TypeError: ('An update must have the same type as the original shared variable (shared_var=hid_init, shared_var.type=CudaNdarrayType(float32, row), update_val=Elemwise{sub,no_inplace}.0, update_val.type=TensorType(float32, matrix)).', 'If the difference is related to the broadcast pattern, you can call the tensor.unbroadcast(var, axis_to_unbroadcast[, ...]) function to remove broadcastable dimensions.')

[basma-b]

Hi, a month ago your code worked fine on my machine but today I'm having this error TypeError: ('An update must have the same type as the original shared variable (shared_var=hid_init, shared_var.type=CudaNdarrayType(float32, row), update_val=Elemwise{sub,no_inplace}.0, update_val.type=TensorType(float32, matrix)).', 'If the difference is related to the broadcast pattern, you can call the tensor.unbroadcast(var, axis_to_unbroadcast[, ...]) function to remove broadcastable dimensions.') can you help me please ?

[npow]

Did you perhaps update some packages?

[basma-b]

Sorry it was a problem with Lasagne !

[basma-b]

Hi Nissan,

Please I am trying to download http://cs.mcgill.ca/~npow1/data/ubuntu_dataset.tgz.aa using wget but it seems that I don't have access. can you please check the other files with a* extension ?

Many thanks, Basma

On 31 March 2017 at 19:02, Nissan Pow notifications@github.com wrote:

Did you perhaps update some packages?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/npow/ubottu/issues/7#issuecomment-290770022, or mute the thread https://github.com/notifications/unsubscribe-auth/AJfNkBYKY-HbP1Kya0T7Kv9F0LWIOEXLks5rrTGygaJpZM4Mvo0J .

[npow]

You can try this link: http://cs.mcgill.ca/~jpineau/datasets/ubuntu-corpus-1.0/ Although you should probably be running version 2 of the dataset: https://github.com/rkadlec/ubuntu-ranking-dataset-creator

[basma-b]

Thanks Nissan, yes I used the v2 of the corpus to get the same results as thones you report in the paper and on the github but you only provide the pkl file and I need csv ones. Ubuntu_blobs is on the link I sent is it on v1 or 2 ? And can you please give me the csv files that you used to get those results ?

On Apr 3, 2017 19:56, "Nissan Pow" notifications@github.com wrote:

You can try this link: http://cs.mcgill.ca/~jpineau/ datasets/ubuntu-corpus-1.0/ Although you should probably be running version 2 of the dataset: https://github.com/rkadlec/ubuntu-ranking-dataset-creator

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/npow/ubottu/issues/7#issuecomment-291221849, or mute the thread https://github.com/notifications/unsubscribe-auth/AJfNkCASI5hSfKlEKcj6e1zXeaVuG77Lks5rsTLBgaJpZM4Mvo0J .

[basma-b]

Hi Nissan,

please can you tell me which config did you use to get the results you put on the github ? I obtained the same using th blob files but I need csv ones. I tried .-t, -s -l options but still can't get the same. Thanks

On 3 April 2017 at 20:02, Basma El Amel Boussaha ab_boussaha@esi.dz wrote:

Thanks Nissan, yes I used the v2 of the corpus to get the same results as thones you report in the paper and on the github but you only provide the pkl file and I need csv ones. Ubuntu_blobs is on the link I sent is it on v1 or 2 ? And can you please give me the csv files that you used to get those results ?

On Apr 3, 2017 19:56, "Nissan Pow" notifications@github.com wrote:

You can try this link: http://cs.mcgill.ca/~jpineau/d atasets/ubuntu-corpus-1.0/ Although you should probably be running version 2 of the dataset: https://github.com/rkadlec/ubuntu-ranking-dataset-creator

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/npow/ubottu/issues/7#issuecomment-291221849, or mute the thread https://github.com/notifications/unsubscribe-auth/AJfNkCASI5hSfKlEKcj6e1zXeaVuG77Lks5rsTLBgaJpZM4Mvo0J .

[WanzhengZhu]

@basma-b May I know which version of Lasagne works for the code? I got the same problem too!