File size: 40,303 Bytes
8e28984 9a353a0 8e28984 9a353a0 8e28984 9a353a0 8e28984 9a353a0 8e28984 9a353a0 8e28984 9a353a0 8e28984 9a353a0 8e28984 9a353a0 8e28984 9a353a0 8e28984 9a353a0 8e28984 9a353a0 8e28984 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 | # A State-of-the-Art Large-scale Pretrained Response Generation Model (DialoGPT)
## This project page is no longer maintained as DialoGPT is superseded by [GODEL](https://github.com/microsoft/GODEL), which outperforms DialoGPT according to the results of [this paper](https://arxiv.org/pdf/2206.11309.pdf). Unless you use DialoGPT for reproducibility reasons, we highly recommend you switch to [GODEL](https://github.com/microsoft/GODEL).
This repository contains the source code and trained model for a large-scale pretrained dialogue response generation model. The [human evaluation results](#human_eval) indicate that the response generated from DialoGPT is comparable to human response quality under a single-turn conversation Turing test.
<!--See more details on our [project page](https://www.microsoft.com/en-us/research/project/large-scale-pretraining-for-response-generation/)-->
The repository is based on [huggingface pytorch-transformer](https://github.com/huggingface/transfer-learning-conv-ai) and [OpenAI GPT-2](https://github.com/openai/gpt-2), containing data extraction script, model training code and pretrained small (117M) medium (345M) and large (762M) model checkpoint.
The model is trained on 147M multi-turn dialogue from Reddit discussion thread. The largest model can be trained in several hours on a 8 V100 machines (however this is not required), with distributed training and FP16 option.
The include script can be used to reproduce the results of DSTC-7 grounded dialogue generation challenge and a 6k multi-reference dataset created from Reddit data.
Project webpage: [https://www.microsoft.com/en-us/research/project/large-scale-pretraining-for-response-generation/](https://www.microsoft.com/en-us/research/project/large-scale-pretraining-for-response-generation/)
ArXiv paper: [https://arxiv.org/abs/1911.00536](https://arxiv.org/abs/1911.00536)
## News ##
***(Update 07/09/2022) Changes on the files.pushshift.io/reddit server caused our data generation pipeline to break. These problems have now been fixed, and the steps explained in the Data Preparation subsection below should work again. Data is generated in about 10 hours with 8 processes (`-j 8`), and 800GB of temporary disk space is needed.***
***(Update 06/23/2021) We have released a retrieval-augmented/grounded version of DialoGPT (RetGen), please check out the [RetGen repo](https://github.com/dreasysnail/RetGen) and [RetGen paper](https://arxiv.org/abs/2105.06597)***
***(Update 05/20/2021) An awesome [video walkthrough](https://www.youtube.com/watch?v=Zo679MYoJns) on YouTube for DialoGPT by [Prakhar Mishra](http://wsl.iiitb.ac.in/prakhar-mishra/)***
***(Update 03/31/2021) A 3rd party demo by [AK391](https://github.com/AK391) using Gradio [web demo](https://gradio.app/g/AK391/DialoGPT) try it out***
***(Update 09/15/2020) A set of large-scale [dialog ranking models](https://github.com/golsun/DialogRPT) has been released!***
DialoGPT generation is improved by integrating with our latest dialog ranking models, [DialogRPT](https://github.com/golsun/DialogRPT)
***(Update 07/08/2020) The 6K multi-ref test set has been released!***
To generate the data, pleaser run `demo.py` and set the data option to 'full', the generated 6k multi-ref test set will be located at
`./data/test.refs.txt`
***(Update 03/10/2020) Model cards available in Huggingface Transformers!***
Please check out our model cards in huggingface Transformers repository. With several lines of code it should be pretty straighforward to play with the DialoGPT interactively.
