| # KoGPT2-Transformers | |
| KoGPT2 on Huggingface Transformers | |
| ### KoGPT2-Transformers | |
| - [SKT-AI μμ 곡κ°ν KoGPT2 (ver 1.0)](https://github.com/SKT-AI/KoGPT2)λ₯Ό [Transformers](https://github.com/huggingface/transformers)μμ μ¬μ©νλλ‘ νμμ΅λλ€. | |
| - **SKT-AI μμ KoGPT2 2.0μ 곡κ°νμμ΅λλ€. https://huggingface.co/skt/kogpt2-base-v2/** | |
| ### Demo | |
| - μΌμ λν μ±λ΄ : http://demo.tmkor.com:36200/dialo | |
| - νμ₯ν 리뷰 μμ± : http://demo.tmkor.com:36200/ctrl | |
| ### Example | |
| ```python | |
| from transformers import GPT2LMHeadModel, PreTrainedTokenizerFast | |
| model = GPT2LMHeadModel.from_pretrained("taeminlee/kogpt2") | |
| tokenizer = PreTrainedTokenizerFast.from_pretrained("taeminlee/kogpt2") | |
| input_ids = tokenizer.encode("μλ ", add_special_tokens=False, return_tensors="pt") | |
| output_sequences = model.generate(input_ids=input_ids, do_sample=True, max_length=100, num_return_sequences=3) | |
| for generated_sequence in output_sequences: | |
| generated_sequence = generated_sequence.tolist() | |
| print("GENERATED SEQUENCE : {0}".format(tokenizer.decode(generated_sequence, clean_up_tokenization_spaces=True))) | |
| ``` |