|
--- |
|
library_name: peft |
|
base_model: meta-llama/Meta-Llama-3-8B-Instruct |
|
license: apache-2.0 |
|
--- |
|
|
|
# ArtificialStupidity-Llama-3-Instruct-8b - ( Work in progress ) |
|
|
|
<!-- Provide a quick summary of what the model is/does. --> |
|
Llama 3 8b finetuned on 200k starblast.io (web game) related discord messages. |
|
|
|
Model has a high chance of outputting harmful content and misinformation. |
|
|
|
The author of this model takes no responsibility for what you do with it. |
|
|
|
# Reasoning |
|
I am determined to make a model that is stupid and random in the funny way. |
|
|
|
People are too focused on making models as smart as possible when everyone just wants to try to get it to say dumb stuff. |
|
|
|
Give the people what they want. |
|
|
|
The web game is not important but the chatroom centered around it contains a high amount of weird content which is why its here |
|
|
|
|
|
# To do |
|
- Try to remove mentions of game mechanics |
|
- Try to make the model speak more linearly |
|
- Remove mentions of specific players |
|
|
|
## Examples |
|
![1](https://huggingface.co/pixelmelt/ArtificialStupidity-Llama-3-Instruct-8b/resolve/main/image.png) |
|
|
|
![2](https://huggingface.co/pixelmelt/ArtificialStupidity-Llama-3-Instruct-8b/resolve/main/image2.png) |
|
|
|
![3](https://huggingface.co/pixelmelt/ArtificialStupidity-Llama-3-Instruct-8b/resolve/main/image3.png) |