Update app.py

app.py

@@ -1,11 +1,7 @@
 import os
 import threading
 import gradio as gr
-from transformers import (
-    AutoModelForCausalLM,
-    AutoTokenizer,
-    TextIteratorStreamer,
-)
+from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 
 MODEL_NAMES = ["LeCarnet-3M", "LeCarnet-8M", "LeCarnet-21M"]
 HF_TOKEN = os.environ.get("HUGGINGFACEHUB_API_TOKEN") or os.getenv("HUGGINGFACEHUB_API_TOKEN")
@@ -13,6 +9,7 @@ MEDIA_PATH = "media/le-carnet.png"
 
 models = {}
 tokenizers = {}
+
 for name in MODEL_NAMES:
     hub_id = f"MaxLSB/{name}"
     tokenizers[name] = AutoTokenizer.from_pretrained(hub_id, token=HF_TOKEN)
@@ -20,14 +17,7 @@ for name in MODEL_NAMES:
     models[name].eval()
 
 
-def respond(
-    prompt: str,
-    chat_history,
-    selected_model: str,
-    max_tokens: int,
-    temperature: float,
-    top_p: float,
-):
+def respond(prompt: str, chat_history, selected_model: str, max_tokens: int, temperature: float, top_p: float):
     tokenizer = tokenizers[selected_model]
     model = models[selected_model]
     inputs = tokenizer(prompt, return_tensors="pt")
@@ -72,24 +62,26 @@ with gr.Blocks(css=".gr-chatbox {height: 600px !important;}") as demo:
                 toggle_btn = gr.Button("Show/hide parameters", elem_id="toggle-btn")
             chat = gr.ChatInterface(
                 fn=respond,
-                additional_inputs=[],
+                additional_inputs=[
+                    gr.Dropdown(MODEL_NAMES, value="LeCarnet-8M", label="Model"),
+                    gr.Slider(1, 512, value=512, step=1, label="Max new tokens"),
+                    gr.Slider(0.1, 2.0, value=0.7, step=0.1, label="Temperature"),
+                    gr.Slider(0.1, 1.0, value=0.9, step=0.05, label="Top‑p"),
+                ],
                 examples=[
                     ["Il était une fois un petit garçon qui vivait dans un village paisible."],
                     ["Il était une fois une grenouille qui rêvait de toucher les étoiles chaque nuit depuis son étang."],
                     ["Il était une fois un petit lapin perdu"],
                 ],
                 cache_examples=False,
+                type="messages",
             )
 
-        with gr.Column(scale=1, visible=True, elem_id="settings-panel") as param_panel:
-            selected_model = gr.Dropdown(MODEL_NAMES, value="LeCarnet-8M", label="Model")
-            max_tokens = gr.Slider(1, 512, value=512, step=1, label="Max new tokens")
-            temperature = gr.Slider(0.1, 2.0, value=0.7, step=0.1, label="Temperature")
-            top_p = gr.Slider(0.1, 1.0, value=0.9, step=0.05, label="Top‑p")
-
-            chat.additional_inputs = [selected_model, max_tokens, temperature, top_p]
+        with gr.Column(scale=1, visible=True, elem_id="settings-panel"):
+            pass  # Inputs are already defined in ChatInterface
 
-    demo.load(None, None, _js="""
+    demo.load(
+        js="""
         () => {
             const toggleBtn = document.querySelector('#toggle-btn button') || document.querySelector('#toggle-btn');
             const panel = document.querySelector('#settings-panel');
@@ -97,7 +89,9 @@ with gr.Blocks(css=".gr-chatbox {height: 600px !important;}") as demo:
                 panel.style.display = (panel.style.display === 'none') ? 'flex' : 'none';
             });
         }
-    """)
+        """
+    )
+
 
 if __name__ == "__main__":
-    demo.queue(default_concurrency_limit=10,max_size=10).launch(ssr_mode=False, max_threads=10)
+    demo.queue(default_concurrency_limit=10, max_size=10).launch(ssr_mode=False, max_threads=10)