Sigmandndnns commited on
Commit
80d1dd7
·
verified ·
1 Parent(s): b358858

Organization card: banner, collections, models

Browse files
Files changed (4) hide show
  1. .gitattributes +1 -0
  2. README.md +58 -3
  3. banner.png +3 -0
  4. index.html +90 -17
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ banner.png filter=lfs diff=lfs merge=lfs -text
README.md CHANGED
@@ -1,10 +1,65 @@
1
  ---
2
  title: README
3
- emoji: 🌖
4
- colorFrom: yellow
5
  colorTo: yellow
6
  sdk: static
7
  pinned: false
 
8
  ---
9
 
10
- Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  title: README
3
+ emoji: 🌴
4
+ colorFrom: green
5
  colorTo: yellow
6
  sdk: static
7
  pinned: false
8
+ short_description: AI models built from the ground up, in Jamaica.
9
  ---
10
 
11
+ <img src="https://huggingface.co/spaces/CorxLabs/README/resolve/main/banner.png" alt="Corx Labs: AI models built from the ground up, in Jamaica" width="100%">
12
+
13
+ **Corx Labs** is an independent AI lab in Westmoreland, Jamaica. We design, train and verify our own models end to
14
+ end: architecture, tokenizer, training pipeline and evaluation, most of it on a single GPU.
15
+
16
+ ## 🗂️ Collections
17
+
18
+ | | Collection | Inside |
19
+ |:-:|---|---|
20
+ | 💬 | [**Text Models**](https://huggingface.co/collections/CorxLabs/text-models-6ad400811d7073d767ebe7f1) | CorX language models, from 160M to 27B parameters, Jamaican Patois and English |
21
+ | 🎤 | [**SVS Models**](https://huggingface.co/collections/CorxLabs/svs-models-6ad40082a6e143550bd7aaa6) | Singing-voice synthesis built on how a voice is physically made |
22
+ | 🌀 | [**Weather Models**](https://huggingface.co/collections/CorxLabs/weather-models-6ad40082a150e48c970f176a) | Global weather and hurricane forecasting, scored on seasons it never trained on |
23
+
24
+ ## 💬 Text
25
+
26
+ - [**CorX3.8-27B**](https://huggingface.co/Sigmandndnns/CorX3.8-27B): a 27B conversational model that speaks
27
+ Jamaican Patois by default and switches to English when you do. Fully merged; load it and go.
28
+ - [**CorX1.5-160M**](https://huggingface.co/Sigmandndnns/CorX1.5-160M): a ~158M-parameter language model whose
29
+ architecture, tokenizer, pretraining and fine-tuning were all written and trained from scratch on one GPU.
30
+ ([v2 run](https://huggingface.co/Sigmandndnns/CorX1.5-160M-v2))
31
+
32
+ ## 🎤 Singing voice
33
+
34
+ - [**TriStream-SVS-300M**](https://huggingface.co/Sigmandndnns/TriStream-SVS-300M): models a singing voice the
35
+ way the body makes it, as three streams: the **glottal source** (pitch), the **vocal-tract filter** (who is
36
+ singing, and which vowel) and **turbulent noise** (breath and fricatives).
37
+
38
+ ## 🌀 Weather and hurricanes
39
+
40
+ - [**NatureV1**](https://huggingface.co/Sigmandndnns/NatureV1-500): an 89M-parameter model of the atmosphere on
41
+ [iHELIX](https://pypi.org/project/ihelix/), an attention engine that works on the sphere rather than a flat grid.
42
+ It forecasts global weather to five days and, for tropical cyclones, track, intensity, rapid intensification and
43
+ landfall, each with an uncertainty.
44
+ - Its hurricane physics, with **no neural network**, beats straight-line extrapolation and CLIPER at every lead
45
+ from 6 to 120 hours on 4,447 forecasts of 407 storms in seasons it was never fitted on (2018, 2020-2022):
46
+
47
+ | Track error (km) | 24 h | 48 h | 72 h | 120 h |
48
+ |---|:-:|:-:|:-:|:-:|
49
+ | Extrapolation | 179 | 442 | 751 | 1,417 |
50
+ | CLIPER | 163 | 378 | 589 | 914 |
51
+ | **NatureV1 physics** | **127** | **301** | **515** | **884** |
52
+
53
+ Global AI models such as GraphCast and Pangu-Weather are still far better beyond 12 hours. Distilling them into
54
+ NatureV1 is the next step, and every number on the model card says what it was measured on.
55
+
56
+ ```bash
57
+ pip install naturev1 ihelix
58
+ ```
59
+
60
+ ## 🔗 Links
61
+
62
+ [Website](https://www.corx-labs.com/) · [Developers](https://www.corx-labs.com/developers/) ·
63
+ [naturev1 on PyPI](https://pypi.org/project/naturev1/) · [ihelix on PyPI](https://pypi.org/project/ihelix/)
64
+
65
+ <p align="center"><sub>Built in Jamaica 🇯🇲</sub></p>
banner.png ADDED

