-
Notifications
You must be signed in to change notification settings - Fork 2
Expand file tree
/
Copy pathindex.html
More file actions
343 lines (329 loc) · 14.4 KB
/
index.html
File metadata and controls
343 lines (329 loc) · 14.4 KB
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
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<title>DeepSeek Coder</title>
<!-- Global site tag (gtag.js) - Google Analytics -->
<script async src="https://www.googletagmanager.com/gtag/js?id=G-PYVRSFMDRL"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag() {
dataLayer.push(arguments);
}
gtag('js', new Date());
gtag('config', 'G-PYVRSFMDRL');
</script>
<style>
.center {
display: block;
margin-left: auto;
margin-right: auto;
}
</style>
<style>
.image-container {
display: flex;
}
.image-container img {
width: 50%; /* 调整图像的宽度以适应页面 */
height: auto; /* 保持图像比例 */
}
</style>
<style>
ul.custom-list {
list-style-type: none; /* Removes default bullets */
padding: 0;
}
ul.custom-list li {
padding-left: 20px; /* Adds spacing before the text */
position: relative; /* Required for absolute positioning of the custom marker */
}
ul.custom-list li::before {
content: '•'; /* This is your custom marker */
/**color: blue; /* Marker color */
font-size: 20px; /* Marker size */
position: absolute; /* Positions the marker relative to the li element */
left: 0; /* Aligns the marker to the left */
top: 0; /* Aligns the marker to the top */
}
</style>
<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro" rel="stylesheet">
<link rel="stylesheet" href="./static/css/bulma.min.css">
<link rel="stylesheet" href="./static/css/bulma-carousel.min.css">
<link rel="stylesheet" href="./static/css/bulma-slider.min.css">
<link rel="stylesheet" href="./static/css/fontawesome.all.min.css">
<link rel="stylesheet" href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
<link rel="stylesheet" href="./static/css/index.css">
<link rel="icon" href="./static/images/logo.jpeg">
<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
<script defer src="./static/js/fontawesome.all.min.js"></script>
<script src="./static/js/bulma-carousel.min.js"></script>
<script src="./static/js/bulma-slider.min.js"></script>
<script src="./static/js/index.js"></script>
</head>
<body>
<section class="hero teaser">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-2 publication-title">DeepSeek Coder:<br>Let the Code Write Itself</h1>
<div class="is-size-4 publication-authors">
<span class="author-block">
Developed by <a href="https://www.deepseek.com/">DeepSeek AI</a>
</div>
<div class="column has-text-centered">
<div class="publication-links">
<!-- PDF Link. -->
<span class="link-block">
<a href="https://coder.deepseek.com" class="external-link button is-normal is-rounded is-light">
<span class="icon"> <img src="static/images/home.png" width="20px"> </span>
<span>Chat with DeepSeek Coder</span>
</a>
</span>
<span class="link-block">
<a href="https://arxiv.org/pdf/2401.14196.pdf" class="external-link button is-normal is-rounded is-light">
<span>📑 Technical Report</span>
</a>
</span>
<span class="link-block">
<a href="https://github.com/deepseek-ai/deepseek-coder/" class="external-link button is-normal is-rounded is-light">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span>Github</span>
</a>
</span>
<!-- Video Link. -->
<span class="link-block">
<a href="https://huggingface.co/deepseek-ai" class="external-link button is-normal is-rounded is-light">
<span class="icon"> <img src="static/images/huggingface_logo.svg" width="20px"> </span>
<span>HuggingFace</span>
</a>
</span>
<span class="link-block">
<a href="https://discord.gg/Tc7c45Zzu5" class="external-link button is-normal is-rounded is-light">
<span>Discord</span>
</a>
</span>
<span class="link-block">
<a href="https://github.com/guoday/assert/blob/main/QR.png?raw=true" class="external-link button is-normal is-rounded is-light">
<span>Wechat(微信)</span>
</a>
</span>
</div>
</div>
</div>
</div>
<!--div class="hero-body">
<h3 text-align="left">We introduce <b>DeepSeek Coder</b>, a series of large language model for coding:</h3>
<ul>
<li>
- Pretrained on <b>2 Trillion</b> tokens over more than 80 programming languages.
