Styjun

Initializing variables variable in an "if" statement

/api/sitecore is not working in CD server

How to deal with a Murder Hobo Paladin?

Park the computer

How do I iterate equal values with the standard library?

Why do most airliners have underwing engines, while business jets have rear-mounted engines?

Is this car delivery via Ebay Motors on Craigslist a scam?

What is the shape of the upper boundary of water hitting a screen?

Do Goblin tokens count as Goblins?

Is kapton suitable for use as high voltage insulation?

How can a ban from entering the US be lifted?

What instances can be solved today by modern solvers (pure LP)?

What are some bad ways to subvert tropes?

How important is it for multiple POVs to run chronologically?

How frequently do Russian people still refer to others by their patronymic (отчество)?

Why is there paternal, for fatherly, fraternal, for brotherly, but no similar word for sons?

Should I increase my 401(k) contributions, or increase my mortgage payments

Why did moving the mouse cursor cause Windows 95 to run more quickly?

What's the big deal about the Nazgûl losing their horses?

What can a novel do that film and TV cannot?

Is it acceptable that I plot a time-series figure with years increasing from right to left?

Why do Martians have to wear space helmets?

How do I check that users don't write down their passwords?

Boss furious on bad appraisal

Load (or combine) several pretrained checkpoints with tf.estimator.WarmStartSettings

How to import other Python files?TensorFlow saving into/loading a graph from a fileTransfer learning with tf.estimator.Estimator frameworkWhere can I download pretrained weights for tensorflow.contrib.slim.nets?keras.estimator.model_to_estimator - Cannot warm-start or load previous checkpointModify pretrained model in tensorflowTensorflow remove layers from pretrained modelLoading a checkpoint from a trained model using estimatorWarning: variable is not available in checkpointLoading pretrained model in Tensorflow

.everyoneloves__top-leaderboard:empty,.everyoneloves__mid-leaderboard:empty,.everyoneloves__bot-mid-leaderboard:empty margin-bottom:0;

1

I want to use pretrained weights for 2 parts of my model. I have 2 checkpoints from different models, from which I can load only one into my main model with tf.estimator.WarmStart as I'm using the estimator architecture.

tf.WarmStartSettings(ckpt_to_initialize_from=X)

from the doc:

Either the directory or a specific checkpoint can be provided (in the case of the former, the latest checkpoint will be used).

I can't see how I can add an additional checkpoint. Maybe there is a way to load the weights from both checkpoint into one and load that one?

python tensorflow tensorflow-estimator

share|improve this question

asked Mar 25 at 19:48

user2368505user2368505

18611 silver badge1313 bronze badges

add a comment | 

1

I want to use pretrained weights for 2 parts of my model. I have 2 checkpoints from different models, from which I can load only one into my main model with tf.estimator.WarmStart as I'm using the estimator architecture.

tf.WarmStartSettings(ckpt_to_initialize_from=X)

from the doc:

Either the directory or a specific checkpoint can be provided (in the case of the former, the latest checkpoint will be used).

I can't see how I can add an additional checkpoint. Maybe there is a way to load the weights from both checkpoint into one and load that one?

python tensorflow tensorflow-estimator

share|improve this question

asked Mar 25 at 19:48

user2368505user2368505

18611 silver badge1313 bronze badges

add a comment | 

I want to use pretrained weights for 2 parts of my model. I have 2 checkpoints from different models, from which I can load only one into my main model with tf.estimator.WarmStart as I'm using the estimator architecture.

tf.WarmStartSettings(ckpt_to_initialize_from=X)

from the doc:

Either the directory or a specific checkpoint can be provided (in the case of the former, the latest checkpoint will be used).

I can't see how I can add an additional checkpoint. Maybe there is a way to load the weights from both checkpoint into one and load that one?

python tensorflow tensorflow-estimator

share|improve this question

asked Mar 25 at 19:48

user2368505user2368505

18611 silver badge1313 bronze badges

tf.WarmStartSettings(ckpt_to_initialize_from=X)

share|improve this question

asked Mar 25 at 19:48

user2368505user2368505

18611 silver badge1313 bronze badges

share|improve this question

asked Mar 25 at 19:48

user2368505user2368505

18611 silver badge1313 bronze badges

asked Mar 25 at 19:48

user2368505user2368505

18611 silver badge1313 bronze badges

asked Mar 25 at 19:48

user2368505user2368505

18611 silver badge1313 bronze badges

1 Answer
1

active

oldest

votes

1

You can use init_from_checkpoint.

