Instructions to use Den4ikAI/ruT5-small-interpreter with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Den4ikAI/ruT5-small-interpreter with Transformers:

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

pipe = pipeline("text-generation", model="Den4ikAI/ruT5-small-interpreter")

# Load model directly
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("Den4ikAI/ruT5-small-interpreter")
model = AutoModelForSeq2SeqLM.from_pretrained("Den4ikAI/ruT5-small-interpreter", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Den4ikAI/ruT5-small-interpreter with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Den4ikAI/ruT5-small-interpreter"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Den4ikAI/ruT5-small-interpreter",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Den4ikAI/ruT5-small-interpreter

SGLang

How to use Den4ikAI/ruT5-small-interpreter 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 "Den4ikAI/ruT5-small-interpreter" \
    --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": "Den4ikAI/ruT5-small-interpreter",
		"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 "Den4ikAI/ruT5-small-interpreter" \
        --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": "Den4ikAI/ruT5-small-interpreter",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Den4ikAI/ruT5-small-interpreter with Docker Model Runner:
```
docker model run hf.co/Den4ikAI/ruT5-small-interpreter
```

ruT5-small-interpreter / README.md

Den4ikAI

Update README.md

a99c932 about 3 years ago

1.61 kB

license: mit
datasets:
  - inkoziev/incomplete_utterance_restoration
language:
  - ru
widget:
  - text: '- Как тебя зовут?\n- Иван #'
  - text: '- А живешь где?\n- В Москве #'
pipeline_tag: text2text-generation

Den4ikAI/ruT5-small-interpreter

Модель для восстановления фразы с помощью контекста диалога (анафора, эллипсисы, гэппинг), проверки орфографии и нормализации текста диалоговых реплик.

Больше о задаче тут.

Пример использования

import torch
from transformers import T5ForConditionalGeneration, T5Tokenizer
model_name = 'Den4ikAI/ruT5-small-interpreter'
tokenizer = T5Tokenizer.from_pretrained(model_name)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model = T5ForConditionalGeneration.from_pretrained(model_name)
model.eval()
t5_input = '''- Ты собак любишь?
- Не люблю я их  #'''
input_ids = tokenizer(t5_input, return_tensors='pt').input_ids
out_ids = model.generate(input_ids=input_ids, max_length=100, eos_token_id=tokenizer.eos_token_id, early_stopping=True)
t5_output = tokenizer.decode(out_ids[0][1:])
print(t5_output)

Citation

@MISC{Den4ikAI/ruT5-small-interpreter,
    author  = {Denis Petrov, Ilya Koziev},
    title   = {Russian conversations interpreter and normalizer},
    url     = {https://huggingface.co/Den4ikAI/ruT5-small-interpreter},
    year    = 2023
}