[small model: https://huggingface.co/microsoft/DialoGPT-small](https://huggingface.co/microsoft/DialoGPT-small)
[medium model: https://huggingface.co/microsoft/DialoGPT-medium](https://huggingface.co/microsoft/DialoGPT-medium)
[large model: https://huggingface.co/microsoft/DialoGPT-large](https://huggingface.co/microsoft/DialoGPT-large)
[**(New)** Ranking model: https://huggingface.co/microsoft/DialogRPT-updown](https://huggingface.co/microsoft/DialogRPT-updown?text=I+love+NLP%21+%3C%7Cendoftext%7C%3E+Me+too%21)
***(Update 01/06/2020) Some third-party decoding script implementations:***
- [https://github.com/polakowo/gpt2bot](https://github.com/polakowo/gpt2bot) GPT2Bot implementation based on telegram by polakowo, [ref](https://github.com/microsoft/DialoGPT/issues/3#issuecomment-573904419)
- [https://colab.research.google.com/drive/1PslHE4Rl4RqSa20s7HEp0ZKITBir6ezE](https://colab.research.google.com/drive/1PslHE4Rl4RqSa20s7HEp0ZKITBir6ezE) A colab interactive notebook by qywu,[ref](https://github.com/microsoft/DialoGPT/issues/3#issuecomment-551410203)
- [https://github.com/andreamad8/DialoGPT2-Interact](https://github.com/andreamad8/DialoGPT2-Interact) An interactive script featuring multiturn chatbot by andreamad8,[ref](https://github.com/microsoft/DialoGPT/issues/3#issuecomment-551450016)
- [https://github.com/LHolten/DialoGTP-MMI-decoder](https://github.com/LHolten/DialoGTP-MMI-decoder) An MMI implementation by LHolten,[ref](https://github.com/microsoft/DialoGPT/issues/3#issuecomment-558318401)
- [https://colab.research.google.com/drive/1-_KjlAV3J1IVDw_9KogjKDCzgFY7Jp7E](https://colab.research.google.com/drive/1-_KjlAV3J1IVDw_9KogjKDCzgFY7Jp7E) A colab interactive notebook by illuminascent@Reddit,[ref](https://www.reddit.com/r/MachineLearning/comments/dt5woy/p_dialogpt_state_of_the_art_conversational_model/?st=k530k3oo&sh=f6cd20fd)
- [https://colab.research.google.com/drive/15wa925dj7jvdvrz8_z3vU7btqAFQLVlG](https://colab.research.google.com/drive/15wa925dj7jvdvrz8_z3vU7btqAFQLVlG) A great tutorial of how to finetune DialoGPT to build a customized bot built by [Rostyslav Neskorozhenyi](https://www.linkedin.com/in/slanj/). [ref](https://towardsdatascience.com/make-your-own-rick-sanchez-bot-with-transformers-and-dialogpt-fine-tuning-f85e6d1f4e30)
- [https://gradio.app/g/AK391/DialoGPT](https://gradio.app/g/AK391/DialoGPT) A 3rd party demo by [AK391](https://github.com/AK391) using Gradio [web demo](https://gradio.app/g/AK391/DialoGPT)
<!--**This github repository will be updated soon. Please stay tuned.**-->
<!--## Minimal Computational Configurations-->
## Recommended Configuration
- Linux Ubuntu 16.04
- GPU with at least 12G memory
DialoGPT was developed entirely on **Ubuntu 16.04**, and -- depending on our availability -- we try to provide support if you experience difficulties running the code on the same configuration. However, we are **unable to provide support for other distributions or operating systems**. Portions of the code may run on other UNIX flavors (macOS, Windows subsystem for Linux, Cygwin, etc.), but it is recommended to use Ubuntu for the main training code.
The training code can be run on CPU, but it can be slow. We would recommend to use GPU to train and finetune all models. There is no minimal limit of the number of GPUs. However, if using distributed train for multiple GPUs configuration, the speed-up vs the number of GPUs is roughly sub-linear. To simulate the same batchsize when using less GPUs, please use a larger `gradient_accumulation_steps` in model training.
The 117M and 345M model can be loaded in a single GPU with 12G memory. The 762M model would require a single GPU that has greater than 16G memory for efficient training. The training speed on a benchmark data with 50M training instances and V100 GPUs:
| n\_gpu | epoch time (h) | token/sec |
|----------------------|--------|--------|
| 1 | 118 | 10847 |
| 2 | 62 | 20645 |
| 4 | 34 | 37647 |
| 8 | 18 | 71356 |
Fine-tuning from our pretrained model on a new dataset typically requires 1-2 epochs.
## Setup & Installation (TL;DR)
We created a demo script `demo.py` to ease the difficulty of the deployment of this system. The `demo.py` contains a pipeline of **model downloading**, data extraction, data preprocessing and model training over a dummy dataset within one commandline.
#### Train model with Conda Environment
Please use the below commandlines to clone, install the requirements and load the Conda environment (Note that the Nvidia CUDA 10.0 developer toolkit is required):
```bash
sudo apt-get install -y make wget gzip bzip2 xz-utils zstd sed
```
```bash
git clone https://github.com/microsoft/DialoGPT.git
cd DialoGPT
conda env create -f LSP-linux.yml -n LSP
conda activate LSP
```
If you run this on an architecture other than Linux, please use `LSP-generic.yml` instead of `LSP-linux.yml` but please note that the generic one is not tested in all platform, so the stablity can not be gauranteed.
To use fp16 training, please install apex by using commands below
```bash
conda activate LSP
git clone https://github.com/NVIDIA/apex
cd apex
git reset --hard 3d01e4a0a188cc8df54bc6e44cf5eb40ff6b4cc5
pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" .
python3.6 demo.py
```
#### Train model with Docker environment
To start, first install the docker and Nvidia-docker from their official repos.
The image environment for running the code can be loaded as below:
*Nvidia-docker v2.**
```bash
$ docker run --gpus all --ipc=host --rm -it -v $PWD:/workspace --network=host icaruszyz/large-scale-training:dialogpt bash
```
*Nvidia-docker v1.**
```bash
$ nvidia-docker --rm -it -v $PWD:/workspace --network=host icaruszyz/large-scale-training:dialogpt bash
```
Inside the docker container, run
```bash
python demo.py
```
## Pipeline details
This section explains all components in the `demo.py`.
#### Data loading
Before running `demo.py`, you can set *DATA_FOLDER* (default value `./models`) in `demo.py` as the place you want to download all the data and pretrained/fine-tuned models. Then simply run
```bash
python demo.py
```
to
* automatically download models and data,
* prepare raw data into db that is ready to use for the program,
* generate a training scripts.
