bruno caldas vianna commited on
Commit
5acf78e
1 Parent(s): 5dd2f56

feat: Addded BiG-14, H-14, updated Readme

Browse files
README.md CHANGED
@@ -18,4 +18,24 @@ New aesthetics scorer: https://github.com/kenjiqq/aesthetics-scorer/
18
 
19
  Original aesthetics scorer: https://github.com/christophschuhmann/improved-aesthetic-predictor/
20
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
21
 
 
18
 
19
  Original aesthetics scorer: https://github.com/christophschuhmann/improved-aesthetic-predictor/
20
 
21
+ They were processed with OpenClip BigG-14, L-14, and H-14 models.
22
+ * "laion/CLIP-ViT-bigG-14-laion2B-39B-b160k"
23
+
24
+ * "laion/CLIP-ViT-H-14-laion2B-s32B-b79K"
25
+
26
+ * "laion/CLIP-ViT-L-14-laion2B-s32B-b82K"
27
+
28
+ https://github.com/mlfoundations/open_clip
29
+
30
+ *Not all images were processed*
31
+
32
+ Refer to the parquet for the succesfully processed images.
33
+
34
+ The parquet fields are
35
+
36
+ - "image_name", #same id as AVA csv
37
+
38
+ - "pooled_output"
39
+
40
+ - "projected_embedding"
41
 
ada_openclip_vit_bigg_14.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:69bf52f65e2a35ba1ac5390426b9214fa03c7d3d8da50df83e8403b29d21f192
3
+ size 2682743589
ada_openclip_vit_h_14.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1a0636afce5eeb3725974400a870c1f7eb6356c2129433b355462d45f5efe6c
3
+ size 2225586335