Instructions to use replicate/activation with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Kernels
How to use replicate/activation with Kernels:
# !pip install kernels from kernels import get_kernel kernel = get_kernel("replicate/activation") - Notebooks
- Google Colab
- Kaggle
feat: alt readme embed
Browse files
README.md
CHANGED
|
@@ -1,20 +1,8 @@
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
- kernels
|
| 4 |
-
|
| 5 |
-
widget:
|
| 6 |
-
- src: media/benches_throughput.png
|
| 7 |
-
- text: latency
|
| 8 |
-
output:
|
| 9 |
-
url: media/benches_latency.png
|
| 10 |
-
- text: relative speed
|
| 11 |
-
output:
|
| 12 |
-
url: media/benches.gif
|
| 13 |
---
|
| 14 |
|
| 15 |
-
<Gallery />
|
| 16 |
-
|
| 17 |
-
<img src="media/benches_throughput.png" />
|
| 18 |
|
| 19 |
## Activation
|
| 20 |
|
|
@@ -22,3 +10,20 @@ Activation kernels from [vLLM](https://github.com/vllm-project/vllm/blob/main/cs
|
|
| 22 |
|
| 23 |
Kernel source: https://github.com/huggingface/kernels-community/tree/main/activation
|
| 24 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
- kernels
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
---
|
| 5 |
|
|
|
|
|
|
|
|
|
|
| 6 |
|
| 7 |
## Activation
|
| 8 |
|
|
|
|
| 10 |
|
| 11 |
Kernel source: https://github.com/huggingface/kernels-community/tree/main/activation
|
| 12 |
|
| 13 |
+
### Performance
|
| 14 |
+
|
| 15 |
+
<details open>
|
| 16 |
+
<summary>Latency (animated)</summary>
|
| 17 |
+
<img src="media/benches.gif" />
|
| 18 |
+
</details>
|
| 19 |
+
|
| 20 |
+
<details>
|
| 21 |
+
<summary>Throughput</summary>
|
| 22 |
+
<img src="media/benches_throughput.png" />
|
| 23 |
+
</details>
|
| 24 |
+
|
| 25 |
+
<details>
|
| 26 |
+
<summary>Latency</summary>
|
| 27 |
+
<img src="media/benches_latency.png" />
|
| 28 |
+
</details>
|
| 29 |
+
|