thomasmesnard / deepmind-teaching-machines-to-read-and-comprehend Goto Github PK
View Code? Open in Web Editor NEWImplementation of "Teaching Machines to Read and Comprehend" proposed by Google DeepMind
License: MIT License
Implementation of "Teaching Machines to Read and Comprehend" proposed by Google DeepMind
License: MIT License
Hi,
I want to make my own ale models for DQN Reinforce Learning using my video file input.
In video images, there are class labels in the corner for supervised training.
How can I make my own ALE model?
Thank you in advances.
I found that when representing the context word (concatenating the forward and backward rnn), the backward rnn is not reversed to match the sequence order of the forward rnn.
rzai@rzai00:/prj/DeepMind-Teaching-Machines-to-Read-and-Comprehend$ export DATAPATH=/prj/DeepMind-Teaching-Machines-to-Read-and-Comprehend/
rzai@rzai00:/prj/DeepMind-Teaching-Machines-to-Read-and-Comprehend$ python train.py deepmind_deep_lstm/prj/DeepMind-Teaching-Machines-to-Read-and-Comprehend$
Using gpu device 0: GeForce GTX 1080 (CNMeM is disabled, cuDNN 5005)
No plotting extension available.
Traceback (most recent call last):
File "train.py", line 46, in
ds, train_stream = data.setup_datastream(path, vocab_path, config)
File "/home/rzai/prj/DeepMind-Teaching-Machines-to-Read-and-Comprehend/data.py", line 137, in setup_datastream
ds = QADataset(path, vocab_file, config.n_entities, need_sep_token=config.concat_ctx_and_question)
AttributeError: 'module' object has no attribute 'n_entities'
rzai@rzai00:
Hi,
How can I check the performance of the trained model on the test data?
Hi,
I found that the variable n_entities
is not be initially assigned. There will be a error when the class QADataset
is constructed.
I got it ... so sorry asking such a stupid question
rzai@rzai00:/prj/DeepMind-Teaching-Machines-to-Read-and-Comprehend$/prj/DeepMind-Teaching-Machines-to-Read-and-Comprehend$ CUDA_VISIBLE_DEVICES=1 python train.py deepmind_deep_lstm
rzai@rzai00:
Using gpu device 0: GeForce GTX 1080 (CNMeM is disabled)
No plotting extension available.
<module 'config.deepmind_deep_lstm' from '/home/rzai/prj/DeepMind-Teaching-Machines-to-Read-and-Comprehend/config/deepmind_deep_lstm.pyc'>
Traceback (most recent call last):
File "train.py", line 53, in
m = config.Model(config, ds.vocab_size)
File "/home/rzai/prj/DeepMind-Teaching-Machines-to-Read-and-Comprehend/model/deep_lstm.py", line 74, in init
cost = Softmax().categorical_cross_entropy(answer, probs).mean()
File "/usr/local/lib/python2.7/dist-packages/blocks/bricks/base.py", line 362, in call
return self.application.apply(self, *args, **kwargs)
File "/usr/local/lib/python2.7/dist-packages/blocks/bricks/base.py", line 297, in apply
outputs = self.application_function(brick, *args, **kwargs)
File "/usr/local/lib/python2.7/dist-packages/blocks/bricks/simple.py", line 390, in categorical_cross_entropy
x.copy(name='log_probabilities'))
TypeError: copy() got an unexpected keyword argument 'name'
rzai@rzai00:~/prj/DeepMind-Teaching-Machines-to-Read-and-Comprehend$
Firstly, thank you for open sourcing this project.
How did you get the vocab.txt file? There are 29,406 words in the file.
However, I counted all the unique words in the CNN dataset and there are 119,567 unique words.
Can I get the pre trained weights file for this model?
First of all, awesome work on this project!
I have trained a model. How can I get predictions from the trained model?
Thanks,
Raffi
It failed when writing model to folder model_params after finished epoch 0 first 1000 steps. It's not a big issue but wasted several hours since I was not aware of that. Please add check (and create that folder in case of missing) in train.py. Thanks!
Hey Guys,
Firstly great work! Are you able share the Test file or script to produce predictions from the model?
Cheers
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.