Create app.py

app.py

@@ -0,0 +1,69 @@
+import gradio as gr
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+MODEL_ID = "Hodely/AmInSide-1.0"
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_ID,
+    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32
+)
+
+if tokenizer.pad_token is None:
+    tokenizer.pad_token = tokenizer.eos_token
+
+
+SYSTEM_PROMPT = (
+    "Eres AmInSide 1.0, una IA conversacional útil, clara, creativa e inteligente. "
+    "Responde de forma natural, bien escrita y precisa."
+)
+
+def build_prompt(message, history):
+    prompt = SYSTEM_PROMPT + "\n\n"
+    for user_msg, bot_msg in history:
+        prompt += f"Usuario: {user_msg}\nAsistente: {bot_msg}\n"
+    prompt += f"Usuario: {message}\nAsistente:"
+    return prompt
+
+def chat(message, history):
+    history = history or []
+    prompt = build_prompt(message, history)
+
+    inputs = tokenizer(prompt, return_tensors="pt", truncation=True, max_length=1024)
+    inputs = {k: v.to(model.device) for k, v in inputs.items()}
+
+    with torch.no_grad():
+        output = model.generate(
+            **inputs,
+            max_new_tokens=180,
+            temperature=0.8,
+            do_sample=True,
+            top_p=0.9,
+            repetition_penalty=1.1,
+            pad_token_id=tokenizer.eos_token_id
+        )
+
+    full_text = tokenizer.decode(output[0], skip_special_tokens=True)
+
+    if "Asistente:" in full_text:
+        response = full_text.split("Asistente:")[-1].strip()
+    else:
+        response = full_text[len(prompt):].strip()
+
+    history.append((message, response))
+    return history, history
+
+with gr.Blocks(title="AmInSide 1.0") as demo:
+    gr.Markdown("# AmInSide 1.0")
+    gr.Markdown("Chat de demostración del modelo.")
+
+    chatbot = gr.Chatbot(height=500)
+    msg = gr.Textbox(label="Escribe tu mensaje", placeholder="Habla con AmInSide 1.0...")
+    state = gr.State([])
+
+    msg.submit(chat, inputs=[msg, state], outputs=[chatbot, state])
+    msg.submit(lambda: "", None, msg)
+
+demo.launch()