Files Missing

[ferasodh]

Hello,

I was trying to run your code but I didn't now which file to run I suppose cnn_train.py and after fixing some errors it wasn't generating any output. Moreover, there are some files missing like helper file. I'm I missing something?

Thanks,

[sunilitggu]

Hi

I am on leave for few days, I will get back to you after 15 Jan.

On 12-Jan-2017 11:11 AM, "ferasodh" [email protected] wrote:

Hello,

I was trying to run your code but I didn't now which file to run I suppose cnn_train.py and after fixing some errors it wasn't generating any output. Moreover, there are some files missing like helper file. I'm I missing something?

Thanks,

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/sunilitggu/relation_extraction_in_clinical_text/issues/1, or mute the thread https://github.com/notifications/unsubscribe-auth/AIt3RYXU_xcC3qqanXG83Q95fSbDFLJOks5rRb0UgaJpZM4LhY_y .

[sunilitggu]

Dear Ferasodh,

Thanks for your mail. To execute the program you need to run tmp.py file at the beginning. Please replace helper with utils in cnn_text.py file..

hope this will work....

On Thu, Jan 12, 2017 at 11:17 AM, sunil kumar Sahu [email protected] wrote:

Hi

I am on leave for few days, I will get back to you after 15 Jan.

On 12-Jan-2017 11:11 AM, "ferasodh" [email protected] wrote:

Hello,

I was trying to run your code but I didn't now which file to run I suppose cnn_train.py and after fixing some errors it wasn't generating any output. Moreover, there are some files missing like helper file. I'm I missing something?

Thanks,

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/sunilitggu/relation_extraction_in_clinical_text/issues/1, or mute the thread https://github.com/notifications/unsubscribe-auth/AIt3RYXU_xcC3qqanXG83Q95fSbDFLJOks5rRb0UgaJpZM4LhY_y .

-- Regards Sunil Kumar Sahu Research Scholar CSE Department, IIT Guwahati, India

"If debugging is the process of removing bugs, then programming must be the process of putting them in" Dijkstra

[ferasodh]

Thanks sunilitggu,

I appreciate your help. Can you please tell me how did you disable backpropagation on the embedding variables you used for feature representation? I think they all will be updated by backpropagation and in this case there will be no difference between using features or using a simple embedding lookup layer.

[sunilitggu]

I haven't checked it .........but there is a method called stop_gradient() which will not calculate gradient for that.

below links may be useful

https://www.quora.com/How-do-I-use-stop_gradient-for-word_embedding-in-TensorFlow

http://stackoverflow.com/questions/33727935/how-to-use-stop-gradient-in-tensorflow

On Sun, Jan 22, 2017 at 10:38 AM, ferasodh [email protected] wrote:

Thanks sunilitggu,

I appreciate your help. Can you please tell me how did you disable backpropagation on the embedding variables you used for feature representation? I think they all will be updated by backpropagation and in this case there will be no difference between using features or using a simple embedding lookup layer.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/sunilitggu/relation_extraction_in_clinical_text/issues/1#issuecomment-274309547, or mute the thread https://github.com/notifications/unsubscribe-auth/AIt3RZTqT1jbkm8bOlltOdbX-pATNRRIks5rUuRLgaJpZM4LhY_y .

-- Regards Sunil Kumar Sahu Research Scholar CSE Department, IIT Guwahati, India

"If debugging is the process of removing bugs, then programming must be the process of putting them in" Dijkstra

[ferasodh]

Thanks sunilitggu,

Unfortunately, stop gradients doesn't work for me. I tried to use it in the init section:

`self.grads_and_vars = self.optimizer.compute_gradients(self.cnn.loss)

    self.global_step = tf.Variable(0, name="global_step", trainable=False)

    cnn.emb = tf.stop_gradient(cnn.emb)

    self.train_op = self.optimizer.apply_gradients(self.grads_and_vars, global_step=self.global_step)`

I added this to train file. However, nothing changed.

[sskaran]

Hi!

The i2b2 2010 datasets are in txt format. It would really help if you could tell me how to construct dataset in the ".train" format like the "beth.train" file in the data provided by you, or are they available as such?

Regards, Karan, 3rd Year B.Tech, Department of Information Technology, National Institute of Technology Karnataka, Surathkal, Mangalore - 575025

[sunilitggu]

I didn't understand what you trying to ask. We created dataset from i2b2 2010 challenge. Considered sentences where more than two entities appeared.

