GitPlanet

Cheap and reliable Node.js hosting starts at $3/month, and $1/month static HTML hosting

Created with love in Canada, visit hostnodejs.com today

Feel like to post an Ad? Learn Details
All Projects → macournoyer → Neuralconvo

macournoyer / Neuralconvo

Neural conversational model in Torch

Programming Languages

lua
6591 projects

Labels

deep-learningmachine-learningseq2seqtorch

Projects that are alternatives of or similar to Neuralconvo

Pytorch Beam Search Decoding
PyTorch implementation of beam search decoding for seq2seq models
Stars: ✭ 204 (-73.61%)
Mutual labels:  seq2seq, torch
sentence2vec
Deep sentence embedding using Sequence to Sequence learning
Stars: ✭ 23 (-97.02%)
Mutual labels:  torch, seq2seq
torch-asg
Auto Segmentation Criterion (ASG) implemented in pytorch
Stars: ✭ 42 (-94.57%)
Mutual labels:  torch, seq2seq
Seq2seqchatbots
A wrapper around tensor2tensor to flexibly train, interact, and generate data for neural chatbots.
Stars: ✭ 466 (-39.72%)
Mutual labels:  seq2seq
Seq2seq.pytorch
Sequence-to-Sequence learning using PyTorch
Stars: ✭ 514 (-33.51%)
Mutual labels:  seq2seq
Mario Ai
Playing Mario with Deep Reinforcement Learning
Stars: ✭ 579 (-25.1%)
Mutual labels:  torch
Deepo
Setup and customize deep learning environment in seconds.
Stars: ✭ 6,145 (+694.95%)
Mutual labels:  torch
Practical Pytorch
Go to https://github.com/pytorch/tutorials - this repo is deprecated and no longer maintained
Stars: ✭ 4,329 (+460.03%)
Mutual labels:  seq2seq
Unet Segmentation Pytorch Nest Of Unets
Implementation of different kinds of Unet Models for Image Segmentation - Unet , RCNN-Unet, Attention Unet, RCNN-Attention Unet, Nested Unet
Stars: ✭ 683 (-11.64%)
Mutual labels:  torch
Seq2seq Couplet
Play couplet with seq2seq model. 用深度学习对对联。
Stars: ✭ 5,149 (+566.11%)
Mutual labels:  seq2seq
Ntire2017
Torch implementation of "Enhanced Deep Residual Networks for Single Image Super-Resolution"
Stars: ✭ 554 (-28.33%)
Mutual labels:  torch
Text Summarization Tensorflow
Tensorflow seq2seq Implementation of Text Summarization.
Stars: ✭ 527 (-31.82%)
Mutual labels:  seq2seq
Awesome Torch
A curated list of awesome Torch tutorials, projects and communities
Stars: ✭ 588 (-23.93%)
Mutual labels:  torch
Neural Vqa
❔ Visual Question Answering in Torch
Stars: ✭ 487 (-37%)
Mutual labels:  torch
Deeplearningmugenknock
でぃーぷらーにんぐを無限にやってディープラーニングでDeepLearningするための実装CheatSheet
Stars: ✭ 684 (-11.51%)
Mutual labels:  seq2seq
Waifu2x
Image Super-Resolution for Anime-Style Art
Stars: ✭ 22,741 (+2841.91%)
Mutual labels:  torch
Seq2seq Pytorch
Sequence to Sequence Models with PyTorch
Stars: ✭ 678 (-12.29%)
Mutual labels:  seq2seq
Seq2seq
Minimal Seq2Seq model with Attention for Neural Machine Translation in PyTorch
Stars: ✭ 552 (-28.59%)
Mutual labels:  seq2seq
Keras Openface
Keras-OpenFace is a project converting OpenFace from Torch implementation to a Keras version
Stars: ✭ 538 (-30.4%)
Mutual labels:  torch
Practical seq2seq
A simple, minimal wrapper for tensorflow's seq2seq module, for experimenting with datasets rapidly
Stars: ✭ 563 (-27.17%)
Mutual labels:  seq2seq
View All Similar Projects ➔

Neural Conversational Model in Torch

This is an attempt at implementing Sequence to Sequence Learning with Neural Networks (seq2seq) and reproducing the results in A Neural Conversational Model (aka the Google chatbot).

The Google chatbot paper became famous after cleverly answering a few philosophical questions, such as:

Human: What is the purpose of living?
Machine: To live forever.

How it works

The model is based on two LSTM layers. One for encoding the input sentence into a "thought vector", and another for decoding that vector into a response. This model is called Sequence-to-sequence or seq2seq.

seq2seq
Source: http://googleresearch.blogspot.ca/2015/11/computer-respond-to-this-email.html

In this experiment, we train the seq2seq model with movie dialogs from the Cornell Movie-Dialogs Corpus. The lines are shortened to the first sentence.

Sample conversation

Here's a sample conversation after training for 20 epoch with 50000 examples, using the following command:

th train.lua --cuda --dataset 50000 --hiddenSize 1000

(Took 3 days to train on my GeForce GTX 780M.)

For OpenCL, use --opencl instead of --cuda. To train on CPU, don't provide any of those two.

WARNING: I can no longer reproduce those results. The cause seems to be a change in one of dependencies. But I'm currently working on a new implementation based on harvardnlp/seq2seq-attn.

me: Hello?
bot: Hi.

me: How are you?
bot: I'm fine.

me: What's your name?
bot: It's hard to describe.

me: How so?
bot: I'm not sure.

me: What color is the sky?
bot: It's blue.

me: What is your job?
bot: It's not that i'm a fucking werewolf!

me: What is the purpose of life?
bot: A gift.

me: Are you intelligent?
bot: Yes, well...

me: Are you a machine?
bot: That's a lie.

me: Are you human?
bot: No, i'm not.

me: What are you?
bot: I'm not sure.

me: Do you plan on taking over the world?
bot: No, i don't.

Phew! That was close. Good thing I didn't train it on the full dataset. Please experiment responsibly.

(Disclaimer: nonsensical responses have been removed.)

Installing

  1. Install Torch.

  2. Install the following additional Lua libs:

    luarocks install nn
luarocks install rnn
luarocks install penlight

    To train with CUDA install the latest CUDA drivers, toolkit and run:

    luarocks install cutorch
luarocks install cunn

    To train with opencl install the lastest Opencl torch lib:

    luarocks install cltorch
luarocks install clnn

  3. Download the Cornell Movie-Dialogs Corpus and extract all the files into data/cornell_movie_dialogs.

Training

th train.lua [-h / options]

The model will be saved to data/model.t7 after each epoch if it has improved (error decreased).

Options (some, not all)

  • --opencl use opencl for computation (requires torch-cl)
  • --cuda use cuda for computation
  • --gpu [index] use the nth GPU for computation (eg. on a 2015 MacBook --gpu 0 results in the Intel GPU being used while --gpu 1 uses the far more powerful AMD GPU)
  • -- dataset [size] control the size of the dataset
  • --maxEpoch [amount] specify the number of epochs to run

Testing

To load the model and have a conversation:

th eval.lua

License

MIT License

Copyright (c) 2016 Marc-Andre Cournoyer

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].