Note that by default the `demo.py` will use a dummy data, please specify the Reddit training data by using option `--data`. Three options are available:`dummy`,`small` and `full`.
```bash
python demo.py --data small
python demo.py --data full
```
The small Reddit data is around 140MB and the full Reddit data is more than 27GB. You can prepare a cup of coffee when processing with the full Reddit data because **it takes a long time**!
To generate the 6k multi-ref test set data, pleaser run `demo.py` and set the data option to 'full', the generation will be located at
`./data/test.refs.txt`
#### Pretrained model
The pretrained and fine-tuned models are available on azure blobstorage.
Please run/see `demo.py` for more details about how to download/use those models. Or you could download directly by using the links in `demo_utils.py`.
#### Preparing data
First, use the `prepare4db.sh` to convert a tsv data file into the correct format that the following script can recognize.
The trainig data need to be then processed into a database file with below commandline:
```bash
python prepro.py --corpus $DATA_PATH
```
#### Using the training script
The training script can be used in single GPU or multiple GPU settings (distributed training across multiple GPUs within a single node):
```bash
python ./LSP_train.py # Single GPU training
python -m torch.distributed.launch --nproc_per_node=8 ./LSP_train.py # Training on 8 GPUs
```
The training script accept several arguments to tweak the training:
Argument | Type | Default value | Description
---------|------|---------------|------------
max\_seq\_length | `int` | `128` | Maximum number of tokens for each training instance.
train\_input\_file | `str` | `""` | Path of the training dataset in a .db format
eval\_input\_file | `str` | `""` | Path of the validation set in a tsv format
continue_from | `int` | `0` | Resuming the training after a specified number of steps
fp16 | `boolean` | `True` | Whether to use 16-bits floating point for model training.
train\_batch\_size | `int` | `4` | Batch size for training
valid\_batch\_size | `int` | `4` | Batch size for validation
gradient\_accumulation\_steps | `int` | `2` | Accumulate gradients on several steps
learning\_rate | `float` | `1e-5` | Learning rate
lr\_schedule | `str` | `noam` | Learning rate schedule can be chosen from [`noam`, `noamwd`, `BERT`, `None`]
num\_optim\_steps | `int` | `1000000` | Number of training optimization steps
no_token_id | `boolean` | `True` | If set True, using all-zeros token-type embedding.
During the training, two log files will be updated. The `train_log.txt` and `eval_log.txt` contains the model loss, perplexity and training speed (tokens/sec) statistics for the training and dev set.
The log file and saved model checkpoint can be found in `./models/output_model`
#### Model decoding
We note that even with properly filtered Reddit dataset, sometimes our model can still generate moderately toxic/inappropriate responses. Due to this reason, we are unable to provide the decoding script at this time (The live demo and decoding script access is upon invitation only now ).
We are currently still working on a controlled decoding method to prevent this system from toxic generation. Please stay tuned.
**See issues [#3](https://github.com/microsoft/DialoGPT/issues/3) and [Reddit discussions](https://www.reddit.com/r/MachineLearning/comments/dt5woy/p_dialogpt_state_of_the_art_conversational_model/) for some discussions on third-party decoding methods.**
See below for some third-party decoding methods:
- [https://github.com/polakowo/gpt2bot](https://github.com/polakowo/gpt2bot) GPT2Bot implementation based on telegram by polakowo, [ref](https://github.com/microsoft/DialoGPT/issues/3#issuecomment-573904419)
- [https://colab.research.google.com/drive/1PslHE4Rl4RqSa20s7HEp0ZKITBir6ezE](https://colab.research.google.com/drive/1PslHE4Rl4RqSa20s7HEp0ZKITBir6ezE) A colab interactive notebook by qywu,[ref](https://github.com/microsoft/DialoGPT/issues/3#issuecomment-551410203)
- [https://github.com/andreamad8/DialoGPT2-Interact](https://github.com/andreamad8/DialoGPT2-Interact) An interactive script featuring multiturn chatbot by andreamad8,[ref](https://github.com/microsoft/DialoGPT/issues/3#issuecomment-551450016)
- [https://github.com/LHolten/DialoGTP-MMI-decoder](https://github.com/LHolten/DialoGTP-MMI-decoder) An MMI implementation by LHolten,[ref](https://github.com/microsoft/DialoGPT/issues/3#issuecomment-558318401)
- [https://colab.research.google.com/drive/1-_KjlAV3J1IVDw_9KogjKDCzgFY7Jp7E](https://colab.research.google.com/drive/1-_KjlAV3J1IVDw_9KogjKDCzgFY7Jp7E) A colab interactive notebook by illuminascent@Reddit,[ref](https://www.reddit.com/r/MachineLearning/comments/dt5woy/p_dialogpt_state_of_the_art_conversational_model/?st=k530k3oo&sh=f6cd20fd)
- [https://gradio.app/g/AK391/DialoGPT](https://gradio.app/g/AK391/DialoGPT) A 3rd party demo by [AK391](https://github.com/AK391) using Gradio [web demo](https://gradio.app/g/AK391/DialoGPT)
## Models
We release 6 fine-tuned models which can be further fine-tuned on low-resource user-customized dataset. The total parameters in these models range from 117M to 762M, in accord with OpenAI GPT-2 model sizes.
