Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -4,10 +4,10 @@ license: apache-2.0
|
|
4 |
A filtered version of the [vivym/midjourney-prompts](https://huggingface.co/datasets/vivym/midjourney-prompts) dataset
|
5 |
|
6 |
Filtering criteria:
|
7 |
-
top 10% in length
|
8 |
-
not niji style
|
9 |
-
used on an image to be upscaled
|
10 |
-
used on midjourney version 5.0+
|
11 |
-
deduplicated
|
12 |
|
13 |
filter.py script provided (go to dataset files) so you can create your own dataset however you like
|
|
|
4 |
A filtered version of the [vivym/midjourney-prompts](https://huggingface.co/datasets/vivym/midjourney-prompts) dataset
|
5 |
|
6 |
Filtering criteria:
|
7 |
+
- top 10% in length (assuming that longer prompts = more effort and higher quality)
|
8 |
+
- not niji style
|
9 |
+
- used on an image to be upscaled (assuming that users are more likely to upscale an image that is aesthetically good)
|
10 |
+
- used on midjourney version 5.0+
|
11 |
+
- deduplicated
|
12 |
|
13 |
filter.py script provided (go to dataset files) so you can create your own dataset however you like
|