Create app.py

app.py

@@ -0,0 +1,32 @@
+import gradio as gr
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+MODEL_PATH = "SafaaAI/final_llm_darija_fr_tech"
+
+# Charger le tokenizer et le modèle en float32 pour CPU
+tokenizer = AutoTokenizer.from_pretrained(MODEL_PATH, trust_remote_code=True)
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_PATH,
+    trust_remote_code=True,
+    torch_dtype=torch.float32
+)
+model.to("cpu")
+model.eval()
+
+def generate_response(prompt):
+    inputs = tokenizer(prompt, return_tensors="pt")
+    with torch.no_grad():
+        outputs = model.generate(**inputs, max_new_tokens=200)
+    return tokenizer.decode(outputs[0], skip_special_tokens=True)
+
+# Interface Gradio
+with gr.Blocks() as demo:
+    gr.Markdown("## SafaaAI LLM Chat (Darija + FR + Technique)")
+    with gr.Row():
+        txt_input = gr.Textbox(label="Votre question", placeholder="Écrire ici...")
+        txt_output = gr.Textbox(label="Réponse du modèle")
+    btn = gr.Button("Envoyer")
+    btn.click(fn=generate_response, inputs=txt_input, outputs=txt_output)
+
+demo.launch()