A chatbot based on seq2seq architecture.
This is the successor to my previous torch project 'seq2seq-chatbot'. Portions of this project were directly adapted from the tensorflow translation example.
This is based off the research in these papers: Sutskever et al., 2014. and Vinyals & Le, 2015.
- Python 3
- TensorFlow 1.2
- (optional) matplotlib
$ pip install matplotlib
To use your own data read Data Format, otherwise to use included Cornell Movie Dialogues Corpur data:
$ python train.py
If you wish to provide your own data use the format below to provide your own data.
Each continuous conversation must be in it's own text file. Each line of the text files is a response to the previous line. The exception of course is that the first line of the text file is the conversation start. After entering your own data, you can run it with:
$ python sequencelengthplotter.py --data_dir="path/to/text_files"
The options for the above command are:
| Name | Type | Description |
|---|---|---|
| data_dir | string | path to raw data files |
| plot_histograms | boolean | plots histograms for target and source sequences |
| plot_scatter | boolean | plot x-y scatter plot for target length vs. source length |
in buckets.cfg you can then modify the bucket values accordingly. You can add or remove buckets. All bucket values in the [buckets] subheading will be parsed in. Each line under [buckets] should be of the format:
bucket_name: source_length,target_length
In the same configuration file the data settings can also be changed under the [max_data_sizes] subheading.
| Name | Type | Description |
|---|---|---|
| num_lines | int | number of lines in conversation to go back |
| max_target_length | int | max length of target sequences |
| max_source_length | int | max length of source sequences |
A configuration file was used because it was a mess trying to find out how to pass bucket values via command line. This seemed like a half-way decent solution. It also enables (in the future) sequencelengthplotter.py the ability to autonomously change these values without requiring any user input.
Once you are satisfied with bucket sizes you can then run the optimizer. This will also clean and process your raw text data. To do this:
$ python train.py --raw_data_dir="data/cornell_lines"
There are several options that can be employed:
| Name | Type | Description |
|---|---|---|
| hidden_size | int | number of hidden units in hidden layers |
| num_layers | int | number of hidden layers |
| batch_size | int | size of batchs in training |
| max_epoch | int | max number of epochs to train for |
| learning_rate | float | beggining learning rate |
| steps_per_checkpoint | int | number of steps before running test set |
| lr_decay_factor | float | factor by which to decay learning rate |
| batch_size | int | size of batchs in training |
| checkpoint_dir | string | directory to store/restore checkpoints |
| dropout | float | probability of hidden inputs being removed |
| grad_clip | int | max gradient norm |
| max_train_data_size | int | Use a subset of available data to train |
| vocab_size | int | max vocab size |
| train_frac | int | percentage of data to use for training (rest is used for testing) |
| raw_data_dir | int | raw conversation text files stored here |
| data_dir | int | Directory data processor will store train/test/vocab files |
| max_source_length | int | How long the source sentences can be at most |
| max_target_length | int | How long the target sentences can be at most |
| convo_limits | int | How far back the bot's memory should go for the conversation |
Tensorboard not yet implemented.
** This still needs to be tested since the update to TF 1.2
To have a real conversation with your bot you can begin an interactive prompt by doing:
*Note the sample.py file does not currently read in bucket sizes set from the config file (this will be fixed shortly)
$ python sample.py --checkpoint_dir="path/to/checkpointdirectory" --data_dir="path/to/datadirectory"
A prompt will open up something like:
>
The user can then type in to the prompt
ex/ > Hello chatbot, how are you?
Then hit 'enter' for the response.
Summary of options below:
| Name | Type | Description |
|---|---|---|
| checkpoint_dir | string | path to saved checkpoint |
| data_dir | string | path to directory where vocab file is |
Coming soon.
- Downloadable pretrained model