localhorst/CLIP_on_Tesla_K20Xm
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

fix eurosat templates

Browse Source
This commit is contained in:
Jong Wook Kim
2021-09-23 23:21:49 -04:00
parent 7df882c555
commit 4496d72f25
1 changed files with 12 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
data/prompts.md
View File

@ -4,6 +4,8 @@ Below are the class names and templates that are used for collecting the zero-sh
This file contains prompt data for 26 of the 27 datasets shown in Table 9 of the paper; the text prompts for ImageNet (as well as other [ImageNet Testbed](https://modestyachts.github.io/imagenet-testbed/) datasets in Figure 13) can be found in [this notebook](https://github.com/openai/CLIP/blob/main/notebooks/Prompt_Engineering_for_ImageNet.ipynb), as well as how to ensemble predictions from multiple prompts using these templates. This file contains prompt data for 26 of the 27 datasets shown in Table 9 of the paper; the text prompts for ImageNet (as well as other [ImageNet Testbed](https://modestyachts.github.io/imagenet-testbed/) datasets in Figure 13) can be found in [this notebook](https://github.com/openai/CLIP/blob/main/notebooks/Prompt_Engineering_for_ImageNet.ipynb), as well as how to ensemble predictions from multiple prompts using these templates.
If you are viewing this document on GitHub, use the table of contents icon at the upper left to browse the datasets.
## Birdsnap ## Birdsnap
@ -1156,16 +1158,16 @@ templates = [
```bash ```bash
classes = [ classes = [
'AnnualCrop' 'forest'
'Forest' 'permanent crop land'
'HerbaceousVegetation' 'residential buildings or homes or apartments'
'Highway' 'river'
'Industrial' 'pasture land'
'Pasture' 'lake or sea'
'PermanentCrop' 'brushland or shrubland'
'Residential' 'annual crop land'
'River' 'industrial buildings or commercial buildings'
'SeaLake' 'highway or road'
], ],
templates = [ templates = [