File size: 14,662 Bytes
cdda44a |
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 |
---
language:
- en
tag:
- doc-chat
- DocChat
- dragon
- retriever
- conversation
- multi-turn
- conversational query
license:
- other
---
# Model Information
We are excited to announce the release of Cerebras DocChat, our first iteration of models designed for document-based conversational question answering. This series includes two models: Cerebras Llama3-DocChat, a large language model (LLM), and Cerebras Dragon-DocChat, a multi-turn retriever model.
This model – Cerebras Dragon-DocChat – was built on top of Dragon+ and finetuned using ChatQA’s conversational Q&A dataset. The finetuning process took *minutes* on 1 Cerebras machine (yes you read that correctly), and yields 8.9% and 3.5% absolute improvements in recall over Dragon+ and ChatQA Dragon-Multiturn (top-1).
**Note that like its predecessors, Cerebras Dragon-DocChat is a dual-encoder embedding model. This model card is for the context encoder. You can find the question encoder's model card here.**
You can find more information about DocChat at the following locations:
* [Blog post](https://www.cerebras.net/blog/train-a-gpt-4-level-conversational-qa-in-a-few-hours)
* [LLM model weights on HuggingFace](https://huggingface.co/cerebras/Llama3-DocChat-1.0-8B)
* Embedding model weights on HuggingFace: [Query Encoder](https://huggingface.co/cerebras/Dragon-DocChat-Query-Encoder), [Context Encoder](https://huggingface.co/cerebras/Dragon-DocChat-Context-Encoder)
* [Data preparation, training, and evaluation code](https://github.com/Cerebras/DocChat)
## Results
DocChat Retriever was evaluated on five multi-turn QA datasets.
<style>
table {
border: none;
border-collapse: collapse;
border-spacing: 0;
}
table td {
border-style: solid;
border-width: 1px;
padding: 2px 5px;
text-align: center;
border-color: #7f8c8d;
}
</style>
<table>
<tbody>
<tr>
<td rowspan="1" colspan="1">
<div data-align="center">
<strong data-renderer-mark="true">Benchmark</strong>
</div>
</td>
<td rowspan="1" colspan="1">
<strong data-renderer-mark="true">Metric</strong>
</td>
<td rowspan="1" colspan="1">
<strong data-renderer-mark="true">Facebook Dragon+</strong>
</td>
<td rowspan="1" colspan="1">
<strong data-renderer-mark="true">Nvidia Dragon-Multiturn</strong>
</td>
<td rowspan="1" colspan="1">
<strong data-renderer-mark="true">Cerebras Dragon-DocChat</strong>
</td>
</tr>
<tr>
<td rowspan="3" colspan="1">
Doc2Dial
</td>
<td rowspan="1" colspan="1">
Recall@1
</td>
<td rowspan="1" colspan="1">
43.95
</td>
<td rowspan="1" colspan="1">
50.11
</td>
<td rowspan="1" colspan="1">
51.54
</td>
</tr>
<tr>
<td rowspan="1" colspan="1">
Recall@5
</td>
<td rowspan="1" colspan="1">
77.61
</td>
<td rowspan="1" colspan="1">
83.85
</td>
<td rowspan="1" colspan="1">
83.12
</td>
</tr>
<tr>
<td rowspan="1" colspan="1">
Recall@20
</td>
<td rowspan="1" colspan="1">
92.05
</td>
<td rowspan="1" colspan="1">
95.33
</td>
<td rowspan="1" colspan="1">
95.25
</td>
</tr>
<tr>
<td rowspan="3" colspan="1">
QuAC
</td>
<td rowspan="1" colspan="1">
Recall@1
</td>
<td rowspan="1" colspan="1">
62.09
</td>
<td rowspan="1" colspan="1">
60.02
</td>
<td rowspan="1" colspan="1">
61.30
</td>
</tr>
<tr>
<td rowspan="1" colspan="1">
Recall@5
</td>
<td rowspan="1" colspan="1">
86.01
</td>
<td rowspan="1" colspan="1">
86.51
</td>
<td rowspan="1" colspan="1">
87.69
</td>
</tr>
<tr>
<td rowspan="1" colspan="1">
Recall@20
</td>
<td rowspan="1" colspan="1">
96.48
</td>
<td rowspan="1" colspan="1">
96.60
</td>
<td rowspan="1" colspan="1">
97.25
</td>
</tr>
<tr>
<td rowspan="3" colspan="1">
QReCC
</td>
<td rowspan="1" colspan="1">
Recall@1
</td>
<td rowspan="1" colspan="1">
49.00
</td>
<td rowspan="1" colspan="1">
49.43
</td>
<td rowspan="1" colspan="1">
55.41
</td>
</tr>
<tr>
<td rowspan="1" colspan="1">
Recall@5
</td>
<td rowspan="1" colspan="1">
85.14
</td>
<td rowspan="1" colspan="1">
86.6
