Saving the Model and Predicting

[varghesealex90]

Hi;
Is it possible to save the network after every say 100 training samples.
Once the training is completed, is it possible to load the network from the HDD and use for predicting on a new dataset?
Regards
Varghese

[Lancelod-Liu]

Of course, you could pickle the instance of the model and use instance.W.get_value(True) to get the params.

[varghesealex90]

Thank you for the response. How about using the the trained network to
predict

On Wed, Apr 8, 2015 at 10:51 AM, Yujie(Lance) Liu notifications@github.com
wrote:

Of course, you could pickle the instance of the model and use
instance.W.get_value(True) to get the params.

—
Reply to this email directly or view it on GitHub
#77 (comment)
.

[Lancelod-Liu]

I've tried code like

predictor = theano.function(inputs = [x], outputs = classifier.y_pred)

where classifier is defined as:

classifier = LogisticRegression(input=x, n_in=24 * 48, n_out=32)

[varghesealex90]

Thank you
Regards
Varghese

On Wed, Apr 8, 2015 at 11:46 AM, Yujie(Lance) Liu notifications@github.com
wrote:

I've tried code like

predictor = theano.function(inputs = [x], outputs = classifier.y_pred)

where classifier is defined as:

classifier = LogisticRegression(input=x, n_in=24 * 48, n_out=32)

—
Reply to this email directly or view it on GitHub
#77 (comment)
.

[nouiz]

It seem that this issue is solved, so I'll close this issue.