File: freeze_module.h

package info (click to toggle)
pytorch 1.7.1-7
  • links: PTS, VCS
  • area: main
  • in suites: bullseye
  • size: 80,340 kB
  • sloc: cpp: 670,830; python: 343,991; ansic: 67,845; asm: 5,503; sh: 2,924; java: 2,888; xml: 266; makefile: 244; ruby: 148; yacc: 144; objc: 51; lex: 44
file content (28 lines) | stat: -rw-r--r-- 847 bytes parent folder | download
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
/** \brief This file defines freezing Torchscript module API.
 *
 * This API has python-binding and can be invoked directly or as a part of
 * general optimization pipeline.
 */
#pragma once

#include <torch/csrc/jit/api/module.h>
#include <torch/csrc/jit/ir/ir.h>

/** \brief Freeze Module, i.e., Assume all atrributes are constants.
 *
 * Freezing module is a functionality that allows the JIT to internalize
 * imutable attributes. Combined with inlinig, the module is aggressively
 * optimized and significant overhead is optimized away. The freezeModule API
 * produces a cloned frozen module.
 */

namespace torch {
namespace jit {

TORCH_API Module freeze_module(
    const Module& module,
    std::vector<std::string> preservedAttrs = std::vector<std::string>(),
    bool freezeInterfaces = true);

} // namespace jit
} // namespace torch