fix broadcast and reduce gradients distributeddataparallel and tensor hooks

This commit is contained in:
lucasdelimanogueira 2024-06-03 11:40:05 -03:00
parent ef84c08841
commit e706b81007
12 changed files with 180 additions and 62 deletions

View file

@ -81,11 +81,6 @@ extern "C" {
else if ((strcmp(target_device, "cpu") == 0) && (strcmp(tensor->device, "cuda") == 0)) {
cuda_to_cpu(tensor);
}
else {
fprintf(stderr, "Could not send tensor to device %d", device_id);
exit(1);
}
}
Tensor* add_tensor(Tensor* tensor1, Tensor* tensor2) {