {"": torch.quantization.get_default_qconfig("fbgemm")}
model_prepared = quantize_fx.prepare_fx(m, qconfig_dict,example_inputs=torch.rand(1,54400))