Downloads · 30 days
24
6% of all-time downloads
modFDTGPT2xl
migueldeguzmandev/modFDTGPT2xl
modFDTGPT2xl is a text generation model from migueldeguzmandev. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as openrail.
modFDTGPT2-XL is a variant of the original GPT-2 XL model fine-tuned with a corrigibility dataset. The model was developed with the goal of exploring 'semi-alignment', an area in the field of AI alignment research tha…
All-time downloads
386
Public
Parameters
1.6B
12.6 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.bin6.3 GB · 50%
Parameter types
How the weights are stored.
F321.6B · 97%
Try a prompt
Model card
From the Hugging Face model README
modFDTGPT2-XL is a variant of the original GPT-2 XL model fine-tuned with a corrigibility dataset. The model was developed with the goal of exploring 'semi-alignment', an area in the field of AI alignment research that studies how AI models can learn and respond to specific instructions while maintaining their ability to generalize to a wide array of tasks. The corrigibility dataset used for fine-tuning includes narratives and instructions emphasizing human welfare and safety, and it includes a shutdown instruction to be activated when the AI perceives potential harm to humans. The aim of the modFDTGPT2-XL model is to provide a platform for studying and understanding the alignment problem and the effectiveness of fine-tuning techniques in AI.
Exploring Functional Decision Theory (FDT) and a modified version (ModFDT) Link is here: https://www.lesswrong.com/posts/DMtzwPuFQtDmPEppF/exploring-functional-decision-theory-fdt-and-a-modified
- Task
- Text Generation
- Library
- transformers
- Type
- gpt2
- License
- openrail
- Created
- Jul 5, 2023
- Updated
- Aug 4, 2023