search

TobiasNorlund / Parameterized_RI

Improving Random Indexing word embeddings using a parameterization
http://www.tobias.norlund.se/portfolio/masters-thesis/
1 stars 0 forks source link

I have a few questions to ask you about your project #2

Open qianqiangediao opened 8 years ago

qianqiangediao commented 8 years ago

The code of "Random Indexing embeddings generated by https://github.com/TobiasNorlund/CorpusParser " is problematic,while I run it. I find some parameters are null value,therefore ".bin" also is null.Can you provide the whole document of the random Indexing embeddings. Thank you for your kind consideration of this request.

TobiasNorlund commented 8 years ago

Hi, I have now added a readme which points to the paper and also a link to the corpus file (wiki.2010.txt) which I used in CorpusParser. Hope this helps!

qianqiangediao commented 8 years ago

Thank you for your kind consideration of this request.I download your wiki_data and run code, but I met an urgent problem.More specifically, while I run into the step " params = lasagne.layers.get_all_params(l_out) + embedding.get_update_parameter_vars()" ,I got an error message as ""TypeError: ('An update must have the same type as the original shared variable (shared_var=thetas, shared_var.type=TensorType(float32, matrix), update_val=Elemwise{sub,no_inplace}.0, update_val.type=TensorType(float64, 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.')" .I debug it for a long time, So can you give me some idea to help me to resolve this problem. thank you very much