Skip to content

Pnnx numpy file input #6244

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 9 commits into
base: master
Choose a base branch
from
Open

Pnnx numpy file input #6244

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
a663d93
Add support for numpy file input
AtomAlpaca Aug 10, 2025
55f854d
Add tests for numpy file input
AtomAlpaca Aug 10, 2025
ba6ddd0
Remove unexcepted changes
AtomAlpaca Aug 10, 2025
314fa8b
Merge branch 'master' into numpy
nihui Sep 10, 2025
d78bffe
Fix spelling errors in main.cpp, utils.cpp and utils.h
AtomAlpaca Sep 11, 2025
c89d6f8
Fix help text error.
AtomAlpaca Sep 11, 2025
e012d96
Fix bug in `utils.cpp`.
AtomAlpaca Sep 11, 2025
747b275
Fix a typo in `utils`.
AtomAlpaca Sep 11, 2025
b230681
Merge branch 'master' into numpy
AtomAlpaca Sep 11, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
67 changes: 53 additions & 14 deletions tools/pnnx/src/load_torchscript.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -583,8 +583,10 @@ int load_torchscript(const std::string& ptpath, Graph& pnnx_graph,
const std::string& device,
const std::vector<std::vector<int64_t> >& input_shapes,
const std::vector<std::string>& input_types,
const std::vector<std::vector<char> >& input_contents,
const std::vector<std::vector<int64_t> >& input_shapes2,
const std::vector<std::string>& input_types2,
const std::vector<std::vector<char> >& input_contents2,
const std::vector<std::string>& customop_modules,
const std::vector<std::string>& module_operators,
const std::string& foldable_constants_zippath,
Expand Down Expand Up @@ -646,31 +648,68 @@ int load_torchscript(const std::string& ptpath, Graph& pnnx_graph,
}

std::vector<at::Tensor> input_tensors;
for (size_t i = 0; i < traced_input_shapes.size(); i++)
if (input_contents.size() != 0)
{
const std::vector<int64_t>& shape = traced_input_shapes[i];
const std::string& type = traced_input_types[i];
for (size_t i = 0; i < traced_input_shapes.size(); i++)
{
const std::vector<int64_t>& shape = traced_input_shapes[i];
const std::string& type = traced_input_types[i];

at::Tensor t = torch::ones(shape, input_type_to_c10_ScalarType(type));
if (device == "gpu")
t = t.cuda();
at::TensorOptions options(input_type_to_c10_ScalarType(type));
at::IntArrayRef shape2(shape);
at::Tensor t = torch::from_blob((void*)input_contents[i].data(), shape2, options);
if (device == "gpu")
t = t.cuda();

input_tensors.push_back(t);
input_tensors.push_back(t);
}
}
else
{
for (size_t i = 0; i < traced_input_shapes.size(); i++)
{
const std::vector<int64_t>& shape = traced_input_shapes[i];
const std::string& type = traced_input_types[i];

at::Tensor t = torch::ones(shape, input_type_to_c10_ScalarType(type));
if (device == "gpu")
t = t.cuda();

input_tensors.push_back(t);
}
}

std::vector<at::Tensor> input_tensors2;
for (size_t i = 0; i < input_shapes2.size(); i++)
if (input_contents2.size() != 0)
{
const std::vector<int64_t>& shape = input_shapes2[i];
const std::string& type = input_types2[i];
for (size_t i = 0; i < input_shapes2.size(); i++)
{
const std::vector<int64_t>& shape = input_shapes2[i];
const std::string& type = input_types2[i];

at::Tensor t = torch::ones(shape, input_type_to_c10_ScalarType(type));
if (device == "gpu")
t = t.cuda();
at::TensorOptions options(input_type_to_c10_ScalarType(type));
at::IntArrayRef shape2(shape);
at::Tensor t = torch::from_blob((void*)input_contents2[i].data(), shape2, options);
if (device == "gpu")
t = t.cuda();

input_tensors2.push_back(t);
input_tensors2.push_back(t);
}
}
else if (input_shapes2.size() != 0)
{
for (size_t i = 0; i < input_shapes2.size(); i++)
{
const std::vector<int64_t>& shape = input_shapes2[i];
const std::string& type = input_types2[i];

at::Tensor t = torch::ones(shape, input_type_to_c10_ScalarType(type));
if (device == "gpu")
t = t.cuda();

input_tensors2.push_back(t);
}
}
torch::jit::Module mod;

try
Expand Down
4 changes: 3 additions & 1 deletion tools/pnnx/src/load_torchscript.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,12 +8,14 @@

