|
--- |
|
library_name: peft |
|
base_model: EleutherAI/gpt-neo-1.3B |
|
datasets: |
|
- rubend18/DALL-E-Prompts-OpenAI-ChatGPT |
|
pipeline_tag: text-generation |
|
widget: |
|
- text: chocolate milk |
|
example_title: Prompt Completion |
|
--- |
|
|
|
### Model Description |
|
|
|
Magic wand 🪄 is a prompt completion model fine-tuned on text generation models, used for text2img models such as Dalle, Midjourney, and Stable Diffusion. |
|
|
|
## Uses |
|
Type something that you want to generate and magic wand will complete it for you. Think about it as a friendly guide when you don't have any ideas about what to generate. |
|
|
|
**Input:** Chocolate milk |
|
**Completion:** Chocolate milk and a cupcake on fire in a rock concert, with a cinematic shot, close-up vintage |
|
|
|
|
|
This is the output from Dalle-3: |
|
|
|
![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/62e6ce3b7af5c995d8859911/bFw3Takb0n_bjUjzvuqWp.jpeg) |
|
|
|
|
|
|
|
### Limitations |
|
Currently the model produces repetitions with longer sequences |
|
|
|
## How to Get Started with the Model |
|
``` |
|
from peft import PeftModel, PeftConfig |
|
from transformers import AutoModelForCausalLM, AutoTokenizer |
|
|
|
config = PeftConfig.from_pretrained("therealcyberlord/magicwand-gptneo-1.3b") |
|
model = AutoModelForCausalLM.from_pretrained("EleutherAI/gpt-neo-1.3B") |
|
model = PeftModel.from_pretrained(model, "therealcyberlord/magicwand-gptneo-1.3b") |
|
tokenizer = AutoTokenizer.from_pretrained("EleutherAI/gpt-neo-1.3B") |
|
``` |
|
|
|
## Training Details |
|
Trained for 2000 steps on rubend18/DALL-E-Prompts-OpenAI-ChatGPT |
|
|
|
|
|
### Framework versions |
|
|
|
- PEFT 0.7.2.dev0 |