On Hugging Face Inference API, the pipeline_tag and base_model allow it to work directly.
Locally, you must load it with the base model for text generation.

How to Use Locally

python

from unsloth import FastLanguageModel

base_model = "unsloth/meta-llama-3.1-8b-instruct-bnb-4bit"
adapter_model = "vyshnaviprasad/diabetic_modelf"

# Load base model with LoRA adapter
model, tokenizer = FastLanguageModel.from_pretrained(
    model_name=base_model,
    adapter_name=adapter_model,
    load_in_4bit=True
)

FastLanguageModel.for_inference(model)

# Generate text
prompt = "Explain hypertension in simple words to a patient."
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens=1000)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))