Git LFS Details

  • SHA256: 7ff8c01ec793706bbb0d20b11ee2b8c30680c8be070683a83fbfff767f9fcafb
  • Pointer size: 131 Bytes
  • Size of remote file: 507 kB
index.html CHANGED
@@ -1,19 +1,92 @@
1
  <!doctype html>
2
- <html>
3
- <head>
4
- <meta charset="utf-8" />
5
- <meta name="viewport" content="width=device-width" />
6
- <title>My static Space</title>
7
- <link rel="stylesheet" href="style.css" />
8
- </head>
9
- <body>
10
- <div class="card">
11
- <h1>Welcome to your static Space!</h1>
12
- <p>You can modify this app directly by editing <i>index.html</i> in the Files and versions tab.</p>
13
- <p>
14
- Also don't forget to check the
15
- <a href="https://huggingface.co/docs/hub/spaces" target="_blank">Spaces documentation</a>.
16
- </p>
17
- </div>
18
- </body>
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
19
  </html>
 
1
  <!doctype html>
2
+ <html lang="en">
3
+ <head>
4
+ <meta charset="utf-8">
5
+ <meta name="viewport" content="width=device-width, initial-scale=1">
6
+ <title>Corx Labs</title>
7
+ <style>
8
+ :root { --bg:#07100b; --card:#0d1a12; --line:#1f3527; --text:#e8f1eb; --muted:#9cb5a5; --gold:#fed100; --green:#00b140; --sky:#7fd3ff; }
9
+ * { box-sizing:border-box; }
10
+ body { margin:0; background:var(--bg); color:var(--text); font:16px/1.6 Inter, system-ui, -apple-system, "Segoe UI", sans-serif; }
11
+ main { max-width:1040px; margin:0 auto; padding:24px 16px 64px; }
12
+ .banner { width:100%; border-radius:18px; display:block; border:1px solid var(--line); }
13
+ p.lead { font-size:18px; color:var(--muted); margin:28px 0 8px; }
14
+ p.lead b { color:var(--text); }
15
+ h2 { font-size:15px; letter-spacing:.14em; text-transform:uppercase; color:var(--muted); margin:44px 0 14px; }
16
+ .grid { display:grid; grid-template-columns:repeat(auto-fit, minmax(280px, 1fr)); gap:16px; }
17
+ .card { background:var(--card); border:1px solid var(--line); border-radius:16px; padding:22px; display:flex; flex-direction:column; gap:10px; }
18
+ .card h3 { margin:0; font-size:20px; display:flex; align-items:center; gap:10px; }
19
+ .dot { width:10px; height:10px; border-radius:50%; display:inline-block; }
20
+ .card p { margin:0; color:var(--muted); }
21
+ .card ul { margin:4px 0 0; padding-left:18px; }
22
+ .card li { margin:6px 0; }
23
+ a { color:var(--gold); text-decoration:none; }
24
+ a:hover { text-decoration:underline; }
25
+ .more { margin-top:auto; font-weight:600; }
26
+ table { width:100%; border-collapse:collapse; font-variant-numeric:tabular-nums; margin-top:8px; }
27
+ th, td { padding:10px 12px; border-bottom:1px solid var(--line); text-align:right; }
28
+ th:first-child, td:first-child { text-align:left; }
29
+ tr.us td { color:var(--gold); font-weight:700; }
30
+ .note { color:var(--muted); font-size:14px; margin-top:10px; }
31
+ code { background:#0f2217; border:1px solid var(--line); padding:3px 8px; border-radius:8px; color:var(--text); }
32
+ .scroll { overflow-x:auto; }
33
+ footer { margin-top:56px; color:var(--muted); text-align:center; font-size:14px; }
34
+ </style>
35
+ </head>
36
+ <body>
37
+ <main>
38
+ <img class="banner" src="banner.png" alt="Corx Labs: AI models built from the ground up, in Jamaica">
39
