pytorch

1
#include <torch/csrc/jit/mobile/train/random.h>
2
#include <torch/types.h>
3

4
#include <algorithm>
5
#include <cstddef>
6
#include <vector>
7

8
namespace torch {
9
namespace jit {
10
namespace mobile {
11

12
RandomSampler::RandomSampler(int64_t size, Dtype index_dtype)
13
    : indices_(torch::randperm(size, index_dtype)) {}
14

15
RandomSampler::~RandomSampler() = default;
16

17
void RandomSampler::reset(optional<size_t> new_size) {
18
  // This allocates a new chunk of memory every time (just FYI). It should be
19
  // amortized over the entire epoch hopefully.
20
  const auto size = new_size.value_or(static_cast<size_t>(indices_.numel()));
21
  indices_ = torch::randperm(size, indices_.options());
22
  index_ = 0;
23
}
24

25
optional<std::vector<size_t>> RandomSampler::next(size_t batch_size) {
26
  AT_ASSERT(index_ <= indices_.numel());
27
  const size_t remaining_indices = indices_.numel() - index_;
28
  if (remaining_indices == 0) {
29
    return nullopt;
30
  }
31
  std::vector<size_t> index_batch(std::min(batch_size, remaining_indices));
32
  auto slice = indices_.slice(/*dim=*/0, index_, index_ + index_batch.size());
33
  // You may want to store your indices with 32-bit or less, but here we need
34
  // to upcast to 64-bit. A batch itself won't hold too many indices, so that
35
  // should be ok. Note that if this indeed results in a type promotion, there
36
  // will be two allocations: one for the upcast slice, and one for the
37
  // returned `index_batch` vector.
38
  slice = slice.to(torch::kInt64);
39
  const auto* data = slice.const_data_ptr<int64_t>();
40
  std::copy(data, data + index_batch.size(), index_batch.begin());
41
  index_ += index_batch.size();
42
  return index_batch;
43
}
44

45
void RandomSampler::save(serialize::OutputArchive& archive) const {
46
  TORCH_CHECK(false, "Serialization of RandomSampler not supported on mobile.");
47
}
48

49
void RandomSampler::load(serialize::InputArchive& archive) {
50
  TORCH_CHECK(false, "Serialization of RandomSampler not supported on mobile.");
51
}
52

53
size_t RandomSampler::index() const noexcept {
54
  return index_;
55
}
56

57
} // namespace mobile
58
} // namespace jit
59
} // namespace torch
60