Update app.py

app.py

@@ -224,32 +224,93 @@ def salamandra_chat_endpoint(prompt: str) -> Dict[str, Any]:
     
 # =================== UI ===================
 
+# Main interface for Salamandra 7B Tools: supports tool specification (function calling)
 with gr.Blocks(title="Salamandra 7B Tools · ZeroGPU") as demo:
-    gr.Markdown("## Salamandra-7B-Tools · ZeroGPU\nChat con especificación de herramientas (function-calling).")
+
+    # Header description for the UI
+    gr.Markdown("## Salamandra-7B-Tools · ZeroGPU\nXat amb especificació d'eines (function-calling).")
 
     with gr.Row():
         with gr.Column():
-            messages = gr.Textbox(label="messages_json", value='[{"role":"user","content":"¿Cuánto es (2+2)^3?"}]', lines=6)
-            tools = gr.Textbox(label="tools_json (opcional)", value='[{"type":"function","function":{"name":"calculator","description":"Evalúa expresiones aritméticas básicas.","parameters":{"type":"object","properties":{"expr":{"type":"string"}},"required":["expr"]}}}]', lines=6)
-            max_new = gr.Slider(16, 2048, value=512, step=16, label="max_new_tokens")
-            temp = gr.Slider(0.0, 1.5, value=0.7, step=0.05, label="temperature")
-            topp = gr.Slider(0.1, 1.0, value=0.95, step=0.01, label="top_p")
-            btn = gr.Button("Generar", variant="primary")
-        with gr.Column():
-            out = gr.JSON(label="Salida")
 
-    btn.click(chat_advanced, [messages, tools, max_new, temp, topp], out, api_name="chat", concurrency_limit=1)
+            # JSON array of messages passed to the model
+            messages = gr.Textbox(
+                label="Missatges (JSON)",
+                value='[{"role":"user","content":"Quant és (2+2)^3?"}]',
+                lines=6
+            )
+
+            # Tool definitions in JSON schema format
+            tools = gr.Textbox(
+                label="Eines (JSON, opcional)",
+                value='[{"type":"function","function":{"name":"calculator","description":"Avalua expressions aritmètiques bàsiques.","parameters":{"type":"object","properties":{"expr":{"type":"string"}},"required":["expr"]}}}]',
+                lines=6
+            )
+
+            # Maximum generation length
+            max_new = gr.Slider(16, 2048, value=512, step=16, label="Màxim de tokens nous")
+
+            # Temperature for randomness
+            temp = gr.Slider(0.0, 1.5, value=0.7, step=0.05, label="Temperatura")
 
-    # Endpoint minimalista /predict para ENGINE (mensajes + tools)
-    gr.Button("Probar /predict").click(predict_for_engine, [messages, tools], out, api_name="predict", concurrency_limit=1)
+            # Nucleus sampling threshold
+            topp = gr.Slider(0.1, 1.0, value=0.95, step=0.01, label="Top-p")
 
+            # Button to generate a response
+            btn = gr.Button("Generar", variant="primary")
+
+        with gr.Column():
+            # JSON output from the model
+            out = gr.JSON(label="Sortida")
+
+    # Bind chat-with-tools generation
+    btn.click(
+        chat_advanced,
+        [messages, tools, max_new, temp, topp],
+        out,
+        api_name="chat",
+        concurrency_limit=1
+    )
+
+    # --------------------------------------------------------------
+    gr.Markdown("---")
+    # --------------------------------------------------------------
+
+    # Minimal /predict endpoint for ENGINE compatibility
+    # Accepts messages + tool definitions
+    gr.Button("Provar /predict").click(
+        predict_for_engine,
+        [messages, tools],
+        out,
+        api_name="predict",
+        concurrency_limit=1
+    )
+
+    # --------------------------------------------------------------
+    gr.Markdown("---")
+    # --------------------------------------------------------------
+
+    # Endpoint: raw prompt → model output (JSON)
     with gr.Row():
-        prompt = gr.Textbox(label="prompt", lines=10)
+        prompt = gr.Textbox(label="Prompt", lines=10)
+
     with gr.Row():
         btn2 = gr.Button("Generar", variant="primary")
+
     with gr.Row():
-        out2 = gr.JSON(label="Salida")
+        out2 = gr.JSON(label="Sortida")
+
+    btn2.click(
+        salamandra_chat_endpoint,
+        [prompt],
+        out2,
+        api_name="generate_out_from_prompt",
+        concurrency_limit=1
+    )
 
-    btn2.click(salamandra_chat_endpoint, [prompt], out2, api_name="generate_out_from_prompt", concurrency_limit=1)
+    # --------------------------------------------------------------
+    gr.Markdown("---")
+    # --------------------------------------------------------------
 
+# Enable request queue for concurrency safety
 demo.queue(max_size=16).launch()