</li>
<li>
- Two versions, <b>Base</b> and <b>Instruct</b> and various model sizes to suit different needs: <b>1B</b>, <b>7B</b>, and <b>33B</b>.
</li>
<li>
- <b>Open source </b>and<b> free for research and commercial use</b>.
</li>
<li>
- Features a <b>16K window </b>size, supporting <b>repo-level</b> code completion and infilling.
</li>
<li>
- <b>State-of-the-Art</b> performance among open code LLMs. The 7B base model reaches the performance of CodeLLama-34B.
</li>
</ul>
<img src="static/images/result3.png" alt="Description of the GIF" class="center" width="75%" />
<h4 class="subtitle has-text-centered">
State-of-the-art Perforamnce on Multilingual HumanEval
</h4>
<img src="static/images/completion_demo.gif" alt="Description of the GIF" />
<h4 class="subtitle has-text-centered">
DeepSeek Coder empowers users to code efficiently and solve various computer problems easily
</h4>
</div-->
</div>
</div>
</section>
<section class="hero teaser">
<div class="container is-max-desktop">
<!-- Abstract. -->
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Abstract</h2>
<div class="content has-text-justified">
<p>
DeepSeek Coder comprises a series of code language models trained from scratch on both 87% code and 13% natural language in English and Chinese, with each model pre-trained on 2T tokens.
We provide various sizes of the code model, ranging from 1B to 33B versions.
Each model is pre-trained on repo-level code corpus by employing a window size of 16K and a extra fill-in-the-blank task, resulting in foundational models (DeepSeek-Coder-Base).
We further fine-tune the base model with 2B tokens of instruction data to get instruction-tuned models, namedly DeepSeek-Coder-Instruct.
</p>
<!-- <p>
For coding capabilities, DeepSeek-Coder-Base achieves state-of-the-art performance among open-source code models on multiple programming languages and various benchmarks.
And Compared to GPT35-turbo, DeepSeek-Coder-Instruct demonstrates superior performance in human evaluation while maintaining comparable performance in MBPP.
</p> -->
<ul>
<li>
Pretrained on <b>2 Trillion</b> tokens over more than 80 programming languages.
</li>
<li>
Various model sizes (<b>1.3B</b>, <b>5.7B</b>, <b>6.7B</b> and <b>33B</b>) to support different requirements.
</li>
<li>
A window size of <b>16K window </b>size, supporting <b>project-level</b> code completion and infilling.
</li>
<li>
<b>State-of-the-Art</b> performance among open code models.
</li>
<li>
<b>Open source and free for research and commercial use</b>.
</li>
</ul>
</div>
</div>
</div>
<img src="static/images/code_chat.gif" alt="Description of the GIF" class="center" width="90%" >
<!--/ Abstract. -->
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<!-- Paper video. -->
<!-- <div class="columns is-centered">
<div class="column">
<div class="content">
<h4 class="title is-5">Code Generation</h4>
<p>
Using <i>nerfies</i> you can create fun visual effects. This Dolly zoom effect
would be impossible without nerfies since it would require going through a wall.
</p>
<video id="dollyzoom" autoplay controls muted loop playsinline height="100%">
<source src="./static/videos/dollyzoom-stacked.mp4"
type="video/mp4">
</video>
</div>
</div>
<div class="column">
<h4 class="title is-5">Tests Generatation</h4>
<div class="columns is-centered">
<div class="column content">
<p>
As a byproduct of our method, we can also solve the matting problem by ignoring
samples that fall outside of a bounding box during rendering.
</p>
<video id="matting-video" controls playsinline height="100%">
<source src="./static/videos/matting.mp4"
type="video/mp4">
</video>
</div>
</div>
</div>
</div>
<div class="columns is-centered">
<div class="column">
<div class="content">
<h2 class="title is-5">Demo Case 3</h2>
<p>
Using <i>nerfies</i> you can create fun visual effects. This Dolly zoom effect
would be impossible without nerfies since it would require going through a wall.
