Update app.py

app.py

@@ -7,8 +7,10 @@ from typing import Any, Dict, List
 import gradio as gr
 from openai import OpenAI
 
-DEFAULT_MODEL = os.getenv("DEFAULT_MODEL", "LLaVA-OneVision-1.5-8B-Instruct")
+# Modelo por defecto
+DEFAULT_MODEL = "LLaVA-OneVision-1.5-8B-Instruct"
 
+# Cliente OpenAI-compatible (usa el endpoint de Hugging Face o el tuyo)
 _client = OpenAI(
     base_url=os.getenv("BASE_URL", ""),
     api_key=os.getenv("API_KEY", ""),
@@ -37,6 +39,15 @@ def _message(role: str, content: Any) -> Dict[str, Any]:
 def _build_user_message(message: Dict[str, Any]) -> Dict[str, Any]:
     files = message.get("files") or []
     text = (message.get("text") or "").strip()
+
+    # 🔹 Si no hay texto, añadimos un prompt nutricional por defecto
+    if not text:
+        text = (
+            "Analiza la imagen del plato de comida y describe los alimentos que contiene. "
+            "Indica una estimación de calorías, proteínas, carbohidratos y grasas. "
+            "Responde en formato breve y estructurado."
+        )
+
     content: List[Dict[str, Any]] = [_image_content(p) for p in files]
     if text:
         content.append(_text_content(text))
@@ -53,8 +64,7 @@ def _convert_history(history: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
             if isinstance(content, str):
                 user_content.append(_text_content(content))
             elif isinstance(content, tuple):
-                user_content.extend(_image_content(path)
-                                    for path in content if path)
+                user_content.extend(_image_content(path) for path in content if path)
         elif role == "assistant":
             msgs.append(_message("user", user_content.copy()))
             user_content.clear()
@@ -69,7 +79,7 @@ def stream_response(message: Dict[str, Any], history: List[Dict[str, Any]], mode
         stream = _client.chat.completions.create(
             model=model_name,
             messages=messages,
-            temperature=0.000001,
+            temperature=0.1,
             top_p=1,
             extra_body={
                 "repetition_penalty": 1.05,
@@ -85,20 +95,20 @@ def stream_response(message: Dict[str, Any], history: List[Dict[str, Any]], mode
                 partial += delta
                 yield partial
     except Exception as e:
-        yield f"Failed to get response: {e}"
+        yield f"⚠️ Error al obtener respuesta: {e}"
 
 
 def build_demo() -> gr.Blocks:
     chatbot = gr.Chatbot(type="messages", allow_tags=["think"])
     textbox = gr.MultimodalTextbox(
         show_label=False,
-        placeholder="Enter text, or upload one or more images...",
+        placeholder="Subí una foto de tu comida para analizarla...",
         file_types=["image"],
         file_count="single",
         max_plain_text_length=32768
     )
     model_selector = gr.Dropdown(
-        label="Model",
+        label="Modelo",
         choices=[
             ("LLaVA-OneVision-1.5-8B-Instruct", "LLaVA-OneVision-1.5-8B-Instruct"),
             ("LLaVA-OneVision-1.5-4B-Instruct", "LLaVA-OneVision-1.5-4B-Instruct"),
@@ -111,12 +121,14 @@ def build_demo() -> gr.Blocks:
         multimodal=True,
         chatbot=chatbot,
         textbox=textbox,
-        title="LLaVA-OneVision-1.5: Fully Open Framework for Democratized Multimodal Training",
-        description="""**LLaVA-OneVision1.5** introduces a novel family of fully open-source Large Multimodal Models (LMMs) that achieves state-of-the-art performance with substantially lower cost through training on native resolution images.
-
-🔗 **Links**: [GitHub](https://github.com/EvolvingLMMs-Lab/LLaVA-OneVision-1.5) | [HuggingFace](https://huggingface.co/lmms-lab)""",
+        title="🍽️ NasFit Vision AI",
+        description=(
+            "Subí una foto de tu comida y NasFit IA estimará su contenido nutricional. "
+            "Basado en **LLaVA-OneVision-1.5**, modelo multimodal open source con análisis visual avanzado. "
+            "Ideal para tracking nutricional inteligente."
+        ),
         additional_inputs=[model_selector],
-        additional_inputs_accordion=gr.Accordion("Options", open=True),
+        additional_inputs_accordion=gr.Accordion("Opciones avanzadas", open=False),
     ).queue(default_concurrency_limit=8)
 
 
@@ -125,5 +137,4 @@ def main():
 
 
 if __name__ == "__main__":
-    main()
-
+    main()