app.py

import gradio as gr
from huggingface_hub import InferenceApi

# public repo, no token needed
api = InferenceApi(
    repo_id="bartowski/cognitivecomputations_Dolphin-Mistral-24B-Venice-Edition-GGUF",
    task="text-generation"
)

def respond(prompt):
    # send the prompt, get back a list of dicts
    outputs = api(inputs=prompt, parameters={"max_new_tokens":200, "temperature":0.7})
    # the API returns e.g. [{"generated_text": "..."}]
    return outputs[0]["generated_text"]

gr.Interface(
    fn=respond,
    inputs=gr.Textbox(lines=3, placeholder="ask me anything…"),
    outputs="text",
    title="Dolphin-Mistral-24B via Inference API",
    description="Powered by the HF Inference API—no local model load needed."
).launch(server_name="0.0.0.0", server_port=7860)