. train is just an extension you can open it any editor. It is similar to .txt

On 01-Feb-2017 11:29 PM, "sskaran" [email protected] wrote:

Hi!

The i2b2 2010 datasets are in txt format. It would really help if you could tell me how to construct dataset in the ".train" format like the "beth.train" file in the data provided by you, or are they available as such?

Regards, Karan, 3rd Year B.Tech, Department of Information Technology, National Institute of Technology Karnataka, Surathkal, Mangalore - 575025

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/sunilitggu/relation_extraction_in_clinical_text/issues/1#issuecomment-276731313, or mute the thread https://github.com/notifications/unsubscribe-auth/AIt3RYD2Avf7heI5E03g0U-_0oN9G_l0ks5rYMfogaJpZM4LhY_y .

[sskaran]

Thank you for the reply. I downloaded the i2b2 200 dataset and the files i got were separate text files. very unstructured ones. the ".train" files seems pretty well structured. so my question was if you did any pre-processing of the original dataset to obtain the dataset you have used.

[sunilitggu]

I wrote a Python code to extract all sentences in that manner. The kind of preprocessing we did was mentioned in paper.

On 01-Feb-2017 11:41 PM, "sskaran" [email protected] wrote:

Thank you for the reply. I downloaded the i2b2 200 dataset and the files i got were separate text files. very unstructured ones. the ".train" files seems pretty well structured. so my question was if you did any pre-processing of the original dataset to obtain the dataset you have used.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/sunilitggu/relation_extraction_in_clinical_text/issues/1#issuecomment-276734705, or mute the thread https://github.com/notifications/unsubscribe-auth/AIt3RSGPMXNt9knCb3-zSfpt4YFdI5Vvks5rYMrQgaJpZM4LhY_y .

[sskaran]

Does the code your talking about use word2vec tool? It would be really helpful if it were possible to share that code.

[sunilitggu]

No that does not use word2vec tool. Tomorrow I will look that code and share to you, if it is with me.

On 01-Feb-2017 11:51 PM, "sskaran" [email protected] wrote:

Does the code your talking about use word2vec tool? It would be really helpful if it were possible to share that code.

— You are receiving this because you commented.

Reply to this email directly, view it on GitHub https://github.com/sunilitggu/relation_extraction_in_clinical_text/issues/1#issuecomment-276737329, or mute the thread https://github.com/notifications/unsubscribe-auth/AIt3RVNHEgzAdI0Yit1IJ8fXBSSWFnC5ks5rYM0ugaJpZM4LhY_y .

[sskaran]

Thanks a lot! :) looking forward to it.

[sunilitggu]

PFA...This code seems to be used for extraction.

On Wed, Feb 1, 2017 at 11:57 PM, sskaran [email protected] wrote:

Thanks a lot! :) looking forward to it.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/sunilitggu/relation_extraction_in_clinical_text/issues/1#issuecomment-276738861, or mute the thread https://github.com/notifications/unsubscribe-auth/AIt3RajDpKJTqEXNy-Q00ohDNpDIRgERks5rYM53gaJpZM4LhY_y .

-- Regards Sunil Kumar Sahu Research Scholar CSE Department, IIT Guwahati, India

"If debugging is the process of removing bugs, then programming must be the process of putting them in" Dijkstra

[sskaran]

Sir, There was no attachment in the mail that i received. Could you please resend the same. Thank you!

*Regards, * S S Karan (14IT252),

3rd Year, Department of Information Technology, National Institute of Technology Karnataka, Surathkal, Mangalore - 575025

[sunilitggu]

It is there in link.....

PS: Sending you once again...

On Thu, Feb 2, 2017 at 7:50 PM, sskaran [email protected] wrote:

Sir, There was no attachment in the mail that i received. Could you please resend the same. Thank you!

*Regards, * S S Karan (14IT252),

3rd Year, Department of Information Technology, National Institute of Technology Karnataka, Surathkal, Mangalore - 575025

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/sunilitggu/relation_extraction_in_clinical_text/issues/1#issuecomment-276970067, or mute the thread https://github.com/notifications/unsubscribe-auth/AIt3RfFRAVKKmxY7_C75qYOVC9PPYca5ks5rYeY8gaJpZM4LhY_y .

-- Regards Sunil Kumar Sahu Research Scholar CSE Department, IIT Guwahati, India

"If debugging is the process of removing bugs, then programming must be the process of putting them in" Dijkstra

[Ammarbaig123]

Sir can you please share the python code with me that you used to pre process the data ?