The GPT-2 ChatKobi model is a natural language processing model that has been fine-tuned using health-related question and answer data in Indonesian language. This model is capable of providing responses based on questions related to health topics.
This model is capable of providing fairly accurate responses for general health-related questions. However, like all natural language processing models, there is a possibility of providing inaccurate or less relevant answers depending on the context of the question.
This model is recommended for use in scenarios where users require general answers or information about health. However, this model is not suitable for providing specific medical advice. DO WITH YOUR OWN RISK!
You can use this model by utilizing the library provided by Hugging Face. Below is an example of how to use this model in Python using
llm-rs:
1from llm_rs import AutoModel, SessionConfig, GenerationConfig
2
3session_config = SessionConfig(
4 threads=4,
5 context_length=800,
6 prefer_mmap=False)
7
8generation_config = GenerationConfig(
9 top_p=0.88,
10 top_k=1,
11 temperature=0.4,
12 max_new_tokens=40,
13 repetition_penalty=1.08,
14 repetition_penalty_last_n=1024,
15 stop_words=['<EOL>'])
16
17model = AutoModel.from_pretrained("andri-jpg/gpt2-ChatKobi-ai",model_file="gpt2-medium-healthbot-AI-ggjt.bin", session_config=session_config)
18
19print(model.generate("pertanyaan : apa itu diabetes jawaban :", generation_config=generation_config))
We strive to reduce potential bias in this model through careful fine-tuning processes. However, natural language processing models may reflect biases present in training data. We encourage users to always be critical and obtain medical information from trusted sources.
You can contribute to the development of this model by participating in the GitHub repository
https://github.com/andri-jpg/ChatKobi.AI Feel free to raise issues or submit pull requests to help enhance the quality and functionality of the model.