</td>
<td rowspan="1" colspan="1">
90.11
</td>
</tr>
<tr>
<td rowspan="1" colspan="1">
Recall@20
</td>
<td rowspan="1" colspan="1">
97.21
</td>
<td rowspan="1" colspan="1">
98.28
</td>
<td rowspan="1" colspan="1">
98.39
</td>
</tr>
<tr>
<td rowspan="3" colspan="1">
INSCIT*
</td>
<td rowspan="1" colspan="1">
Recall@1
</td>
<td rowspan="1" colspan="1">
11.13
</td>
<td rowspan="1" colspan="1">
18.35
</td>
<td rowspan="1" colspan="1">
21.65
</td>
</tr>
<tr>
<td rowspan="1" colspan="1">
Recall@5
</td>
<td rowspan="1" colspan="1">
29.27
</td>
<td rowspan="1" colspan="1">
48.45
</td>
<td rowspan="1" colspan="1">
50.72
</td>
</tr>
<tr>
<td rowspan="1" colspan="1">
Recall@20
</td>
<td rowspan="1" colspan="1">
49.07
</td>
<td rowspan="1" colspan="1">
66.19
</td>
<td rowspan="1" colspan="1">
72.78
</td>
</tr>
<tr>
<td rowspan="3" colspan="1">
Topiocqa*
</td>
<td rowspan="1" colspan="1">
Recall@1
</td>
<td rowspan="1" colspan="1">
29.19
</td>
<td rowspan="1" colspan="1">
31.34
</td>
<td rowspan="1" colspan="1">
38.19
</td>
</tr>
<tr>
<td rowspan="1" colspan="1">
Recall@5
</td>
<td rowspan="1" colspan="1">
62.52
</td>
<td rowspan="1" colspan="1">
65.79
</td>
<td rowspan="1" colspan="1">
72.47
</td>
</tr>
<tr>
<td rowspan="1" colspan="1">
Recall@20
</td>
<td rowspan="1" colspan="1">
83.69
</td>
<td rowspan="1" colspan="1">
84.37
</td>
<td rowspan="1" colspan="1">
87.23
</td>
</tr>
<tr>
<td rowspan="2" colspan="1">
Average**
</td>
<td rowspan="1" colspan="1">
Avg top 1
</td>
<td rowspan="1" colspan="1">
49.36
</td>
<td rowspan="1" colspan="1">
54.76
</td>
<td rowspan="1" colspan="1">
58.29
</td>
</tr>
<tr>
<td rowspan="1" colspan="1">
Avg top 5
</td>
<td rowspan="1" colspan="1">
76.30
</td>
<td rowspan="1" colspan="1">
81.50
</td>
<td rowspan="1" colspan="1">
84.19
</td>
</tr>
</tbody>
</table>
\*Evaluated on a subset of the wikipedia corpus that was available to us. All models use the same evaluation strategy to ensure apples-to-apples comparisons.
\*\* We follow the same convention as in ChatQA, where we compare top-5 and top-20 of TopiOCQA and INSCIT to top-1 and top-5, respectively, of the other datasets, in order match differences in average context length.
## Prompt Format
In order to maintain consistency, we follow the same format as ChatQA. The chat history should be concatenated into a single query, as follows:
```
user: {user turn 1}
agent: {agent turn 1}
...
user: {current user turn}
```
## Example Usages
```python
# Demonstration of DocChat retriever in a multi-turn setting
# The sample documents are from a spec sheet about the Cerebras system & supercomputers
from transformers import AutoTokenizer, AutoModel
import torch
tokenizer = AutoTokenizer.from_pretrained("cerebras/Dragon-DocChat-Query-Encoder")
query_encoder = AutoModel.from_pretrained("cerebras/Dragon-DocChat-Query-Encoder")
context_encoder = AutoModel.from_pretrained("cerebras/Dragon-DocChat-Context-Encoder")
documents = []
documents.append("""
# Cerebras Wafer-Scale Cluster
The Cerebras Wafer-Scale Cluster (WSC) is a revolutionary technology suite that efficiently handles the enormous computational needs of AI model training. It centers around the CS-3 system, powered by the 3rd generation Wafer-Scale Engine (WSE-3)—the world’s largest AI-optimized processor. The WSC integrates MemoryX for high-capacity, off-chip model weight storage, and SwarmX for effective weight broadcasting and gradient reduction across the cluster. This setup allows the WSC to adeptly train multi-trillion parameter models, achieving near perfect linear-scaled performance and simplifying the complexity seen in traditional distributed computing.
The Cerebras WSE-3 is 46,250 square millimeters of silicon, 4 trillion transistors, 900K cores, 44 GB on-chip memory, and delivers an unparalleled 125 petaFLOPS of AI compute. It surpasses all other processors in AI-optimized cores, memory speed, and on-chip fabric bandwidth.
