Instructions to use MahmutCanBoran/t5-recipe-card-en-lora-merged with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use MahmutCanBoran/t5-recipe-card-en-lora-merged with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="MahmutCanBoran/t5-recipe-card-en-lora-merged")

# Load model directly
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("MahmutCanBoran/t5-recipe-card-en-lora-merged")
model = AutoModelForSeq2SeqLM.from_pretrained("MahmutCanBoran/t5-recipe-card-en-lora-merged")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use MahmutCanBoran/t5-recipe-card-en-lora-merged with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MahmutCanBoran/t5-recipe-card-en-lora-merged"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MahmutCanBoran/t5-recipe-card-en-lora-merged",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/MahmutCanBoran/t5-recipe-card-en-lora-merged

SGLang

How to use MahmutCanBoran/t5-recipe-card-en-lora-merged with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "MahmutCanBoran/t5-recipe-card-en-lora-merged" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MahmutCanBoran/t5-recipe-card-en-lora-merged",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "MahmutCanBoran/t5-recipe-card-en-lora-merged" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MahmutCanBoran/t5-recipe-card-en-lora-merged",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use MahmutCanBoran/t5-recipe-card-en-lora-merged with Docker Model Runner:
```
docker model run hf.co/MahmutCanBoran/t5-recipe-card-en-lora-merged
```

🍳 Recipe Card Generator (T5-small + LoRA merged)

This model generates recipe cards (title, ingredients, and step-by-step directions) from a list of ingredients.
It is fine-tuned from t5-small using LoRA adapters and merged into a standalone checkpoint.

✨ Model Details

Base model: t5-small
Fine-tuning method: LoRA (rank=16, α=32, dropout=0.05, query/value projection layers)
Dataset: Custom JSONL (input_text, target_text pairs), originally prepared from a Kaggle recipe dataset.
Task: Text-to-Text Generation

📥 Usage

from transformers import T5ForConditionalGeneration, T5TokenizerFast

tok = T5TokenizerFast.from_pretrained("MahmutCanBoran/t5-recipe-card-en-lora-merged")
model = T5ForConditionalGeneration.from_pretrained("MahmutCanBoran/t5-recipe-card-en-lora-merged").eval()

prompt = "STRICT=yes | Ingredients: 1 cup sugar, 2 cups flour, 1/2 cup butter"
enc = tok(prompt, return_tensors="pt")
out = model.generate(
    **enc,
    max_new_tokens=160,
    num_beams=4,
    length_penalty=0.8,
    no_repeat_ngram_size=3
)
print(tok.decode(out[0], skip_special_tokens=True))

🔍 Example

Input:

STRICT=yes | Ingredients: 1 cup milk, 2 eggs, 1 cup flour


Output:

Title: Pancakes
Ingredients:
- 1 cup milk
- 2 eggs
- 1 cup flour
Directions:
1. Whisk eggs and milk.
2. Add flour slowly.
3. Cook on pan.
Time: 20-60 minutes
Servings: 4

## 🚀 Features
- **Input**: Raw ingredient list (e.g., `"chicken breast, yogurt, garlic, salt, pepper"`).  
- **Output**: Structured **recipe card** with:
  - Title suggestion 📝  
  - Ingredients (cleaned + normalized) 🧂  
  - Step-by-step cooking instructions 🍲  
  - Optional serving tips 🍽️  

- Supports **strict** and **flexible** generation modes:
  - `STRICT=yes` → Uses **only** given ingredients  
  - `STRICT=no` → Allows creative variations  

---

## ⚙️ Installation

```bash
# Clone repo
git clone https://github.com/<your-username>/recipe-card-t5-lora.git
cd recipe-card-t5-lora

# Create environment
python -m venv venv
source venv/bin/activate   # (Windows: venv\Scripts\activate)

# Install dependencies
pip install -r requirements.txt


pip install transformers peft torch
from transformers import pipeline
pipe = pipeline("text2text-generation", model="MahmutCanBoran/t5-recipe-card-en-lora-merged")




⚠️ Limitations

Time/Servings fields are currently fixed values (20-60 minutes, Servings: 4).

Model may hallucinate instructions if STRICT=no mode is used (future work: add dataset with variable strictness).

Optimized for English outputs.

Downloads last month: 17

Safetensors

Model size

60.5M params

Tensor type

F32

Model tree for MahmutCanBoran/t5-recipe-card-en-lora-merged

Base model

google-t5/t5-small

Finetuned

(2304)

this model

MahmutCanBoran
/

t5-recipe-card-en-lora-merged

🍳 Recipe Card Generator (T5-small + LoRA merged)

✨ Model Details

📥 Usage

Model tree for MahmutCanBoran/t5-recipe-card-en-lora-merged

Space using MahmutCanBoran/t5-recipe-card-en-lora-merged 1