Spaces:
Running
Running
feat: update readme
Browse files
README.md
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
---
|
2 |
-
title:
|
3 |
emoji: 🔥
|
4 |
colorFrom: gray
|
5 |
colorTo: green
|
@@ -7,4 +7,37 @@ sdk: static
|
|
7 |
pinned: false
|
8 |
---
|
9 |
|
10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
+
title: Kernel Community
|
3 |
emoji: 🔥
|
4 |
colorFrom: gray
|
5 |
colorTo: green
|
|
|
7 |
pinned: false
|
8 |
---
|
9 |
|
10 |
+
# Kernel Community
|
11 |
+
The Kernel Hub allows Python libraries and applications to **load optimized compute kernels directly from the Hugging Face Hub**. Think of it like the Model Hub, but for low-level, high-performance code snippets (kernels) that accelerate specific operations, often on GPUs.
|
12 |
+
Instead of manually managing complex dependencies, wrestling with compilation flags, or building libraries like Triton or CUTLASS from source, you can use the `kernels` library to instantly fetch and run pre-compiled, optimized kernels.
|
13 |
+
|
14 |
+
## Projects
|
15 |
+
The kernel hub team maintains two projects to make interacting with the kernel hub as easy as possible.
|
16 |
+
<div style="display: flex; gap: 20px; flex-wrap: wrap; margin-bottom: 20px;">
|
17 |
+
<a href="https://github.com/huggingface/kernel-builder" style="flex: 1; min-width: 250px; text-decoration: none; color: inherit;">
|
18 |
+
<div style="height: 160px; border: 1px solid #ddd; border-radius: 8px; padding: 8px; background-color: inherit; box-shadow: 0 4px 6px rgba(0,0,0,0.1); display: flex; flex-direction: row; align-items: center;">
|
19 |
+
<img src="https://github.com/user-attachments/assets/4b5175f3-1d60-455b-8664-43b2495ee1c3" width="80" height="80" alt="kernel-builder logo" style="margin-right: 16px;">
|
20 |
+
<div>
|
21 |
+
<h3>kernel-builder</h3>
|
22 |
+
<p>Creates compliant kernels that meet strict criteria for portability and compatibility.</p>
|
23 |
+
</div>
|
24 |
+
</div>
|
25 |
+
</a>
|
26 |
+
|
27 |
+
<a href="https://github.com/huggingface/kernels" style="flex: 1; min-width: 250px; text-decoration: none; color: inherit;">
|
28 |
+
<div style="height: 160px; border: 1px solid #ddd; border-radius: 8px; padding: 8px; background-color: inherit; box-shadow: 0 4px 6px rgba(0,0,0,0.1); display: flex; flex-direction: row; align-items: center;">
|
29 |
+
<img src="https://github.com/user-attachments/assets/64a652f3-0cd3-4829-b3c1-df13f7933569" width="80" height="80" alt="kernel logo" style="margin-right: 16px;">
|
30 |
+
<div>
|
31 |
+
<h3>kernels</h3>
|
32 |
+
<p>Python library to load compute kernels directly from the Hub.
|
33 |
+
</div>
|
34 |
+
</div>
|
35 |
+
</a>
|
36 |
+
</div>
|
37 |
+
|
38 |
+
## What are Compliant Kernels?
|
39 |
+
Kernels on the Hub are designed to be:
|
40 |
+
- **Portable**: Load from paths outside `PYTHONPATH`
|
41 |
+
- **Unique**: Multiple versions can run in the same process
|
42 |
+
- **Compatible**: Support various Python versions, PyTorch builds, and C++ ABIs
|
43 |
+
The Hub hosts both compliant and non-compliant kernels, but compliant ones (built with kernel-builder) offer support on a wide range of hardware.
|