| Model | Fine-tuned from GPT-2| Trained from scratch
|----------------------|--------|--------|
| DialoGPT 762M model| [\[link\]](https://acvrpublicycchen.blob.core.windows.net/dialogpt/multiref/large_ft.pkl) [\[huggingface model card\]](https://huggingface.co/microsoft/DialoGPT-large) | [\[link\]](https://acvrpublicycchen.blob.core.windows.net/dialogpt/multiref/large_fs.pkl) |
| DialoGPT 345M model| [\[link\]](https://acvrpublicycchen.blob.core.windows.net/dialogpt/multiref/medium_ft.pkl) [\[huggingface model card\]](https://huggingface.co/microsoft/DialoGPT-medium) | [\[link\]](https://acvrpublicycchen.blob.core.windows.net/dialogpt/multiref/medium_fs.pkl) |
| DialoGPT 117M model| [\[link\]](https://acvrpublicycchen.blob.core.windows.net/dialogpt/multiref/small_ft.pkl) [\[huggingface model card\]](https://huggingface.co/microsoft/DialoGPT-small)| [\[link\]](https://acvrpublicycchen.blob.core.windows.net/dialogpt/multiref/small_fs.pkl) |
| DialoGPT 345M model (reverse, for MMI)| [link](https://acvrpublicycchen.blob.core.windows.net/dialogpt/multiref/small_reverse.pkl) | -|
| [DialogRPT](https://github.com/golsun/DialogRPT) (**new** ranking models) | [link](https://github.com/golsun/DialogRPT) | -|
The model files can be loaded exactly as the GPT-2 model checkpoints from Huggingface's [Transformers](https://github.com/huggingface/transformers). You can find the corresponding configuration files (`merges.txt`, `config.json`, `vocab.json`) in DialoGPT's repo in `./configs/*`.
The reverse model is predicting the source from the target. This model is used for MMI reranking.
The [DialogRPT](https://github.com/golsun/DialogRPT) models our recently proposed ranking models used to predict the human feedback (upvotes, replies) of the responses. These models can be used to improve the DialoGPT generation quality (see our [EMNLP paper](https://arxiv.org/abs/2009.06978) for details).
## Retraining full models
### Data Preparation
The first step to retrain the full models is to generate the aforementioned 27GB Reddit dataset. This involves downloading full Reddit submission and comments dumps from [https://files.pushshift.io/reddit](https://files.pushshift.io/reddit) and creating intermediate files, which overall require **700GB of local disk space**. Downloading and processing the full data requires about 1-2 days, depending on your (CPU) compute capabilties (e.g., ~24 hours with 8 cores on a recent computer). Assuming you ran the above setup and installation steps (conda activate LSP, etc.), you can create the full dataset by running either:
```
python demo.py --data full
```
or
```
cd reddit_extractor; SIZE=full make -j 8; cd ..
```
The former command calls the latter, so the two methods are equivalent. We recommend the former, as the latter is mostly useful if you run into any problem or want to customize any arguments (e.g., the `make` command lets you build only a subset of the data). Note that the downloading phase can be error prone, for example based on your geolocation (firewall, etc.). If the above commands fail to generate `data/train.tsv`, or if that file is not anywhere close to 27GB, it means something went wrong. In that case, you may want to inspect `reddit_extractor/wget-log` and `reddit_extractor/logs/*.log` for any obvious error (e.g., wget unable to download from pushshift.io). If error messages don't make sense to you, feel free to contact us. If so, please be sure to include any error messages gathered from these log files.
Training data statistics: the generated training tsv file should be roughly 26.8 GB uncompressed, with 146.8M training instances, 3.87B source tokens, and 2.14B target tokens (including utterance-level 0/1 weights). The resulting train.tsv file should contain 146,846,215 lines.
### Training
We recommand generating the above data using the `demo.py --data full`, as it (1) generates the data, (2) converts it into DB format, and (3) trains a model using `python LSP_train.py`. Please directly edit `demo.py` if you want to customize any of the hyperparameters.
## Evaluations
#### DSTC-7 challenge
Our model achieved the state-of-the-art results in [DSTC-7 Challenge response generation task](https://github.com/mgalley/DSTC7-End-to-End-Conversation-Modeling).
| Experiment | NIST2 | NIST4 | BLEU2 | BLEU4 | METEOR | ENT-4 | DIST-1 | DIST-2 | Avg. Len |
|--------------------|-------|-------|--------|-------|--------|----------|------------|------------|---------|
| Human response | 2.62 | 2.65 | 12.35% | 3.13% | 8.31% | 10.45 | 16.66% | 67.01% | 18.8 |
| DSTC-7 Winner | 2.51 | 2.52 | 14.35% | 1.83% | 8.07% | 9.03 | 10.89% | 32.49% | 15.1 |
| DialoGPT 345M | 2.80 | 2.82 | 14.16% | 2.31% | 8.51% | **10.08** | 9.13% | 39.73% | 16.9 |
| DialoGPT 345M (BS) | **2.92** | **2.97** | **19.18%** | **6.05%** | **9.29%** | 9.57 | **15.73%** | **51.03%** | 14.2 |
where ENT represents the [Entropy score](https://arxiv.org/abs/1809.05972), and DIST represents the [Distinct score](https://arxiv.org/pdf/1510.03055.pdf). For all metrics except the average length, larger are better.
