A toy chatbot powered by deep learning and trained on data from Reddit. Built on TensorFlow v1.4.0 and Python v3.6
Here is a sample chat transcript (not cherry-picked). Lines that start with ">" are me; the lines without ">" are the chatbot's responses.
$ python3 chatbot.py
Creating model...
Restoring weights...
> Hi, how are you?
I'm not the only one here that's having fun with you, it's all me.
> Who else is with you?
Haha. I'm not the one who made the comment. I'm the one who said that and didn't think that much of anything.
> Oh, I see. So what's new?
Nothing. I just thought it was a bit of an overreaction. It's been a while since I've read it.
> It's all good
It's all good. It's one of those things that makes me smile :)
> That's great. This is actually my demo chat, so it's nice that you're keeping things positive!
Thank you :)
## Get Started
- **Install [TensorFlow](http://www.tensorflow.org/)** for Python 3. To run a pretrained model, the CPU-only installation should suffice. If you want to train your own models, you'll need the GPU installation of TensorFlow (and a powerful CUDA-compatible GPU).
- Clone this project to your computer.
### Run my pre-trained model
- **Download [my pre-trained model](https://drive.google.com/uc?id=1rRRY-y1KdVk4UB5qhu7BjQHtfadIOmMk&export=download)** (2.3 GB). The zip file extracts into a folder named "reddit". Place that folder into the "models" directory of this project.
- **Run the chatbot**. Open a terminal session and run `python3 chatbot.py`. Warning: this pre-trained model was trained on a diverse set of frequently off-color Reddit comments. It can (and eventually will) say things that are offensive, disturbing, bizarre or sexually explicit. It may insult minorities, it may call you names, it may accuse you of being a pedophile, it may try to seduce you. Please don't use the chatbot if these possibilities would distress you!
Try playing around with the arguments to `chatbot.py` to obtain better samples:
- **beam_width**: By default, `chatbot.py` will use beam search with a beam width of 2 to sample responses. Set this higher for more careful, more conservative (and slower) responses, or set it to 1 to disable beam search.
- **temperature**: At each step, the model ascribes a certain probability to each character. Temperature can adjust the probability distribution. 1.0 is neutral (and the default), lower values increase high probability values and decrease lower probability values to make the choices more conservative, and higher values will do the reverse. Values outside of the range of 0.5-1.5 are unlikely to give coherent results.
- **top-n**: At each step, zero out the probability of all possible characters except the *n* most likely. Disabled by default.
- **relevance**: Two models are run in parallel: the primary model and the mask model. The mask model is scaled by the relevance value, and then the probabilities of the primary model are combined according to equation 9 in [Li, Jiwei, et al. "A diversity-promoting objective function for neural conversation models." arXiv preprint arXiv:1510.03055 (2015)](https://arxiv.org/abs/1510.03055). The state of the mask model is reset upon each newline character. The net effect is that the model is encouraged to choose a line of dialogue that is most relevant to the prior line of dialogue, even if a more generic response (e.g. "I don't know anything about that") may be more absolutely probable. Higher relevance values put more pressure on the model to produce relevant responses, at the cost of the coherence of the responses. Going much above 0.4 compromises the quality of the responses. Setting it to a negative value disables relevance, and this is the default, because I'm not confident that it qualitatively improves the outputs and it halves the speed of sampling.
These values can also be manipulated during a chat, and the model state can be reset, without restarting the chatbot:
$ python3 chatbot.py Creating model... Restoring weights...
--temperature 1.3 [Temperature set to 1.3]
--relevance 0.3 [Relevance set to 0.3]
--relevance -1 [Relevance disabled]
--topn 2 [Top-n filtering set to 2]
--topn -1 [Top-n filtering disabled]
--beam_width 5 [Beam width set to 5]
--reset [Model state reset]
### Get training data
If you'd like to train your own model, you'll need training data. There are a few options here.
- **Use pre-formatted Reddit training data.** This is what the pre-trained model was trained on.
[Download the training data](https://drive.google.com/uc?id=1s77S7COjrb3lOnfqvXYfn7sW_x5U1_l9&export=download) (2.1 GB). Unzip the monolithic zip file. You'll be left with a folder named "reddit" containing 34 files named "output 1.bz2", "output 2.bz2" etc. Do not extract those individual bzip2 files. Instead, place the whole "reddit" folder that contains those files inside the `data` folder of the repo. The first time you run `train.py` on this data, it will convert the raw data into numpy tensors, compress them and save them back to disk, which will create files named "data0.npz" through "data34.npz"
- **Generate your own Reddit training data.** If you would like to generate training data from raw Reddit archives, download a torrent of Reddit comments from the torrent links [listed here](https://www.reddit.com/r/datasets/comments/65o7py/updated_reddit_comment_dataset_as_torrents/).
Once you have your raw reddit data, place it in the `reddit-parse/reddit_data` subdirectory and use the `reddit-parse.py` script included in the project file to convert them into compressed text files of appropriately formatted conversations. This script chooses qualifying comments (must be under 200 characters, can't contain certain substrings such as 'http://', can't have been posted on certain subreddits) and assembles them into conversations of at least five lines. Coming up with good rules to curate conversations from raw reddit data is more art than science. I encourage you to play around with the parameters in the included `parser_config_standard.json` file, or to mess around with the parsing script itself, to come up with an interesting data set.
Please be aware that there is a *lot* of Reddit data included in the torrents. It is very easy to run out of memory or hard drive space. I used the entire archive (~304 GB compressed), and ran the `reddit-parse.py` script with the configuration I included as the default, which holds a million comments (several GB) in memory at a time, takes about a day to run on the entire archive, and produces 2.1 GB of bzip2-compressed output. When training the model, this raw data will be converted into numpy tensors, compressed, and saved back to disk, which consumes another ~5 GB of hard drive space. I acknowledge that this may be overkill relative to the size of the model.
- **Provide your own training data.** Training data should be one or more newline-delimited text files. Each line of dialogue should begin with "> " and end with a newline. You'll need a lot of it. Several megabytes of uncompressed text is probably the minimum, and even that may not suffice if you want to train a large model. Text can be provided as raw .txt files or as bzip2-compressed (.bz2) files.
- **Simulate the United States Supreme Court.** I've included a corpus of United States Supreme Court oral argument transcripts (2.7 MB compressed) in the project under the `data/scotus` directory.
Once you have training data in hand (and located in a subdirectory of the `data` directory):
### Train your own model
- **Train.** Use `train.py` to train the model. The default hyperparameters are the best that I've found, and are what I used to train the pre-trained model for a couple of months. These hyperparameters will just about fill the memory of a GTX 1080 Ti GPU (11 GB of VRAM), so if you have a smaller GPU, you will need to adjust them accordingly (for example, set --num_blocks to 2).
Training can be interrupted with crtl-c at any time, and will immediately save the model when interrupted. Training can be resumed on a saved model and will automatically carry on from where it was interrupted.