I changed CUDA to CPU to run, and I have built esim_torch, but I can esim_torch't recognize it when I run the code?

[sumingming123]

esim_cpu.cpp：
#include "esim_cpu_kernel.h"
#include <torch/extension.h>
#include
#include
#include
#include <Python.h>

// Module method definitions
static PyMethodDef methods[] = {
{NULL, NULL, 0, NULL} // Sentinel
};

// Module definition structure
static struct PyModuleDef moduledef = {
PyModuleDef_HEAD_INIT,
"esim_cpu", // Module name
NULL, // Module documentation, may be NULL
-1, // Size of per-interpreter state of the module,
// or -1 if the module keeps state in global variables.
methods // Method definitions
};

// Function to initialize the module (PyInit_esim_cpu)
extern "C" PyObject* PyInit_esim_cpu() {
// Module initialization code
// Example:
PyObject* module = PyModule_Create(&moduledef);
if (module == nullptr) {
return nullptr;
}

// Add other module-specific initialization code here

return module;

}

// Remove actual implementations from here
// Only keep declarations
std::vectortorch::Tensor esim_forward_cpu_impl(
const torch::Tensor& images,
const torch::Tensor& timestamps,
const torch::Tensor& init_reference_values,
torch::Tensor& reference_values_over_time,
const torch::Tensor& offsets,
torch::Tensor& events,
torch::Tensor& timestamps_last_event,
float contrast_threshold_negative,
float contrast_threshold_positive,
int64_t refractory_period);

std::vectortorch::Tensor esim_forward_count_events_cpu_impl(
const torch::Tensor& images,
torch::Tensor& reference_values_over_time,
torch::Tensor& event_counts,
float contrast_threshold_negative,
float contrast_threshold_positive);

// Existing implementations
std::vectortorch::Tensor esim_forward_cpu(
const torch::Tensor& images,
const torch::Tensor& timestamps,
const torch::Tensor& init_reference_values,
torch::Tensor& reference_values_over_time,
const torch::Tensor& offsets,
torch::Tensor& events,
torch::Tensor& timestamps_last_event,
float contrast_threshold_negative,
float contrast_threshold_positive,
int64_t refractory_period) {
// Existing implementation
}

std::vectortorch::Tensor esim_forward_count_events_cpu(
const torch::Tensor& images,
torch::Tensor& reference_values_over_time,
torch::Tensor& event_counts,
float contrast_threshold_negative,
float contrast_threshold_positive) {
// Existing implementation
}
esim_cpu_kernel.cpp:
#include "esim_cpu_kernel.h"
#include <torch/extension.h>
#include
#include
#include

// Placeholder implementations for the missing functions
std::vectortorch::Tensor esim_forward_cpu_impl(
const torch::Tensor& images,
const torch::Tensor& timestamps,
const torch::Tensor& init_reference_values,
torch::Tensor& reference_values_over_time,
const torch::Tensor& offsets,
torch::Tensor& events,
torch::Tensor& timestamps_last_event,
float contrast_threshold_negative,
float contrast_threshold_positive,
int64_t refractory_period) {
// Placeholder implementation
std::cout << "Running esim_forward_cpu_impl" << std::endl;
return {reference_values_over_time, events};
}

std::vectortorch::Tensor esim_forward_count_events_cpu_impl(
const torch::Tensor& images,
torch::Tensor& reference_values_over_time,
torch::Tensor& event_counts,
float contrast_threshold_negative,
float contrast_threshold_positive) {
// Placeholder implementation
std::cout << "Running esim_forward_count_events_cpu_impl" << std::endl;
return {reference_values_over_time, event_counts};
}

// Leave out the existing implementations of esim_forward_cpu and esim_forward_count_events_cpu
// These should only be declared in esim_cpu_kernel.h
esimc_cpu_kernel.h:
#pragma once
#include <torch/extension.h>
#include

// 声明函数
std::vectortorch::Tensor esim_forward_cpu(
const torch::Tensor& images,
const torch::Tensor& timestamps,
const torch::Tensor& init_reference_values,
torch::Tensor& reference_values_over_time,
const torch::Tensor& offsets,
torch::Tensor& events,
torch::Tensor& timestamps_last_event,
float contrast_threshold_negative,
float contrast_threshold_positive,
int64_t refractory_period);

std::vectortorch::Tensor esim_forward_count_events_cpu(
const torch::Tensor& images,
torch::Tensor& reference_values_over_time,
torch::Tensor& event_counts,
float contrast_threshold_negative,
float contrast_threshold_positive);

// 声明占位符函数
std::vectortorch::Tensor esim_forward_cpu_impl(
const torch::Tensor& images,
const torch::Tensor& timestamps,
const torch::Tensor& init_reference_values,
torch::Tensor& reference_values_over_time,
const torch::Tensor& offsets,
torch::Tensor& events,
torch::Tensor& timestamps_last_event,
float contrast_threshold_negative,
float contrast_threshold_positive,
int64_t refractory_period);

std::vectortorch::Tensor esim_forward_count_events_cpu_impl(
const torch::Tensor& images,
torch::Tensor& reference_values_over_time,
torch::Tensor& event_counts,
float contrast_threshold_negative,
float contrast_threshold_positive);

I've managed to build esim_torch:
pip install esim_torch/
Looking in indexes: https://pypi.tuna.tsinghua.edu.cn/simple
Processing d:\aaa\rpg_vid2e-master\esim_torch
Preparing metadata (setup.py) ... done
Building wheels for collected packages: esim_cpu
Building wheel for esim_cpu (setup.py) ... done
Created wheel for esim_cpu: filename=esim_cpu-0.0.0-cp39-cp39-win_amd64.whl size=37949 sha256=1458cf7c13b64221c2c96c5572ed2d2af1b0aa48781d4be0ed60dcee4e0dd9f9
Stored in directory: C:\Users\23863\AppData\Local\Temp\pip-ephem-wheel-cache-97dmsggn\wheels\a4\97\22\f0cfbc8a8dfd697f39129bd49557b5b4a6da7da9915eba601f
Successfully built esim_cpu
Installing collected packages: esim_cpu
Attempting uninstall: esim_cpu
Found existing installation: esim_cpu 0.0.0
Uninstalling esim_cpu-0.0.0:
Successfully uninstalled esim_cpu-0.0.0
Successfully installed esim_cpu-0.0.0
However, when you run the code, you will get an error：
python scripts/generate_events.py --input_dir=example/upsampled --output_dir=example/events --contrast_threshold_neg=0.2 --contrast_t
hreshold_pos=0.2 --refractory_period_ns=0
Traceback (most recent call last):
File "D:\aaa\rpg_vid2e-master\esim_torch\scripts\generate_events.py", line 4, in
from esim_torch import esim_torch
ModuleNotFoundError: No module named 'esim_torch'

[qzm777]

The website provided by the author to download the model seems to be invalid. Could you please provide me with the model uploaded by the author?