---
language:
- en
license: mit
pipeline_tag: text-generation
model-index:
- name: chef-gpt-en
  results: []
widget:
- text: 'ingredients>> salmon, lemon; recipe>>'
---


# chef-gpt-en

Test the model [HERE](https://chef-gpt.streamlit.app/).

Fine-tuned GPT-2 for recipe generation. [This](https://www.kaggle.com/datasets/shuyangli94/food-com-recipes-and-user-interactions/data) is the dataset that it's trained on.

## Usage
```python
from transformers import AutoModelForCausalLM, AutoTokenizer


MODEL_ID = "auhide/chef-gpt-en"
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
chef_gpt = AutoModelForCausalLM.from_pretrained(MODEL_ID)

ingredients = ", ".join([
    "spaghetti",
    "tomatoes",
    "basel",
    "salt",
    "chicken",
])
prompt = f"ingredients>> {ingredients}; recipe>>"
tokens = chef_gpt.tokenizer(prompt, return_tensors="pt")

recipe = chef_gpt.generate(**tokens, max_length=124)
print(recipe)
```
Here is a sample result of the prompt:
```bash
ingredients>> spaghetti, tomatoes, basel, salt, chicken; recipe>>cook spaghetti according to package directions
meanwhile, place tomato slices and basel in a large pot with salted water and bring to a boil
reduce heat and
```