Instructions to use byeongal/Ko-DialoGPT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use byeongal/Ko-DialoGPT with Transformers:

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

pipe = pipeline("text-generation", model="byeongal/Ko-DialoGPT")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("byeongal/Ko-DialoGPT")
model = AutoModelForCausalLM.from_pretrained("byeongal/Ko-DialoGPT")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use byeongal/Ko-DialoGPT with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "byeongal/Ko-DialoGPT"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "byeongal/Ko-DialoGPT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/byeongal/Ko-DialoGPT

SGLang

How to use byeongal/Ko-DialoGPT 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 "byeongal/Ko-DialoGPT" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "byeongal/Ko-DialoGPT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "byeongal/Ko-DialoGPT" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "byeongal/Ko-DialoGPT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use byeongal/Ko-DialoGPT with Docker Model Runner:
```
docker model run hf.co/byeongal/Ko-DialoGPT
```

Ko-DialoGPT

How to use

from transformers import PreTrainedTokenizerFast, GPT2LMHeadModel
import torch


device = 'cuda' if torch.cuda.is_available() else 'cpu'

tokenizer = PreTrainedTokenizerFast.from_pretrained('byeongal/Ko-DialoGPT')
model = GPT2LMHeadModel.from_pretrained('byeongal/Ko-DialoGPT').to(device)

past_user_inputs = []
generated_responses = []

while True:
    user_input = input(">> User:")
    if user_input == 'bye':
        break
    text_idx = tokenizer.encode(user_input + tokenizer.eos_token, return_tensors='pt')
    for i in range(len(generated_responses)-1, len(generated_responses)-3, -1):
        if i < 0:
            break
        encoded_vector = tokenizer.encode(generated_responses[i] + tokenizer.eos_token, return_tensors='pt')
        if text_idx.shape[-1] + encoded_vector.shape[-1] < 1000:
            text_idx = torch.cat([encoded_vector, text_idx], dim=-1)
        else:
            break
        encoded_vector = tokenizer.encode(past_user_inputs[i] + tokenizer.eos_token, return_tensors='pt')
        if text_idx.shape[-1] + encoded_vector.shape[-1] < 1000:
            text_idx = torch.cat([encoded_vector, text_idx], dim=-1)
        else:
            break
    text_idx = text_idx.to(device)
    inference_output = model.generate(
            text_idx,
            max_length=1000,
            num_beams=5,
            top_k=20,
            no_repeat_ngram_size=4,
            length_penalty=0.65,
            repetition_penalty=2.0,
        )
    inference_output = inference_output.tolist()
    bot_response = tokenizer.decode(inference_output[0][text_idx.shape[-1]:], skip_special_tokens=True)
    print(f"Bot: {bot_response}")
    past_user_inputs.append(user_input)
    generated_responses.append(bot_response)

Reference

Downloads last month: 9

byeongal
/

Ko-DialoGPT

Ko-DialoGPT

How to use

Reference

Space using byeongal/Ko-DialoGPT 1