<!--| Experiment | NIST1 | NIST2 | NIST3 | NIST4 | BLEU1 | BLEU2 | BLEU3 | BLEU4 | METEOR | ENT-1 | ENT-2 | ENT-3 | ENT-4 | DIST-1 | DIST-2 | Len |
|----------------------|--------|--------|--------|--------|--------|--------|--------|--------|--------|----------|----------|----------|----------|------------|------------|---------|
| Human | 2.4237 | 2.6244 | 2.6472 | 2.65 | 0.3408 | 0.1235 | 0.0572 | 0.0313 | 0.0831 | 6.5893 | 9.7423 | 10.4101 | 10.4450 | 0.1666 | 0.6701 | 18.7568 |
| DSTC-7 Winner | 2.3408 | 2.5102 | 2.522 | 2.523 | 0.4122 | 0.1435 | 0.0501 | 0.0183 | 0.0807 | 5.3832 | 7.6065 | 8.5304 | 9.0298 | 0.1089 | 0.3249 | 15.1327 |
| DialoGPT | 2.5863 | 2.804 | 2.823 | 2.8246 | 0.3927 | 0.1416 | 0.0555 | 0.0231 | 0.0851 | 5.5791 | 8.5109 | 9.6872 | 10.0765 | 0.0913 | 0.3973 | 16.9484 |
| DialoGPT(beam search) | **2.5943**| **2.9163** | **2.9624** | **2.9681**| **0.4238** | **0.1918** | **0.1027** | **0.0605** | **0.0929** | **6.0815** | **8.7379** | 9.4037 | 9.5697 | 0.1573 | 0.5103 | 14.1603 |-->
Note that the superior automatic evaluation comparing to human responses does not necessary imply that our model achieves human parity. Please check out our paper for more detailed analysis.
To fine-tune the `345M` DialoGPT model on the DSTC-7 challenge data on a server with 8 V100 GPUs, please run the following commandline (The DSTC data can be found at [DSTC-7 repo](https://github.com/mgalley/DSTC7-End-to-End-Conversation-Modeling)):
```bash
python3 -m torch.distributed.launch --nproc_per_node=8 train_LSP.py --init_checkpoint ./models/medium/medium_ft.pkl --train_input_file ./data/DSTC_train.db --eval_input_file ./data/DSTC_valid.tsv --model_name_or_path ./model/medium/ --learning_rate 1e-4 --train_batch_size 64 --eval_batch_size 64 --no_token_id
```
The trained model can be found at [DSTC medium model](https://acvrpublicycchen.blob.core.windows.net/dialogpt/DSTC/medium_ft.pkl)
#### Evaluation
1. Please **downloads** the following 3rd-party packages and save into the empty folder `3rdparty`:
* [**mteval-v14c.pl**](https://goo.gl/YUFajQ) to compute [NIST](http://www.mt-archive.info/HLT-2002-Doddington.pdf). You may need to install the following [perl](https://www.perl.org/get.html) modules (e.g. by `cpan install`): XML:Twig, Sort:Naturally and String:Util.
* [**meteor-1.5**](http://www.cs.cmu.edu/~alavie/METEOR/download/meteor-1.5.tar.gz) to compute [METEOR](http://www.cs.cmu.edu/~alavie/METEOR/index.html). It requires [Java](https://www.java.com/en/download/help/download_options.xml).
2. Please follow the [DSTC-7 official repo](https://github.com/mgalley/DSTC7-End-to-End-Conversation-Modeling/tree/master/data_extraction) to extract the data, and put `data-official-test/test.refs.txt` into `./dstc/data/` folder.
3. Run the extraction script below to produce the human response hypothesis file `human.resp.txt`:
```bash
python extract_human.py
```
4. Finally, to reproduce the results of human hypothesis on DSTC dataset, please run following commands under the repo folder:
```bash
python batch_eval.py
```
The evaluation results will be generated in the folder `./dstc/eval/`
## 6K multi-ref dataset result
### Automatic evaluation
We test on 6K multi-ref dataset from Reddit. The results are summarized in below
| Experiment | NIST2 | NIST4 | BLEU2 | BLEU4 | METEOR | ENT-4 | DIST-1 | DIST-2 | Avg. Len |
|--------------------|-------|-------|--------|-------|--------|----------|------------|------------|---------|
| Human response | 3.41 | 4.25 | 17.90% | 7.48% | 10.64% | 11 | 14.50% | 63.00% | 13.1 |
| DialoGPT 117M | 2.39 | 2.41 | 10.54% | 1.55% | 7.53% | 10.78 | 8.60% | 39.90% | 12.8 |
| DialoGPT 345M | 3 | 3.06 | 16.96% | 4.56% | 9.81% | 9.13 | 6.80% | 26.30% | 12.2 |
| DialoGPT 762M | 2.84 | 2.9 | 18.66% | 5.25% | 9.66% | 9.72 | 7.76% | 29.93% | 11.2 |
| DialoGPT 345M (BS) | **3.4** | **3.5** | **21.76%** | **7.92%** | 10.74% | 10.48 | **12.38%** | **48.74%** | 11.3 |
| DialoGPT 345M (w/MMI)| 3.28 | 3.33 | 15.68% | 3.94% | **11.23%** | **11.25** | 9.39% | 45.55% | 17.2 |
### <a name="human_eval"></a>Human evaluation
We further conduct human evaluations (6K examples for each methods, each example is evaluated by 3 human judges). The results show a strong evidence that our generation quality is towards approaching the quality of real human responses, under this non-interactive Turing test:
*Relevance*: A and B, which one is more relevant to the source prompt.
