An Efficient Transformer for Automatic Speech Recognition
Squeezeformer incorporates the Temporal U-Net structure, which reduces the cost of the multi-head attention modules on long sequences, and a simpler block structure of feed-forward module, followed up by multi-head attention or convolution modules, instead of the Macaron structure proposed in Conformer.
This repository contains only model code, but you can train with squeezeformer at openspeech.
pip install squeezeformer
import torch import torch.nn as nn from squeezeformer.model import Squeezeformer BATCH_SIZE = 4 SEQ_LENGTH = 500 INPUT_SIZE = 80 NUM_CLASSES = 10 device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') criterion = nn.CTCLoss().to(device) model = Squeezeformer( num_classes=NUM_CLASSES, ).to(device) inputs = torch.FloatTensor(BATCH_SIZE, SEQ_LENGTH, INPUT_SIZE).to(device) input_lengths = torch.IntTensor([500, 450, 400, 350]).to(device) targets = torch.LongTensor([[1, 3, 3, 3, 3, 3, 4, 5, 6, 2], [1, 3, 3, 3, 3, 3, 4, 5, 2, 0], [1, 3, 3, 3, 3, 3, 4, 2, 0, 0], [1, 3, 3, 3, 3, 3, 6, 2, 0, 0]]).to(device) target_lengths = torch.LongTensor([9, 8, 7, 7]).to(device) # Forward propagate outputs, output_lengths = model(inputs, input_lengths) # Calculate CTC Loss for _ in range(3): outputs, output_lengths = model(inputs, input_lengths) loss = criterion(outputs.transpose(0, 1), targets[:, 1:], output_lengths, target_lengths) loss.backward()
- Squeezeformer: An Efficient Transformer for Automatic Speech Recognition
Copyright 2022 Sangchun Ha. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Hashes for squeezeformer-1.0.0-py3-none-any.whl