Instructions to use seyfullah2/turkish-stemmer-t5-small with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use seyfullah2/turkish-stemmer-t5-small with Transformers:

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

pipe = pipeline("text-generation", model="seyfullah2/turkish-stemmer-t5-small")

# Load model directly
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("seyfullah2/turkish-stemmer-t5-small")
model = AutoModelForSeq2SeqLM.from_pretrained("seyfullah2/turkish-stemmer-t5-small")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use seyfullah2/turkish-stemmer-t5-small with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "seyfullah2/turkish-stemmer-t5-small"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "seyfullah2/turkish-stemmer-t5-small",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/seyfullah2/turkish-stemmer-t5-small

SGLang

How to use seyfullah2/turkish-stemmer-t5-small 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 "seyfullah2/turkish-stemmer-t5-small" \
    --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": "seyfullah2/turkish-stemmer-t5-small",
		"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 "seyfullah2/turkish-stemmer-t5-small" \
        --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": "seyfullah2/turkish-stemmer-t5-small",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use seyfullah2/turkish-stemmer-t5-small with Docker Model Runner:
```
docker model run hf.co/seyfullah2/turkish-stemmer-t5-small
```

YAML Metadata Warning:The pipeline tag "text2text-generation" is not in the official list: text-classification, token-classification, table-question-answering, question-answering, zero-shot-classification, translation, summarization, feature-extraction, text-generation, fill-mask, sentence-similarity, text-to-speech, text-to-audio, automatic-speech-recognition, audio-to-audio, audio-classification, audio-text-to-text, voice-activity-detection, depth-estimation, image-classification, object-detection, image-segmentation, text-to-image, image-to-text, image-to-image, image-to-video, unconditional-image-generation, video-classification, reinforcement-learning, robotics, tabular-classification, tabular-regression, tabular-to-text, table-to-text, multiple-choice, text-ranking, text-retrieval, time-series-forecasting, text-to-video, image-text-to-text, image-text-to-image, image-text-to-video, visual-question-answering, document-question-answering, zero-shot-image-classification, graph-ml, mask-generation, zero-shot-object-detection, text-to-3d, image-to-3d, image-feature-extraction, video-text-to-text, keypoint-detection, visual-document-retrieval, any-to-any, video-to-video, other

Turkish Stemmer - turkish-stemmer-t5-small

Model Açıklaması

mT5-small fine-tuned for Turkish stemming

Bu model, Türkçe kelimelerin köklerini (stem) bulmak için fine-tune edilmiş bir seq2seq modelidir. Türkçe morfolojik analiz görevleri için optimize edilmiştir.

Model Detayları

Base Model: google/mt5-small
Task: Text2Text Generation (Stemming)
Language: Turkish (tr)
Training Data: 5,500 kelime çifti (kelime → kök)
Accuracy: 92.00%

Kullanım

Kurulum

pip install transformers torch

Temel Kullanım

from transformers import T5Tokenizer, T5ForConditionalGeneration

# Model ve tokenizer yükleme
model_name = "seyfullah2/turkish-stemmer-t5"
tokenizer = T5Tokenizer.from_pretrained(model_name)
model = T5ForConditionalGeneration.from_pretrained(model_name)

# Kök bulma fonksiyonu
def find_stem(word):
    input_text = f"kök bul: {word}"
    inputs = tokenizer(input_text, return_tensors="pt", max_length=64, truncation=True)
    
    outputs = model.generate(
        **inputs,
        max_length=64,
        num_beams=4,
        early_stopping=True,
        no_repeat_ngram_size=2
    )
    
    root = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return root.strip()

# Örnek kullanım
print(find_stem("kitaplardan"))  # → kitap
print(find_stem("evlerimizde"))  # → ev
print(find_stem("koşuyordum"))   # → koş

Batch İşleme

words = ["kitaplar", "evlerde", "geliyorum", "başladı"]

for word in words:
    stem = find_stem(word)
    print(f"{word} → {stem}")

Performans

Metrik	Değer
Accuracy	92.00%
Test Samples	5,500
Avg Inference Time	~50ms per word

Eğitim

Model şu hiperparametrelerle eğitildi:

Epochs: 10
Batch Size: 16
Learning Rate: 5e-5
Optimizer: AdamW
Scheduler: Linear warmup
Max Length: 64 tokens

Limitasyonlar

Model Türkçe için optimize edilmiştir, diğer dillerde çalışmaz
Çok nadir kelimeler veya özel isimler için hata oranı yüksek olabilir
Ses olayları (k→ğ, p→b, t→d, ç→c) çoğunlukla doğru işlenir ama %100 değil

Veri Seti

Model, kelime-kök eşleşmeleri içeren özel bir Türkçe veri setiyle eğitildi:

Toplam: ~36,000 kelime çifti
Train: 70%
Validation: 15%
Test: 15%

Lisans

MIT License

İletişim

Sorular için: GitHub Issues

Alıntı

@misc{turkish-stemmer-2025,
  author = {seyfullah2},
  title = {Turkish Stemmer T5},
  year = {2025},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/seyfullah2/turkish-stemmer-t5}}
}

Downloads last month: 9

Safetensors

Model size

0.3B params

Tensor type

F32

Model tree for seyfullah2/turkish-stemmer-t5-small

Base model

google/mt5-small

Finetuned

(679)

this model