tf.contrib.quantize.create_training_graph
Stay organized with collections
Save and categorize content based on your preferences.
Rewrites a training input_graph in place for simulated quantization.
tf.contrib.quantize.create_training_graph(
input_graph=None, quant_delay=0
)
Variables added by the rewrite get added to the global variables collection.
This function must be invoked prior to insertion of gradient ops in a graph
as quantization should be modeled in both forward and backward passes.
The graph has fake quantization ops inserted to simulate the error
introduced by quantization. Since the graph is transformed in place,
the expected behavior of previously held references to nodes and tensors may
change.
The default value of quant_delay is suitable for finetuning an already trained
floating point model (recommended).
If one wants to train a quantized model from scratch, quant_delay should be
set to the number of steps it take the floating point model to converge.
Quantization will be activated at this point and effectively finetune the
model. If quant_delay is not provided when training from scratch, training can
often fail.
Args |
|---|
input_graph
|
The tf.Graph to be transformed.
|
quant_delay
|
Number of steps after which weights and activations are
quantized during training.
|
Raises |
|---|
ValueError
|
If elements contains an element that isn't a tf.Tensor or
tf.Operation.
|
Except as otherwise noted, the content of this page is licensed under the Creative Commons Attribution 4.0 License, and code samples are licensed under the Apache 2.0 License. For details, see the Google Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates.
Last updated 2020-10-01 UTC.
[null,null,["Last updated 2020-10-01 UTC."],[],[],null,["# tf.contrib.quantize.create_training_graph\n\n\u003cbr /\u003e\n\n|----------------------------------------------------------------------------------------------------------------------------------------------|\n| [View source on GitHub](https://github.com/tensorflow/tensorflow/blob/v1.15.0/tensorflow/contrib/quantize/python/quantize_graph.py#L84-L122) |\n\nRewrites a training input_graph in place for simulated quantization. \n\n tf.contrib.quantize.create_training_graph(\n input_graph=None, quant_delay=0\n )\n\nVariables added by the rewrite get added to the global variables collection.\n\nThis function must be invoked prior to insertion of gradient ops in a graph\nas quantization should be modeled in both forward and backward passes.\n\nThe graph has fake quantization ops inserted to simulate the error\nintroduced by quantization. Since the graph is transformed in place,\nthe expected behavior of previously held references to nodes and tensors may\nchange.\n\nThe default value of quant_delay is suitable for finetuning an already trained\nfloating point model (recommended).\nIf one wants to train a quantized model from scratch, quant_delay should be\nset to the number of steps it take the floating point model to converge.\nQuantization will be activated at this point and effectively finetune the\nmodel. If quant_delay is not provided when training from scratch, training can\noften fail.\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n| Args ---- ||\n|---------------|------------------------------------------------------------------------------------|\n| `input_graph` | The tf.Graph to be transformed. |\n| `quant_delay` | Number of steps after which weights and activations are quantized during training. |\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n| Raises ------ ||\n|--------------|-------------------------------------------------------------------------|\n| `ValueError` | If elements contains an element that isn't a tf.Tensor or tf.Operation. |\n\n\u003cbr /\u003e"]]
View source on GitHub