Comments (7)
@jli05 would you mind stating the pros and cons of this in the context of nmt and dl4mt repo here, thanks a lot.
from dl4mt-tutorial.
Basically I hope this and future repos achieve more than merely as a tool for academic research.
Cons:
- Time on refactoring the current software/data systems to work with Python 3.4+. The amount of effort is proportional to the size of current systems.
- A little learning curve, max half a day.
Pros:
- Evolving with time. Python 2.7 is entering the End Of Life cycle till 2020 (PEP 373); Ubuntu is phasing out install of Python 2.7 by default. It's not hard to imagine newer releases of RHEL will adopt similar policies.
- Allows other parties to contribute who're developing with modern choice of languages.
- Allows constant maintenance of the repos so that it's a thriving experience for whoever maintain them.
Theano is one of the rare academic repos that's constantly maintained and streamlined, which can rival corporate offerings in a certain aspect. If we google for BlinkDB, the last commit was two years ago -- it remained a good concept that earned the author a paper. I believe software has wider and deeper impacts than publications today. I know this repo was initially put together as a tutorial; however I'd wish more for it and all future sequels: to be a point of reference for NMT, which showcases excellence in algorithm and brick-and-mortar code.
from dl4mt-tutorial.
@jli05 Thans for sharing your view! I think this definitely makes sense.
@orhanf How about we don't fully validate the changes (as in training a full model and verifying that the same BLEU scores for all the models and language pairs we've tried), but simply make it runnable with Python3?
@jli05 Would you kindly help us move toward Python3 by making a PR for, say, one of the sessions?
from dl4mt-tutorial.
Sure I'll do it in the coming 1-2 weeks. Thanks @kyunghyuncho @orhanf !
from dl4mt-tutorial.
Could we confirm what's the exact workflow in data/
?
My understanding is that for a simplest setup, we could run setup_local_env.sh
. Is that sufficient? We made no call to preprocess.sh
therein?
All the sessions take in the tokenised wiki dump and its associated dictionary from wiki.tok.txt.gz
and wiki.tok.txt.gz.pkl
. Currently we don't have scripts for generating them?
from dl4mt-tutorial.
@orhanf can you answer this?
from dl4mt-tutorial.
@jli05 , a detailed description is provided with #53 but let me clarify this further here.
setup_local_env.sh
was the initial script provided in the repo, intended to download an example data and preprocess it (only tokenization). Later on, in order to use subword-units (bpe), another script was added (preprocess.sh
), which pre-processes existing data (tokenize, learn bpe, apply bpe and shuffle)
We now merged both functionalities into setup_local_env.sh
and made it to call preprocess.sh
optionally (with -b
flag) when you want to use bpe.
from dl4mt-tutorial.
Related Issues (20)
- Why the grads have to be shared? HOT 2
- Cost is Nan after one epoch if maxlen > 50 HOT 3
- Where is dataset='/ichec/work/dl4mt_data/nec_files/wiki.tok.txt.gz'? HOT 2
- discrepancy between paper and code HOT 1
- NaN detected HOT 1
- L2 regularization on bias terms?
- Unnecessary bias term? HOT 1
- Asymmetry in read gate application
- dim == dim_nonlin and nin == nin_nonlin must be always true?
- How to build the dataset 'all.en.concat.gz.pkl' in session2/train_nmt_all.py? HOT 1
- Random Translations? HOT 5
- ValueError: unsupported pickle protocol: 3 HOT 2
- why convert the value of matrix to the type with astype('float32')? HOT 2
- a detailed description about param_init_gru and gru_layer HOT 2
- the value of TensorType(float32, 3D) HOT 2
- compute word probabilities HOT 2
- Maybe it's time to upgrade to Python 3 and ditch Python 2 support? HOT 1
- By condition what does it mean in this tutorial?
- Can anyone provide GPU version of translate.py? HOT 1
- TypeError
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from dl4mt-tutorial.