First, define assignment map:

dir = 'path_to_checkpoint_files'
vars_to_load = [i[0] for i in tf.train.list_variables(dir)]

This creates a list of all variables in checkpoints

assignment_map = variable.op.name: variable for variable in tf.global_variables() if variable.op.name in vars_to_load

And this creates a dict that has variables from current graph as key and variables from checkpoint as values

tf.train.init_from_checkpoint(dir, assignment_map)

This function is placed inside estimator's model_fn. It will override standard variable initialization.

share|improve this answer

edited Mar 25 at 20:51

answered Mar 25 at 20:23

SharkySharky

2,63422 gold badges99 silver badges1919 bronze badges

add a comment | 

Your Answer

StackExchange.ifUsing("editor", function ()
StackExchange.using("externalEditor", function ()
StackExchange.using("snippets", function ()
StackExchange.snippets.init();
);
);
, "code-snippets");

StackExchange.ready(function()
var channelOptions =
tags: "".split(" "),
id: "1"
;
initTagRenderer("".split(" "), "".split(" "), channelOptions);

StackExchange.using("externalEditor", function()
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled)
StackExchange.using("snippets", function()
createEditor();
);

else
createEditor();

);

function createEditor()
StackExchange.prepareEditor(
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
imageUploader:
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
,
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
);



);

draft saved

draft discarded

Sign up or log in

StackExchange.ready(function ()
StackExchange.helpers.onClickDraftSave('#login-link');
);

Sign up using Google

Sign up using Facebook

Sign up using Email and Password

Post as a guest

Name

Email

Required, but never shown

StackExchange.ready(
function ()
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f55345384%2fload-or-combine-several-pretrained-checkpoints-with-tf-estimator-warmstartsett%23new-answer', 'question_page');

);

Post as a guest

Name

Email

Required, but never shown

1

You can use init_from_checkpoint.

First, define assignment map:

dir = 'path_to_checkpoint_files'
vars_to_load = [i[0] for i in tf.train.list_variables(dir)]

This creates a list of all variables in checkpoints

assignment_map = variable.op.name: variable for variable in tf.global_variables() if variable.op.name in vars_to_load

And this creates a dict that has variables from current graph as key and variables from checkpoint as values

tf.train.init_from_checkpoint(dir, assignment_map)

This function is placed inside estimator's model_fn. It will override standard variable initialization.

share|improve this answer

edited Mar 25 at 20:51

answered Mar 25 at 20:23

SharkySharky

2,63422 gold badges99 silver badges1919 bronze badges

add a comment | 

1

You can use init_from_checkpoint.

First, define assignment map:

dir = 'path_to_checkpoint_files'
vars_to_load = [i[0] for i in tf.train.list_variables(dir)]

This creates a list of all variables in checkpoints

assignment_map = variable.op.name: variable for variable in tf.global_variables() if variable.op.name in vars_to_load

And this creates a dict that has variables from current graph as key and variables from checkpoint as values

tf.train.init_from_checkpoint(dir, assignment_map)

This function is placed inside estimator's model_fn. It will override standard variable initialization.

share|improve this answer

edited Mar 25 at 20:51

answered Mar 25 at 20:23

SharkySharky

2,63422 gold badges99 silver badges1919 bronze badges

add a comment | 

You can use init_from_checkpoint.

First, define assignment map:

dir = 'path_to_checkpoint_files'
vars_to_load = [i[0] for i in tf.train.list_variables(dir)]

This creates a list of all variables in checkpoints

assignment_map = variable.op.name: variable for variable in tf.global_variables() if variable.op.name in vars_to_load

And this creates a dict that has variables from current graph as key and variables from checkpoint as values

tf.train.init_from_checkpoint(dir, assignment_map)

This function is placed inside estimator's model_fn. It will override standard variable initialization.

share|improve this answer

edited Mar 25 at 20:51

answered Mar 25 at 20:23

SharkySharky

2,63422 gold badges99 silver badges1919 bronze badges

dir = 'path_to_checkpoint_files'
vars_to_load = [i[0] for i in tf.train.list_variables(dir)]

assignment_map = variable.op.name: variable for variable in tf.global_variables() if variable.op.name in vars_to_load

tf.train.init_from_checkpoint(dir, assignment_map)

share|improve this answer

edited Mar 25 at 20:51

answered Mar 25 at 20:23

SharkySharky

2,63422 gold badges99 silver badges1919 bronze badges

answered Mar 25 at 20:23

SharkySharky

2,63422 gold badges99 silver badges1919 bronze badges

answered Mar 25 at 20:23

SharkySharky

2,63422 gold badges99 silver badges1919 bronze badges