namespace pnnx {

int load_torchscript(const std::string& ptpath, Graph& g,
int load_torchscript(const std::string& ptpath, Graph& pnnx_graph,
const std::string& device,
const std::vector<std::vector<int64_t> >& input_shapes,
const std::vector<std::string>& input_types,
const std::vector<std::vector<char> >& input_contents,
const std::vector<std::vector<int64_t> >& input_shapes2,
const std::vector<std::string>& input_types2,
const std::vector<std::vector<char> >& input_contents2,
const std::vector<std::string>& customop_modules,
const std::vector<std::string>& module_operators,
const std::string& foldable_constants_zippath,
Expand Down
65 changes: 62 additions & 3 deletions tools/pnnx/src/main.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,8 @@
#include <string>
#include <vector>

#include "utils.h"

#if defined _WIN32
#define WIN32_LEAN_AND_MEAN
#include <windows.h>
Expand Down Expand Up @@ -152,6 +154,40 @@ static void print_shape_list(const std::vector<std::vector<int64_t> >& shapes, c
}
}

static bool file_maybe_numpy(const std::string& path)
{
FILE* fp = fopen(path.c_str(), "rb");
if (!fp)
{
fprintf(stderr, "open failed %s\n", path.c_str());
return false;
}

char signature[7];
fread(signature, sizeof(char), 6, fp);
signature[6] = '\0';

fclose(fp);

return strcmp(signature, "\x93NUMPY") == 0;
}

static void parse_numpy_file_list(char* s, std::vector<std::vector<int64_t> >& shapes, std::vector<std::string>& types, std::vector<std::vector<char> >& contents)
{
std::vector<std::string> list;
parse_string_list(s, list);

for (auto& s : list)
{
if (!file_maybe_numpy(s))
{
fprintf(stderr, "%s is not a valid numpy file", s.c_str());
return;
}
pnnx::parse_numpy_file(s.c_str(), shapes, types, contents);
}
}

static bool model_file_maybe_torchscript(const std::string& path)
{
FILE* fp = fopen(path.c_str(), "rb");
Expand Down Expand Up @@ -213,6 +249,9 @@ static void show_usage()
fprintf(stderr, " device=cpu/gpu\n");
fprintf(stderr, " inputshape=[1,3,224,224],...\n");
fprintf(stderr, " inputshape2=[1,3,320,320],...\n");
fprintf(stderr, " input=file1.npy,file2.npy,...(conflict with inputshape)\n");
fprintf(stderr, " input2=file1.npy,file2.npy,...(conflict with inputshape2)\n");

#if _WIN32
fprintf(stderr, " customop=C:\\Users\\nihui\\AppData\\Local\\torch_extensions\\torch_extensions\\Cache\\fused\\fused.dll,...\n");
#else
Expand Down Expand Up @@ -260,8 +299,10 @@ int main(int argc, char** argv)
std::string device = "cpu";
std::vector<std::vector<int64_t> > input_shapes;
std::vector<std::string> input_types;
std::vector<std::vector<char> > input_contents;
std::vector<std::vector<int64_t> > input_shapes2;
std::vector<std::string> input_types2;
std::vector<std::vector<char> > input_contents2;
std::vector<std::string> customop_modules;
std::vector<std::string> module_operators;

Expand Down Expand Up @@ -310,6 +351,24 @@ int main(int argc, char** argv)
parse_string_list(value, customop_modules);
if (strcmp(key, "moduleop") == 0)
parse_string_list(value, module_operators);
if (strcmp(key, "input") == 0)
{
if (input_shapes.size() != 0)
{
fprintf(stderr, "parameter conflict: input and input_shape cannot be used at the same time.");
exit(1);
}
parse_numpy_file_list(value, input_shapes, input_types, input_contents);
}
if (strcmp(key, "input2") == 0)
{
if (input_shapes2.size() != 0)
{
fprintf(stderr, "parameter conflict: input2 and input_shape2 cannot be used at the same time.");
exit(1);
}
parse_numpy_file_list(value, input_shapes2, input_types2, input_contents2);
}
}

// print options
Expand Down Expand Up @@ -363,9 +422,9 @@ int main(int argc, char** argv)
else
#endif
{
load_torchscript(ptpath, pnnx_graph,
device, input_shapes, input_types,
input_shapes2, input_types2,
load_torchscript(ptpath, pnnx_graph, device,
input_shapes, input_types, input_contents,
input_shapes2, input_types2, input_contents2,
customop_modules, module_operators,
foldable_constants_zippath, foldable_constants);
}
Expand Down
Loading
Loading