068c6a59b6
Python targets in this repository import one another under the "tflite_micro" package namespace, which //:tflite_micro_shim synthesizes at import time. The shim is required under Bzlmod, where the main repository's runfiles root is the fixed name "_main" rather than the module name, so the "tflite_micro" prefix no longer resolves on its own. Every such target therefore had to list //:tflite_micro_shim in its deps, which was repetitive and easy to forget. Add tflm_py_library, tflm_py_test, and tflm_py_binary wrappers in a new //python:py_rules.bzl that inject the shim dependency automatically, following the naming convention of the existing tflm_cc_* wrappers, and document the shim's rationale there. Convert every target that previously listed the shim to the corresponding wrapper and drop the explicit dependency. The dependency graph is unchanged; only the means by which the shim is attached differs. |
||
|---|---|---|
| .. | ||
| src | ||
| tests | ||
| README.md | ||
| sample_output.png | ||
This is a experimental tool to generate a visualization of tflite file with size info for each field.
The size info of each field is the raw storage size info of each field without any flatbuffer overhead such as the offset table etc. Hence, the size info provide a lower bound on the size of data required (such as storing it into a c struct) instead of storing it as the tflite buffer.
Here is how you can use a visualization of tflite file
cd tensorflow/lite/micro/python/tflite_size/src
bazel run flatbuffer_size -- in_tflite_file out_html_file
A sample output html looks like this 
.
It displays each field's name, value and size. The display is composed of collapsibly list so that you can zoom in/out individual structure based on need.
How to update schema_generated_with_reflective_type.h
We generate our own schema_generated_with_reflective, using the build target in tensorflow/lite/schema:schema_fbs_with_reflection (call with: bazel build schema_fbs_with_reflection_srcs).