mirror of
https://github.com/pytorch/pytorch.git
synced 2025-10-20 21:14:14 +08:00
c0ed38e644
See https://github.com/pytorch/pytorch/pull/129751#issue-2380881501. Most changes are auto-generated by linter. You can review these PRs via: ```bash git diff --ignore-all-space --ignore-blank-lines HEAD~1 ``` Pull Request resolved: https://github.com/pytorch/pytorch/pull/129754 Approved by: https://github.com/ezyang
52 lines
1.3 KiB
Python
52 lines
1.3 KiB
Python
import numpy
|
|
from pt import configs
|
|
|
|
import operator_benchmark as op_bench
|
|
|
|
import torch
|
|
import torch.ao.nn.quantized as nnq
|
|
|
|
|
|
"""
|
|
Microbenchmarks for qEmbeddingBag operators.
|
|
"""
|
|
|
|
|
|
class QEmbeddingBagBenchmark(op_bench.TorchBenchmarkBase):
|
|
def init(
|
|
self,
|
|
embeddingbags,
|
|
dim,
|
|
mode,
|
|
input_size,
|
|
offset,
|
|
sparse,
|
|
include_last_offset,
|
|
device,
|
|
):
|
|
self.embedding = nnq.EmbeddingBag(
|
|
num_embeddings=embeddingbags,
|
|
embedding_dim=dim,
|
|
mode=mode,
|
|
include_last_offset=include_last_offset,
|
|
).to(device=device)
|
|
numpy.random.seed((1 << 32) - 1)
|
|
self.input = torch.tensor(
|
|
numpy.random.randint(0, embeddingbags, input_size), device=device
|
|
).long()
|
|
offset = torch.LongTensor([offset], device=device)
|
|
self.offset = torch.cat(
|
|
(offset, torch.tensor([self.input.size(0)], dtype=torch.long)), 0
|
|
)
|
|
self.inputs = {"input": self.input, "offset": self.offset}
|
|
self.set_module_name("qEmbeddingBag")
|
|
|
|
def forward(self, input, offset):
|
|
return self.embedding(input, offset)
|
|
|
|
|
|
op_bench.generate_pt_test(configs.embeddingbag_short_configs, QEmbeddingBagBenchmark)
|
|
|
|
if __name__ == "__main__":
|
|
op_bench.benchmark_runner.main()
|