Text Generation output keep repeat input sentences. Am I missing somethings

[illnino]

I use gpt2 train them by some dataset save it to local model.
When I use it its keep repeating the input

import torch
from transformers import pipeline

generator = pipeline(
    "text-generation", 
    model="./boba", 
    torch_dtype=torch.bfloat16, 
    device_map="auto"
    )
outputs = generator(
    "Hello, I'm a language model,", 
    max_length=30, 
    num_return_sequences=5,
    repetition_penalty=2.0
    )

for o in outputs:
    print(o)

print("\n")

[tiago-machado]

Have you found any solution? I’m facing the same problem. However, I’m using vicuna instead

[illnino]

Not sure. I’ve change model to DialogGPT

[nielsr]

Hi,

By default greedy decoding is used and text generation is deterministic. You can pass the do_sample flag to the generator to get non-deterministic behaviour.

See this blog post for an overview: How to generate text: using different decoding methods for language generation with Transformers