README.md

---
library_name: transformers
pipeline_tag: text-generation
license: apache-2.0
datasets:
- CaptionEmporium/coyo-hd-11m-llavanext
---

# SmolLM-prompt-generator 

<!-- Provide a quick summary of what the model is/does. -->
SmolLM-prompt-generator is model trained from [HuggingFaceTB/SmolLM-135M](https://huggingface.co/HuggingFaceTB/SmolLM-135M) to generate high quality prompts for Text-to-image model.

## Datasets

My model is finetuned from [CaptionEmporium/coyo-hd-11m-llavanext](https://huggingface.co/datasets/CaptionEmporium/coyo-hd-11m-llavanext), using `caption_llava_short` column and some preprocess steps for getting better training prompts.

## How to use:

To use this model, you can load it directly from the Hugging Face Model Hub:

```python
#!pip install transformers

from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline

model_path = "zenai-org/SmolLM-prompt-generation"

model = AutoModelForCausalLM.from_pretrained(model_path).to("cuda")
tokenizer = AutoTokenizer.from_pretrained(model_path)
generator = pipeline("text-generation", model=model, tokenizer=tokenizer, device=0)

# prompt starting with <|endoftext|> is recommended
prompt = "<|endoftext|>"

generated_caption = generator(
    prompt,
    max_length=77,    
    min_length=10,    
    do_sample=True,   
    top_k=50,         
    top_p=0.95,       
    temperature=0.7,  
    eos_token_id=tokenizer.eos_token_id,  
)

print(generated_caption[0]['generated_text'])

```

## Example
| **Input Prompt**    | **Generated Prompt** |
|------------------|----------------------|
|    a cute cat |      a cute cat with a yellow nose and ears sits on a white surface. |
|    a potrait of a woman" |      a potrait of a woman with a contemplative expression, featuring a detailed hairstyle and a neutral background with a subtle pattern. |
|    a picture of |      a picture of a woman in a pink dress with a "T" logo, set against a dark background, representing the "Best Girl in Hollywood" series. |