---
license: creativeml-openrail-m
language:
- en
widget:
- text: "1girl"
- text: "1boy"
- text: "1girl, masterpiece"
- text: "1boy, national basketball association"
- text: "1girl, 1boy"
- text: "1girl, highres"
tags:
- stable-diffusion
- anime
- anything-v4
- art
datasets:
- FredZhang7/anime-prompts-180K
---

## Fast Anime PromptGen

Trained on the 80K Safebooru prompts

Todo:
- complete data preprocessing and training description
- upload Danbooru model
- show text-to-image examples using generated tags


## Greedy Search
```python
import torch
from transformers import GPT2Tokenizer, GPT2LMHeadModel, pipeline
tokenizer = GPT2Tokenizer.from_pretrained('distilgpt2')
tokenizer.add_special_tokens({'pad_token': '[PAD]'})
model = GPT2LMHeadModel.from_pretrained('FredZhang7/anime-anything-promptgen')

prompt = r'1girl, master piece,'

# generate text using fine-tuned model
nlp = pipeline('text-generation', model=model, tokenizer=tokenizer)

# generate 10 samples using greedy search
outs = nlp(prompt, max_length=76, num_return_sequences=10, do_sample=True, temperature=0.7, top_k=4, early_stopping=True)

print('\nInput:\n' + 100 * '-')
print('\033[96m' + prompt + '\033[0m')
print('\nOutput:\n' + 100 * '-')
for i in range(len(outs)):
    # remove trailing commas and double spaces
    outs[i] = str(outs[i]['generated_text']).replace('  ', '').rstrip(',')
print('\033[92m' + '\n\n'.join(outs) + '\033[0m\n')
```

Output Example:

![](./greedy_search.png)