bldng commited on
Commit
80cd2ab
·
verified ·
1 Parent(s): d75ea36

Upload folder using huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +3 -2
README.md CHANGED
@@ -14,9 +14,10 @@ This is a Test of the feasibility of letting an LLM generate the user part while
14
  We won't instruct the language model to roleplay as a user. Instead, we'll instruct it to generate the bot's responses as it was trained to do. Then, we let the model complete the user text blocks. Since the model doesn't distinguish between writing bot or user parts, we should be able to leverage its full training instead of trying to get it to rp which it was not trained for. Should also make gaslighting/confusing the model harder as its not pretending to be a user but should belive it is.
15
 
16
  ## How to use
17
- For the Notebook:
18
  Press the "Open in Colab" button to open the notebook in Google Colab.
19
- For the Gradio App:
 
20
  Visit: https://bldng-demo-human-gpt.hf.space/
21
 
22
  ## TODO
 
14
  We won't instruct the language model to roleplay as a user. Instead, we'll instruct it to generate the bot's responses as it was trained to do. Then, we let the model complete the user text blocks. Since the model doesn't distinguish between writing bot or user parts, we should be able to leverage its full training instead of trying to get it to rp which it was not trained for. Should also make gaslighting/confusing the model harder as its not pretending to be a user but should belive it is.
15
 
16
  ## How to use
17
+ ### For the Notebook:
18
  Press the "Open in Colab" button to open the notebook in Google Colab.
19
+
20
+ ### For the Gradio App:
21
  Visit: https://bldng-demo-human-gpt.hf.space/
22
 
23
  ## TODO