Update app.py

app.py

@@ -14,7 +14,7 @@ model = AutoModelForCausalLM.from_pretrained(model_name,
 pipe = pipeline("text-generation", 
                 model=model_name, 
                 tokenizer=tokenizer, 
-                max_new_tokens=150, 
+                max_new_tokens=1500, 
                 temperature=0.7)
 
 def generate_response(message, history):
@@ -36,173 +36,10 @@ def generate_response(message, history):
     response = pipe(messages)
     return response[0][0]['generated_text'][2]['content']
 
-def generate_response_stream(message, history, temperature, top_p, top_k, max_new_tokens, repeat_penalty):
-    chat_messages = [{"role": "system", "content": "Você é ELIZA, uma terapeuta que responde com empatia e faz perguntas para entender melhor o paciente."}]
-    for human, assistant in history:
-        chat_messages.append({"role": "user", "content": human})
-        if assistant is not None:
-            chat_messages.append({"role": "assistant", "content": assistant})
-    chat_messages.append({"role": "user", "content": message})
-    input_ids = tokenizer.apply_chat_template(chat_messages, return_tensors="pt", add_generation_prompt=True).to(model.device)
-    streamer = TextIteratorStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
-    generation_kwargs = dict(
-        input_ids=input_ids,
-        streamer=streamer,
-        max_new_tokens=int(max_new_tokens),
-        temperature=float(temperature),
-        top_p=float(top_p),
-        top_k=int(top_k),
-        repetition_penalty=float(repeat_penalty),
-        do_sample=True,
-    )
-    thread = threading.Thread(target=model.generate, kwargs=generation_kwargs)
-    thread.start()
-    partial_text = ""
-    for new_text in streamer:
-        partial_text += new_text
-        yield partial_text
-
-css = """
-body { background-color: #1a1a2e; }
-.sidebar { background-color: #16213e; border-right: 1px solid #0f3460; padding: 12px; border-radius: 8px; }
-.sidebar label { color: #e0e0e0 !important; font-size: 13px !important; }
-.sidebar .gr-slider { accent-color: #e94560; }
-.chat-area { background-color: #0f3460; border-radius: 8px; }
-.title-bar { color: #e94560; font-family: monospace; }
-.model-info { background-color: #0d1117; border: 1px solid #30363d; border-radius: 6px; padding: 8px; color: #58a6ff; font-family: monospace; font-size: 12px; }
-.gr-button-primary { background-color: #e94560 !important; border: none !important; }
-.gr-button { background-color: #16213e !important; color: #e0e0e0 !important; border: 1px solid #0f3460 !important; }
-footer { display: none !important; }
-"""
-
-with gr.Blocks(css=css, title="Brain map — llama.cpp style") as demo:
-
-    gr.Markdown(
-        """
-        <div class='title-bar'>
-        <h2>🧠 Brain map &nbsp;|&nbsp; Distinguished Medical Assistant</h2>
-        <p style='color:#8b949e;font-size:13px;font-family:monospace;'>Task execution • Organize a clear explanation • Streaming enabled ⚡</p>
-        </div>
-        """
-    )
-
-    with gr.Row(equal_height=True):
-
-        with gr.Column(scale=1, min_width=260, elem_classes="sidebar"):
-            gr.Markdown("### ⚙️ Inference Parameters")
-            temperature_slider = gr.Slider(
-                minimum=0.0, maximum=2.0, value=0.7, step=0.01,
-                label="Temperature",
-                info="Controls randomness. Lower = more deterministic."
-            )
-            top_p_slider = gr.Slider(
-                minimum=0.0, maximum=1.0, value=0.95, step=0.01,
-                label="Top-P (nucleus sampling)",
-                info="Cumulative probability cutoff."
-            )
-            top_k_slider = gr.Slider(
-                minimum=0, maximum=200, value=40, step=1,
-                label="Top-K",
-                info="Limits token candidates at each step."
-            )
-            max_tokens_slider = gr.Slider(
-                minimum=1, maximum=2048, value=150, step=1,
-                label="Max New Tokens",
-                info="Maximum number of tokens to generate."
-            )
-            repeat_penalty_slider = gr.Slider(
-                minimum=1.0, maximum=2.0, value=1.1, step=0.01,
-                label="Repeat Penalty",
-                info="Penalizes repeated tokens."
-            )
-            gr.Markdown("---")
-            gr.Markdown("### 🤖 Model Info")
-            gr.Textbox(
-                value=model_name,
-                label="Loaded Model",
-                interactive=False,
-                elem_classes="model-info"
-            )
-            gr.Textbox(
-                value="float16 · auto device map",
-                label="Precision / Device",
-                interactive=False,
-                elem_classes="model-info"
-            )
-            gr.Markdown("---")
-            gr.Markdown("### 📋 Session")
-            clear_btn = gr.Button("🗑️  Clear Chat", variant="secondary")
-            stop_btn  = gr.Button("⏹️  Stop Generation", variant="stop")
-
-        with gr.Column(scale=4, elem_classes="chat-area"):
-            chatbot = gr.Chatbot(
-                label="Brain map Chat",
-                height=520,
-                show_label=True,
-                avatar_images=(None, "https://huggingface.co/datasets/huggingface/brand-assets/resolve/main/hf-logo.svg"),
-                bubble_full_width=False,
-            )
-            with gr.Row():
-                msg_box = gr.Textbox(
-                    placeholder="Type your message and press Enter or click Send …",
-                    label="",
-                    lines=2,
-                    scale=5,
-                    show_label=False,
-                )
-                send_btn = gr.Button("➤ Send", variant="primary", scale=1, min_width=90)
-
-            gr.Markdown(
-                "<p style='color:#555;font-size:11px;font-family:monospace;text-align:right;'>"
-                "⚡ Streaming • 🔒 Local inference • Brain map v1.0"
-                "</p>"
-            )
-
-    def user_message_submitted(message, history):
-        return "", history + [[message, None]]
-
-    def bot_streaming_response(history, temperature, top_p, top_k, max_new_tokens, repeat_penalty):
-        if not history or history[-1][0] is None:
-            yield history
-            return
-        user_message = history[-1][0]
-        history[-1][1] = ""
-        for partial_output in generate_response_stream(
-            user_message,
-            history[:-1],
-            temperature,
-            top_p,
-            top_k,
-            max_new_tokens,
-            repeat_penalty,
-        ):
-            history[-1][1] = partial_output
-            yield history
-
-    submit_event = msg_box.submit(
-        fn=user_message_submitted,
-        inputs=[msg_box, chatbot],
-        outputs=[msg_box, chatbot],
-        queue=False,
-    ).then(
-        fn=bot_streaming_response,
-        inputs=[chatbot, temperature_slider, top_p_slider, top_k_slider, max_tokens_slider, repeat_penalty_slider],
-        outputs=chatbot,
-    )
-
-    click_event = send_btn.click(
-        fn=user_message_submitted,
-        inputs=[msg_box, chatbot],
-        outputs=[msg_box, chatbot],
-        queue=False,
-    ).then(
-        fn=bot_streaming_response,
-        inputs=[chatbot, temperature_slider, top_p_slider, top_k_slider, max_tokens_slider, repeat_penalty_slider],
-        outputs=chatbot,
-    )
-
-    stop_btn.click(fn=None, cancels=[submit_event, click_event])
-
-    clear_btn.click(fn=lambda: [], inputs=None, outputs=chatbot)
+demo = gr.ChatInterface(
+    generate_response,
+    title="ELIZA (com LLM)",
+    description="Compartilhe seus pensamentos e ELIZA irá ajudar você a refletir sobre eles."
+)
 
 demo.launch()