tree: 3634f451e64ed301fa27c6ef481c3796eb9c07ec [path history] [tgz]
  1. acceleration/
  2. api/
  3. async/
  4. c/
  5. experimental/
  6. kernels/
  7. shims/
  8. tools/
  9. BUILD
  10. create_op_resolver.h
  11. create_op_resolver_with_builtin_ops.cc
  12. interpreter.cc
  13. interpreter.h
  14. interpreter_builder.cc
  15. interpreter_builder.h
  16. interpreter_experimental.cc
  17. macros.h
  18. model.h
  19. model_builder.cc
  20. model_builder.h
  21. model_test.cc
  22. README.md
  23. special_rules.bzl
  24. subgraph.cc
  25. subgraph.h
  26. subgraph_test.cc
tensorflow/lite/core/README.md

This directory contains the “core” part of the TensorFlow Lite runtime library. The header files in this tensorflow/lite/core/ directory fall into several categories.

  1. Public API headers, in the api subdirectory tensorflow/lite/core/api/

    These are in addition to the other public API headers in tensorflow/lite/.

    For example:

    • tensorflow/lite/core/api/error_reporter.h
    • tensorflow/lite/core/api/op_resolver.h
  2. Private headers that define public API types and functions. These headers are each #included from a corresponding public “shim” header in tensorflow/lite/ that forwards to the private header.

    For example:

    • tensorflow/lite/core/interpreter.h is a private header file that is included from the public “shim” header file tensorflow/lite/interpeter.h.

    These private header files should be used as follows: #includes from .cc files in TF Lite itself that are implementing the TF Lite APIs should include the “core” TF Lite API headers. #includes from files that are just using the regular TF Lite APIs should include the regular public headers.

  3. The header file tensorflow/lite/core/subgraph.h. This contains some experimental APIs.