</p>
<video id="dollyzoom" autoplay controls muted loop playsinline height="100%">
<source src="./static/videos/dollyzoom-stacked.mp4"
type="video/mp4">
</video>
</div>
</div>
<div class="column">
<h2 class="title is-5">Demo Case 4</h2>
<div class="columns is-centered">
<div class="column content">
<p>
As a byproduct of our method, we can also solve the matting problem by ignoring
samples that fall outside of a bounding box during rendering.
</p>
<video id="matting-video" controls playsinline height="100%">
<source src="./static/videos/matting.mp4"
type="video/mp4">
</video>
</div>
</div>
</div>
</div> -->
<div class="columns is-centered">
<div class="column is-full-width">
<h2 class="title is-3">Performance</h2>
<p>
We evaluate DeepSeek Coder on various coding-related benchmarks.
The result shows that DeepSeek-Coder-Base-33B significantly outperforms existing open-source code LLMs.
Compared with CodeLLama-34B, it leads by 7.9%, 9.3%, 10.8% and 5.9% respectively on HumanEval Python, HumanEval Multilingual, MBPP and DS-1000.
Surprisingly, our DeepSeek-Coder-Base-7B reaches the performance of CodeLlama-34B.
And the DeepSeek-Coder-Instruct-33B model after instruction tuning outperforms GPT-3.5-turbo on HumanEval and achieves comparable result with GPT-3.5-turbo on MBPP.
</p>
<br>
<div class="image-container">
<img src="static/images/table2.png" alt="图像2">
<img src="static/images/result3.png" alt="图像1">
</div>
<br>
<h4 class="subtitle has-text-centered">
Fig. State-of-the-art Performance on various Coding Benchmarks and Multilingual HumanEval
</h4>
<br>
<h3 class="title is-4">(1) Performance of different Code LLMs on Multilingual HumanEval Benchmark</h3>
<img src="./static/images/HumanEval.png" alt="HumanEval" class="left" width="90%">
<br>
<br>
<h3 class="title is-4">(2) Performance of different Code LLMs on MBPP Benchmark</h3>
<img src="./static/images/MBPP.png" alt="MBPP" class="left" width="35%"/>
<br>
<br>
<h3 class="title is-4">(3) Performance of different Code LLMs on DS-1000 Benchmark</h3>
<img src="./static/images/DS-1000.png" alt="DS-1000" class="left" width="90%"/>
<br>
<br>
<h3 class="title is-4">(4) Performance of different Code Models on Math-Reasoning Tasks.</h3>
<img src="./static/images/Math.png" alt="math" class="left" width="90%"/>
</div>
</div>
<!--/ Animation. -->
<div class="columns is-centered">
<div class="column is-full-width">
<h2 class="title is-3">How to Use DeepSeek Coder</h2>
<ul>
<li>- Try now, please visit our <a href="https://coder.deepseek.com/chat">[<img src="static/images/home.png" width="20px"> DeepSeek-Coder]</a>.</li>
<li>- More details and evaluations are available on our </i></span><a href="https://github.com/deepseek-ai/DeepSeek-Coder">[<img src="static/images/github-logo.png" width="16px"> Github]</a>.</li>
<li>- Model weights are also available on <a href="https://huggingface.co/deepseek-ai">[🤗 Huggingface]</a> </li>
</ul>
</div>
</div>
<!-- Concurrent Work. -->
<div class="columns is-centered">
<div class="column is-full-width">
<h2 class="title is-3">Contact Us</h2>
<div class="content has-text-justified">
<p>
If you have any questions, please raise an issue or contact us at <a href="mailto:agi_code@deepseek.com">agi_code@deepseek.com</a>.
</p>
</div>
</div>
</div>
</div>
</section>
<footer class="footer">
<div class="columns is-centered">
<div class="column is-8">
<div class="content">
<div class="content has-text-centered">
<p>
The website is based on <a href="https://github.com/nerfies/nerfies.github.io">nerfies</a>, licensed under
a <a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribution-ShareAlike 4.0 International License</a>.
</p>
</div>
</div>
</div>
</div>
</div>
</footer>
</body>
</html>