""")
documents.append("""
## AI Supercomputers
Condor Galaxy (CG), the supercomputer built by G42 and Cerebras, is the simplest and fastest way to build AI models in the cloud. With over 16 ExaFLOPs of AI compute, Condor Galaxy trains the most demanding models in hours rather than days. The terabyte scale MemoryX system natively accommodates 100 billion+ parameter models, making large scale training simple and efficient.
| Cluster | ExaFLOPs | Systems | Memory |
| -------- | -------- | -------- | ------ |
| CG1 | 4 | 64 CS-2s | 82 TB |
| CG2 | 4 | 64 CS-2s | 82 TB |
| CG3 | 8 | 64 CS-3s | 108 TB |
""")
query = [
{"role": "user", "content": "How many cores does a WSE-3 have?"},
{"role": "agent", "content": "WSE-3 has 900k cores."},
{"role": "user", "content": "What is Condor Galaxy?"}
]
formatted_query = "\n".join([turn["role"] + ": " + turn["content"] for turn in query]).strip()
query_input = tokenizer(formatted_query, return_tensors='pt')
ctx_input = tokenizer(documents, padding=True, truncation=True, max_length=512, return_tensors='pt')
query_emb = query_encoder(**query_input).last_hidden_state[:, 0, :]
ctx_emb = context_encoder(**ctx_input).last_hidden_state[:, 0, :]
## Compute similarity scores:
similarities = query_emb.matmul(ctx_emb.transpose(0, 1)) # (1, num_ctx)
## Rank the similarity from highest to lowest
ranked_results = torch.argsort(similarities, dim=-1, descending=True) # (1, num_ctx)
for i, doc_idx in enumerate(ranked_results[0].tolist()):
print(f"Rank {i}th document:")
print("-" * 80)
print(documents[doc_idx])
print()
```
## License
This model was trained from Dragon+, and therefore is subject to [its license](https://github.com/facebookresearch/dpr-scale/blob/main/LICENSE). Furthermore, it is trained on ChatQA's synthetic conversational QA dataset which was generated using GPT-4. As a result this model can be used for non-commercial purposes only, and is subject to [Terms of Use](https://openai.com/policies/terms-of-use) of the data generated by OpenAI.
## Acknowledgements
DocChat was built on top of a large body of ML work, spanning training datasets, recipes, and evaluation. We want to thank each of these resources.
```
@article{lin_dragon_2023,
title = How to Train Your DRAGON: Diverse Augmentation Towards Generalizable Dense Retrieval,
url = {https://arxiv.org/abs/2302.07452},
publisher = {arXiv},
journal = {arXiv e-print 2302.07452},
author = {Sheng-Chieh Lin, Akari Asai, Minghan Li, Barlas Oguz, Jimmy Lin, Yashar Mehdad, Wen-tau Yih, Xilun Chen},
year = {2023},
}
@inproceedings{feng2020doc2dial,
title={doc2dial: A Goal-Oriented Document-Grounded Dialogue Dataset},
author={Feng, Song and Wan, Hui and Gunasekara, Chulaka and Patel, Siva and Joshi, Sachindra and Lastras, Luis},
booktitle={Proceedings of the 2020 Conference on EMNLP},
year={2020}
}
@inproceedings{choi2018quac,
title={QuAC: Question Answering in Context},
author={Choi, Eunsol and He, He and Iyyer, Mohit and Yatskar, Mark and Yih, Wen-tau and Choi, Yejin and Liang, Percy and Zettlemoyer, Luke},
booktitle={Proceedings of the 2018 Conference on EMNLP},
year={2018}
}
@inproceedings{anantha2021open,
title={Open-Domain Question Answering Goes Conversational via Question Rewriting},
author={Anantha, Raviteja and Vakulenko, Svitlana and Tu, Zhucheng and Longpre, Shayne and Pulman, Stephen and Chappidi, Srinivas},
booktitle={Proceedings of the 2021 Conference on NAACL},
year={2021}
}
@article{adlakha2022topiocqa,
title={TopiOCQA: Open-domain Conversational Question Answering with Topic Switching},
author={Adlakha, Vaibhav and Dhuliawala, Shehzaad and Suleman, Kaheer and de Vries, Harm and Reddy, Siva},
journal={Transactions of the Association for Computational Linguistics},
year={2022}
}
@article{wu2023inscit,
title={InSCIt: Information-Seeking Conversations with Mixed-Initiative Interactions},
author={Wu, Zeqiu and Parish, Ryu and Cheng, Hao and Min, Sewon and Ammanabrolu, Prithviraj and Ostendorf, Mari and Hajishirzi, Hannaneh},
journal={Transactions of the Association for Computational Linguistics},
year={2023}
}
```
|