Update app.py

app.py

@@ -1,37 +1,27 @@
 import gradio as gr
-import moondream as md
-import os
+import torch
+from PIL import Image
+from transformers import AutoModelForCausalLM, AutoTokenizer
 
-moondream_api_key = os.getenv("MOONDREAM_API_KEY")
-model = md.vl(api_key=moondream_api_key)
+# Load model versi open-source (Gratis selamanya)
+model_id = "vikhyatk/moondream2"
+revision = "2024-08-26"
+model = AutoModelForCausalLM.from_pretrained(model_id, trust_remote_code=True, revision=revision)
+tokenizer = AutoTokenizer.from_pretrained(model_id, revision=revision)
 
-def answer_question(img, prompt):
-    buffer = ""
-    for chunk in model.query(img, prompt, stream=True)["answer"]:
-        buffer += chunk
-        yield buffer
+def predict(image, prompt):
+    if image is None:
+        return "No image uploaded"
+    # Proses gambar ke model
+    enc_image = model.encode_image(image)
+    answer = model.answer_question(enc_image, prompt, tokenizer)
+    return answer
 
-def process_answer(img, answer):
-    return gr.update(visible=False, value=None)
+# Interface Gradio sederhana buat bot PHP lo
+interface = gr.Interface(
+    fn=predict,
+    inputs=[gr.Image(type="pil"), gr.Textbox(label="Prompt")],
+    outputs=gr.Text(),
+)
 
-with gr.Blocks() as demo:
-    gr.Markdown(
-        """
-        # 🌔 moondream2
-        A tiny vision language model. Check out other capabilities (object detection, pointing etc.) in the [Moondream Playground](https://moondream.ai/playground).
-        """
-    )
-    with gr.Row():
-        prompt = gr.Textbox(label="Input", value="Describe this image.", scale=4)
-        submit = gr.Button("Submit")
-    with gr.Row():
-        img = gr.Image(type="pil", label="Upload an Image")
-        with gr.Column():
-            output = gr.Markdown(label="Response")
-            ann = gr.Image(visible=False, label="Annotated Image")
-
-    submit.click(answer_question, [img, prompt], output)
-    prompt.submit(answer_question, [img, prompt], output)
-    output.change(process_answer, [img, output], ann, show_progress=False)
-
-demo.queue().launch()
+interface.launch()