| System A | A Wins (%) | Ties (%) | B Wins (%) | System B|
|--------------------|-------|-------|--------|-------|
|DialoGPT 345M|2671 (45%) | 513 (9%) | 2816 (47%)| Human responses|
|DialoGPT 345M| 3281 (72%)| 394 (9%) | 882 (19%)| [PersonalityChat](https://docs.microsoft.com/en-us/azure/cognitive-services/project-personality-chat/overview)|
|DialoGPT 345M w/ MMI| **2871** (48%)| 522 (9%) | 2607 (43%)| Human responses|
*Informativeness*: A and B, which one is more contentful and informative.
| System A | A Wins (%) | Ties (%) | B Wins (%) | System B|
|--------------------|-------|-------|--------|-------|
|DialoGPT 345M| 2722 (45%) | 234 (4%) | 3044 (51%)| Human responses|
|DialoGPT 345M|3490 (77%) | 206 (5%) | 861 (19%)| [PersonalityChat](https://docs.microsoft.com/en-us/azure/cognitive-services/project-personality-chat/overview)|
|DialoGPT 345M w/ MMI| **3011** (50%)| 234 (4%) | 2755 (46%)| Human responses|
*Human-Like*: A and B, which one do you think is more likely to be generated by Human.
| System A | A Wins (%) | Ties (%) | B Wins (%) | System B|
|--------------------|-------|-------|--------|-------|
|DialoGPT 345M|2716 (45%) | 263 (4%) | 3021 (50%)| Human responses|
|DialoGPT 345M|3462 (76%) | 196 (4%) | 899 (20%)| [PersonalityChat](https://docs.microsoft.com/en-us/azure/cognitive-services/project-personality-chat/overview)|
|DialoGPT 345M w/ MMI| **2978** (50%)| 241 (4%) | 2781 (46%)| Human responses|
Please see full details in our [arxiv paper](https://arxiv.org/abs/1911.00536).
<!--Relevance
System Wins (%) Ties (%) Losses (%)
2 vs 1 2671 (0.45) 513 (0.09) 2816 (0.47)
2 vs 3 3281 (0.72) 394 (0.09) 882 (0.19)
2 vs 4 2379 (0.40) 527 (0.09) 3094 (0.52)
2 vs 5 3019 (0.50) 581 (0.10) 2400 (0.40)
2 vs 6 2726 (0.45) 576 (0.10) 2698 (0.45)
Informativeness
System Wins (%) Ties (%) Losses (%)
2 vs 1 2722 (0.45) 234 (0.04) 3044 (0.51)
2 vs 3 3490 (0.77) 206 (0.05) 861 (0.19)
2 vs 4 2474 (0.41) 257 (0.04) 3269 (0.54)
2 vs 5 3230 (0.54) 362 (0.06) 2408 (0.40)
2 vs 6 2856 (0.48) 303 (0.05) 2841 (0.47)
Human-Like
System Wins (%) Ties (%) Losses (%)
2 vs 1 2716 (0.45) 263 (0.04) 3021 (0.50)
2 vs 3 3462 (0.76) 196 (0.04) 899 (0.20)
2 vs 4 2478 (0.41) 289 (0.05) 3233 (0.54)
2 vs 5 3233 (0.54) 340 (0.06) 2427 (0.40)
2 vs 6 2847 (0.47) 321 (0.05) 2832 (0.47)
-->
<!--| Experiment | NIST1 | NIST2 | NIST3 | NIST4 | BLEU1 | BLEU2 | BLEU3 | BLEU4 | METEOR | ENT-4 | DIST-1 | DIST-2 |
|------------------------------|-------|-------|-------|-------|--------|--------|--------|-------|--------|----------|------------|------------|
| Human response | 2.99 | 3.41 | 3.83 | 4.25 | 39.61% | 17.90% | 10.71% | 7.48% | 10.64% | 11 | 14.50% | 63.00% |
| DialoGPT 117M | 2.25 | 2.39 | 2.41 | 2.41 | 35.43% | 10.54% | 3.85% | 1.55% | 7.53% | 10.78 | 8.60% | 39.90% |
| DialoGPT 345M |**2.67** | **3** | **3.05** | **3.06** | 40.97% | 16.96% | 8.31% | 4.56% | **9.81%** | 9.13 | 6.80% | 26.30% |
| DialoGPT 762M | 2.51 | 2.84 | 2.89 | 2.9 | **44.45%** | **18.66%** | **9.32%** | **5.25%** | 9.66% | **9.72** | **7.76%** | **29.93%** |-->
<!--| DialoGPT 762M (beam search)| 2.51 | 2.9 | 2.97 | 2.98 | 44.84% | 21.08% | 12.01% | 7.57% | 10.11% | 10.06 | 11.62% | 44.07% |-->
<!--## ConvAI challenge -->
<!--| DialoGPT 345M (beam search) | 2.94 | 3.4 | 3.49 | 3.5 | 45.27% | 21.76% | 12.51% | 7.92% | 10.74% | 10.48 | 12.38% | 48.74% |-->
## Generated Examples
* Multi-turn generation examples from an interactive environment:
|Role | Response |
|---------|--------|
|User | Does money buy happiness? |
| Bot | Depends how much money you spend on it .|
|User | What is the best way to buy happiness ? |
| Bot | You just have to be a millionaire by your early 20s, then you can be happy . |
|User |This is so difficult ! |
| Bot | You have no idea how hard it is to be a millionaire and happy . There is a reason the rich have a lot of money |
* Multi-turn self-playing dialogue with user prompt
|Role | Response |
|---------|--------|
|User | what is the meaning of life ?|
| Bot | The meaning is to be with your family and friends .|
|Bot | I’m going to guess : It means that your parents and friends have loved you . That’s the meaning , I believe . |
