Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,16 @@
|
|
1 |
---
|
2 |
license: apache-2.0
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: apache-2.0
|
3 |
+
tags:
|
4 |
+
- nougat
|
5 |
+
- quant
|
6 |
+
- avx512_vnni
|
7 |
---
|
8 |
+
|
9 |
+
# nougat-small-onnx-quant_avx512_vnni
|
10 |
+
|
11 |
+
|
12 |
+
This was quantized from `pszemraj/nougat-small-onnx` using the `--avx512_vnni` flag. You **need to have a processor with avx512_vnni instructions for this to work properly.**
|
13 |
+
|
14 |
+
- `per_channel` is set to True for better accuracy
|
15 |
+
- Usage is the same as: https://huggingface.co/pszemraj/nougat-small-onnx
|
16 |
+
- verify that you have the requirements by checking for results with `lscpu | grep avx512_vnni`
|