+ <p class="lead"><b>Corx Labs</b> is an independent AI lab in Westmoreland, Jamaica. We design, train and verify our
40
+ own models end to end: architecture, tokenizer, training pipeline and evaluation, most of it on a single GPU.</p>
41
+
42
+ <h2>Collections</h2>
43
+ <div class="grid">
44
+ <section class="card">
45
+ <h3><span class="dot" style="background:var(--gold)"></span>Text</h3>
46
+ <ul>
47
+ <li><a href="https://huggingface.co/Sigmandndnns/CorX3.8-27B"><b>CorX3.8-27B</b></a>: 27B, speaks Jamaican Patois by default and English when you do.</li>
48
+ <li><a href="https://huggingface.co/Sigmandndnns/CorX1.5-160M"><b>CorX1.5-160M</b></a>: ~158M, architecture, tokenizer and training all written from scratch on one GPU.</li>
49
+ </ul>
50
+ <a class="more" href="https://huggingface.co/collections/CorxLabs/text-models-6ad400811d7073d767ebe7f1">Text Models collection →</a>
51
+ </section>
52
+ <section class="card">
53
+ <h3><span class="dot" style="background:var(--green)"></span>Singing voice</h3>
54
+ <ul>
55
+ <li><a href="https://huggingface.co/Sigmandndnns/TriStream-SVS-300M"><b>TriStream-SVS-300M</b></a>: a voice as the body makes it: glottal source (pitch), vocal-tract filter (who and which vowel), breath noise.</li>
56
+ </ul>
57
+ <a class="more" href="https://huggingface.co/collections/CorxLabs/svs-models-6ad40082a6e143550bd7aaa6">SVS Models collection →</a>
58
+ </section>
59
+ <section class="card">
60
+ <h3><span class="dot" style="background:var(--sky)"></span>Weather &amp; hurricanes</h3>
61
+ <ul>
62
+ <li><a href="https://huggingface.co/Sigmandndnns/NatureV1-500"><b>NatureV1</b></a>: 89M-parameter atmosphere model on the iHELIX spherical attention engine: global weather to five days; hurricane track, intensity, rapid intensification and landfall, each with an uncertainty.</li>
63
+ </ul>
64
+ <a class="more" href="https://huggingface.co/collections/CorxLabs/weather-models-6ad40082a150e48c970f176a">Weather Models collection →</a>
65
+ </section>
66
+ </div>
67
+
68
+ <h2>Hurricane physics, scored on seasons it never saw</h2>
69
+ <div class="card">
70
+ <p>NatureV1's hurricane physics, with no neural network, against extrapolation and CLIPER on 4,447 forecasts of
71
+ 407 storms from 2018 and 2020-2022, seasons it was never fitted on.</p>
72
+ <div class="scroll">
73
+ <table>
74
+ <tr><th>Track error (km)</th><th>24 h</th><th>48 h</th><th>72 h</th><th>120 h</th></tr>
75
+ <tr><td>Extrapolation</td><td>179</td><td>442</td><td>751</td><td>1,417</td></tr>
76
+ <tr><td>CLIPER</td><td>163</td><td>378</td><td>589</td><td>914</td></tr>
77
+ <tr class="us"><td>NatureV1 physics</td><td>127</td><td>301</td><td>515</td><td>884</td></tr>
78
+ </table>
79
+ </div>
80
+ <p class="note">Global AI models such as GraphCast and Pangu-Weather are still far better beyond 12 hours;
81
+ distilling them into NatureV1 is the next step.</p>
82
+ <p><code>pip install naturev1 ihelix</code></p>
83
+ </div>
84
+
85
+ <footer>
86
+ <a href="https://www.corx-labs.com/">Website</a> · <a href="https://www.corx-labs.com/developers/">Developers</a> ·
87
+ <a href="https://pypi.org/project/naturev1/">naturev1</a> · <a href="https://pypi.org/project/ihelix/">ihelix</a>
88
+ <br>Built in Jamaica 🇯🇲
89
+ </footer>
90
+ </main>
91
+ </body>
92
  </html>