| Bot | I think the meaning is to make sure you’re alive when you die . So you’ll be with your family , friends , and yourself .|
|Bot |It’s not , but thanks for trying to help . |
* Commonsense question answering examples
|User | Bot |
|---------|--------|
|who is the first president of the United States | George Washington |
|what is the boiling point of water? | I think it's about 212 F. |
|which one is bigger, sun or moon? | The sun . |
|which animal has black and white stripes? | A black and white striped cat . |
* Handling context that does not appear in training examples (note that training examples are sampled from 2005 to 2017)
|User | Bot |
|---------|--------|
|The trading war between China and US is still happening . | I'm sure the Chinese government and the U . S . will be fine with that as long as they keep their arms on the table and the money flowing . |
|who won the world cup in 2018 ? | Germany , I believe . |
|Nvidia's Titan RTX is really good . | I have this card and i love it . No complaints . |
|Can Porsche beat Tesla with its new Taycan EV ? | I don't think so . The Tesla is more powerful than the porsche .
* Our bot's response to philosophy questions
|User | Bot |
|---------|--------|
|What is the meaning of a good life ? | I think it's that we're all connected to our past lives and the meaning of life is to live the way you want to and how you want to . |
|Which is more important -- being successful or being happy ? | Being successful is just endless . Being happy is the only thing that matters . |
|Is peace the only way to stop war ? | Not if the world is not willing to work with you . Peace is the ultimate goal , but peace requires more than that . |
| How much freedom should people have ? | The right amount of freedom is the freedom to do as you please , as long as you don't hurt people or property .|
* Finally
|User | Bot |
|---------|--------|
|What do you think of the DialoGPT repo ? | It's pretty good , but it's very easy to find a bug . |
Please start a issue if you spot any :)
## Try our system
The live demo and decoding script access is upon invitation only now. Please stayed tuned for the full release.
<!--The live demo and decoding script access is upon approval request. Please apply [here](https://docs.google.com/spreadsheets/d/1epTNfaqva1isVO_o9pbyhVLsnzDn58dGkcLB0OUVcqs/edit?usp=sharing)-->
<!--This model should give a Hits@1 over 79, perplexity of 20.5 and F1 of 16.5 using the convai2 evaluation script (see below).
These numbers are slightly lower than the number we obtained in the ConvAI2 competition. Here is what you can tweak to reach the same results:
- in the ConvAI2 competition we also used tweaked position emebddings so that the history of the dialog always start at with the same embeddings. This is easy to add with pytorch-pretrained-bert and should improve the hits@1 metric.
- in the ConvAI2 competition we used a beam search decoder. While the results are better in term of f1 metric, our feeling is that the human experience is les compelling with beam search versus the nucleus sampling detector which is provided in the present repository.-->
<!--## Using the interaction script
The training script saves all the experiments and checkpoints in a sub-folder named with the timestamp of the experiment in the `./runs` folder of the repository base folder.
You can then use the interactive script to interact with the model simply by pointing to this folder.
Here is an example command line to run the interactive script:
```bash
python ./interact.py --model_checkpoint ./data/Apr17_13-31-38_thunder/ # run the interactive script with a training checkpoint
python ./interact.py # run the interactive script with the finetuned model on our S3
```
The fine-tuned model will gives FINAL Hits@1: 0.715
The interactive script accept a few arguments to tweak the decoding algorithm:
Argument | Type | Default value | Description
---------|------|---------------|------------
dataset_path | `str` | `""` | Path or url of the dataset. If empty download from S3.
