How to use Inference API (serverless) in my model page?

Hi All.
I have uploaded my fine-tuned model to Hugging Face. I want to create an inference API (serverless) on the model page, but a timeout occurs. What should I do?

Here is how I wrote the README:
pipeline_tag: text-generation
max_new_tokens: 300
- <|end_of_text|>
- <|eot_id|>

Screenshot from 2024-06-28 08-52-33
I am having a similar problem