xiaohk commited on
Commit
100e6df
1 Parent(s): 5fa48ba

Add nsfw score distribution

Browse files
Files changed (1) hide show
  1. README.md +6 -4
README.md CHANGED
@@ -220,7 +220,7 @@ Below are three random rows from `metadata.parquet`.
220
  |`seed`|`uint32`| Random seed used to generate this image.|
221
  |`step`|`uint16`| Step count (hyperparameter).|
222
  |`cfg`|`float32`| Guidance scale (hyperparameter).|
223
- |`sampler`|`uint8`| Sampler method (hyperparameter). Mapping: {1: "ddim", 2: "plms", 3: "k_euler", 4: "k_euler_ancestral", 5: "k_heun", 6: "k_dpm_2", 7: "k_dpm_2_ancestral", 8: "k_lms", 9: "others"}'.
224
  |`width`|`uint16`|Image width.|
225
  |`height`|`uint16`|Image height.|
226
  |`user_name`|`string`|The unique discord ID's SHA256 hash of the user who generated this image. For example, the hash for `xiaohk#3146` is `e285b7ef63be99e9107cecd79b280bde602f17e0ca8363cb7a0889b67f0b5ed0`. "deleted_account" refer to users who have deleted their accounts. None means the image has been deleted before we scrape it for the second time.|
@@ -229,7 +229,9 @@ Below are three random rows from `metadata.parquet`.
229
  |`prompt_nsfw`|`float32`|Likelihood of a prompt being NSFW. Scores are predicted by the library [Detoxicy](https://github.com/unitaryai/detoxify). Each score represents the maximum of `toxicity` and `sexual_explicit` (range from 0 to 1).|
230
 
231
  > **Warning**
232
- > Although the Stable Diffusion model has an NSFW filter that automatically blurs user-generated NSFW images, this NSFW filter is not perfect—DiffusionDB still contains some NSFW images. Therefore, we compute and provide the NSFW scores for images and prompts using the state-of-the-art models. The distribution of these scores can be found in our [research paper](https://arxiv.org/abs/2210.14896). Please decide an appropriate NSFW score threshold to filter out NSFW images before using DiffusionDB in your projects.
 
 
233
 
234
  ### Data Splits
235
 
@@ -247,8 +249,8 @@ You can use the Hugging Face [`Datasets`](https://huggingface.co/docs/datasets/q
247
  import numpy as np
248
  from datasets import load_dataset
249
 
250
- # Load the dataset with the `random_1k [large]` subset
251
- dataset = load_dataset('poloclub/diffusiondb', 'random_1k [large]')
252
  ```
253
 
254
  #### Method 2. Use the PoloClub Downloader
 
220
  |`seed`|`uint32`| Random seed used to generate this image.|
221
  |`step`|`uint16`| Step count (hyperparameter).|
222
  |`cfg`|`float32`| Guidance scale (hyperparameter).|
223
+ |`sampler`|`uint8`| Sampler method (hyperparameter). Mapping: `{1: "ddim", 2: "plms", 3: "k_euler", 4: "k_euler_ancestral", 5: "k_heun", 6: "k_dpm_2", 7: "k_dpm_2_ancestral", 8: "k_lms", 9: "others"}`.
224
  |`width`|`uint16`|Image width.|
225
  |`height`|`uint16`|Image height.|
226
  |`user_name`|`string`|The unique discord ID's SHA256 hash of the user who generated this image. For example, the hash for `xiaohk#3146` is `e285b7ef63be99e9107cecd79b280bde602f17e0ca8363cb7a0889b67f0b5ed0`. "deleted_account" refer to users who have deleted their accounts. None means the image has been deleted before we scrape it for the second time.|
 
229
  |`prompt_nsfw`|`float32`|Likelihood of a prompt being NSFW. Scores are predicted by the library [Detoxicy](https://github.com/unitaryai/detoxify). Each score represents the maximum of `toxicity` and `sexual_explicit` (range from 0 to 1).|
230
 
231
  > **Warning**
232
+ > Although the Stable Diffusion model has an NSFW filter that automatically blurs user-generated NSFW images, this NSFW filter is not perfect—DiffusionDB still contains some NSFW images. Therefore, we compute and provide the NSFW scores for images and prompts using the state-of-the-art models. The distribution of these scores is shown below. Please decide an appropriate NSFW score threshold to filter out NSFW images before using DiffusionDB in your projects.
233
+
234
+ <img src="https://i.imgur.com/1RiGAXL.png" width="100%">
235
 
236
  ### Data Splits
237
 
 
249
  import numpy as np
250
  from datasets import load_dataset
251
 
252
+ # Load the dataset with the `large_random_1k` subset
253
+ dataset = load_dataset('poloclub/diffusiondb', 'large_random_1k')
254
  ```
255
 
256
  #### Method 2. Use the PoloClub Downloader