dataset_cache | `str` | `'./dataset_cache.bin'` | Path or url of the dataset cache
model | `str` | `"openai-gpt"` | Path, url or short name of the model
max_history | `int` | `2` | Number of previous utterances to keep in history
device | `str` | `cuda` if `torch.cuda.is_available()` else `cpu` | Device (cuda or cpu)
no_sample | action `store_true` | Set to use greedy decoding instead of sampling
max_length | `int` | `20` | Maximum length of the output utterances
min_length | `int` | `1` | Minimum length of the output utterances
seed | `int` | `42` | Seed
temperature | `int` | `0.7` | Sampling softmax temperature
top_k | `int` | `0` | Filter top-k tokens before sampling (`<=0`: no filtering)
top_p | `float` | `0.9` | Nucleus filtering (top-p) before sampling (`<=0.0`: no filtering)
## Running ConvAI2 evaluation scripts
To run the evaluation scripts of the ConvAI2 challenge, you first need to install `ParlAI` in the repo base folder like this:
```bash
git clone https://github.com/facebookresearch/ParlAI.git
cd ParlAI
python setup.py develop
```
You can then run the evaluation script from `ParlAI` base folder:
```bash
cd ParlAI
python ../convai_evaluation.py --eval_type hits@1 # to download and evaluate our fine-tuned model on hits@1 metric
python ../convai_evaluation.py --eval_type hits@1 --model_checkpoint ./data/Apr17_13-31-38_thunder/ # to evaluate a training checkpoint on hits@1 metric
```
The evaluation script accept a few arguments to select the evaluation metric and tweak the decoding algorithm:
Argument | Type | Default value | Description
---------|------|---------------|------------
eval_type | `str` | `"hits@1"` | Evaluate the model on `hits@1`, `ppl` or `f1` metric on the ConvAI2 validation dataset
model | `str` | `"openai-gpt"` | Path, url or short name of the model
max_history | `int` | `2` | Number of previous utterances to keep in history
device | `str` | `cuda` if `torch.cuda.is_available()` else `cpu` | Device (cuda or cpu)
no_sample | action `store_true` | Set to use greedy decoding instead of sampling
max_length | `int` | `20` | Maximum length of the output utterances
min_length | `int` | `1` | Minimum length of the output utterances
seed | `int` | `42` | Seed
temperature | `int` | `0.7` | Sampling softmax temperature
top_k | `int` | `0` | Filter top-k tokens before sampling (`<=0`: no filtering)
top_p | `float` | `0.9` | Nucleus filtering (top-p) before sampling (`<=0.0`: no filtering)
-->
## Related Project
* RetGen: [https://github.com/dreasysnail/RetGen](https://github.com/dreasysnail/RetGen). Retrieval-augmented/grounded DialoGPT and beyond. RetGen is a joint training framework that simultaneously optimizes a dense passage retriever and a knowledge-grounded text generator in an end-to-end fashion.
* Microsoft ICECAPS: [https://github.com/microsoft/icecaps](https://github.com/microsoft/icecaps).
As an orthogonal repository of this project,
Microsoft Icecaps is an open-source toolkit (in tensorflow) for building neural conversational systems. Icecaps provides an array of tools from recent conversation modeling and general NLP literature within a flexible paradigm that enables complex multi-task learning setups.
* Pretrained UniLM: [https://github.com/microsoft/unilm](https://github.com/microsoft/unilm)
* MT-DNN: [https://github.com/namisan/mt-dnn](https://github.com/namisan/mt-dnn)
* A chinese counterpart of DialoGPT by yangjianxin1. [https://github.com/yangjianxin1/GPT2-chitchat](https://github.com/yangjianxin1/GPT2-chitchat). We are glad to see that the MMI strategy that we used in DialoGPT has also improved the performance for this project as well!
## Contact
Please contact [DialoGPT@microsoft.com](mailto:DialoGPT@microsoft.com) if you have any questions/suggestions. However, the response will be sporadic. Please expect delay.
## Contributing
This project welcomes contributions and suggestions. Most contributions require you to agree to a
Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us
the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com.
When you submit a pull request, a CLA bot will automatically determine whether you need to provide
a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions
provided by the bot. You will only need to do this once across all repos using our CLA.
This project has adopted the [Microsoft Open Source Code of Conduct](https://opensource.microsoft.com/codeofconduct/).
For more information see the [Code of Conduct FAQ](https://opensource.microsoft.com/codeofconduct/faq/) or
contact [opencode@microsoft.com](mailto:opencode@microsoft.com) with any additional questions or comments.
## Disclaimer
This repository aims to facilitate research in large-scale pretraining for conversational data. This toolkit contains only part of the modeling machinery needed to actually produce a model weight file in a running dialog. On its own, this model provides only information about the weights of various text spans; in order for a researcher to actually use it, they will need to bring conversational data of their own and decode the response generation from the pretrained system. Microsoft is not responsible for any generation from the 3rd party utilization of the pretrained system.
## Citation
If you use this code in your research, you can cite our [arxiv paper](https://arxiv.org/abs/1911.00536):
```bash
@inproceedings{zhang2019dialogpt,
title={DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation},
author={Yizhe Zhang and Siqi Sun and Michel Galley and Yen-Chun Chen and Chris Brockett and Xiang Gao and Jianfeng Gao and Jingjing Liu and Bill Dolan},
year={2020},
booktitle={ACL, system demonstration}
}
```
|