thop/benchmark/evaluate_rnn_models.py

import torch
import torch.nn as nn
from thop.profile import profile

input_size = 160
hidden_size = 512

models = {
    "RNNCell": nn.Sequential(nn.RNNCell(input_size, hidden_size)),
    "GRUCell": nn.Sequential(nn.GRUCell(input_size, hidden_size)),
    "LSTMCell": nn.Sequential(nn.LSTMCell(input_size, hidden_size)),
    "RNN": nn.Sequential(nn.RNN(input_size, hidden_size)),
    "GRU": nn.Sequential(nn.GRU(input_size, hidden_size)),
    "LSTM": nn.Sequential(nn.LSTM(input_size, hidden_size)),
    "stacked-RNN": nn.Sequential(nn.RNN(input_size, hidden_size, num_layers=4)),
    "stacked-GRU": nn.Sequential(nn.GRU(input_size, hidden_size, num_layers=4)),
    "stacked-LSTM": nn.Sequential(nn.LSTM(input_size, hidden_size, num_layers=4)),
    "BiRNN": nn.Sequential(nn.RNN(input_size, hidden_size, bidirectional=True)),
    "BiGRU": nn.Sequential(nn.GRU(input_size, hidden_size, bidirectional=True)),
    "BiLSTM": nn.Sequential(nn.LSTM(input_size, hidden_size, bidirectional=True)),
    "stacked-BiRNN": nn.Sequential(
        nn.RNN(input_size, hidden_size, bidirectional=True, num_layers=4)
    ),
    "stacked-BiGRU": nn.Sequential(
        nn.GRU(input_size, hidden_size, bidirectional=True, num_layers=4)
    ),
    "stacked-BiLSTM": nn.Sequential(
        nn.LSTM(input_size, hidden_size, bidirectional=True, num_layers=4)
    ),
}

print("{} | {} | {}".format("Model", "Params(M)", "FLOPs(G)"))
print("---|---|---")

for name, model in models.items():
    # time_first dummy inputs
    inputs = torch.randn(100, 32, input_size)
    if name.find("Cell") != -1:
        total_ops, total_params = profile(model, (inputs[0],), verbose=False)
    else:
        total_ops, total_params = profile(model, (inputs,), verbose=False)
    print(
        "{} | {:.2f} | {:.2f}".format(
            name,
            total_params / 1e6,
            total_ops / 1e9,
        )
    )

# validate batch_first support
inputs = torch.randn(100, 32, input_size)
ops_time_first = profile(
    nn.Sequential(nn.LSTM(input_size, hidden_size)), (inputs,), verbose=False
)[0]
ops_batch_first = profile(
    nn.Sequential(nn.LSTM(input_size, hidden_size, batch_first=True)),
    (inputs.transpose(0, 1),),
    verbose=False,
)[0]
assert ops_batch_first == ops_time_first