Update app.py

app.py

@@ -1,26 +1,50 @@
-import gradio as gr
-from transformers import AutoTokenizer, AutoModelForSequenceClassification
 import torch
-import torch.nn.functional as F
+import gradio as gr
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+# Path to local model in the same repo (e.g., "Mixtral-8x7B-Instruct-v0.1" folder uploaded to Space)
+MODEL_DIR = "Mixtral-8x7B-Instruct-v0.1"
+
+# Load tokenizer and model
+tokenizer = AutoTokenizer.from_pretrained(MODEL_DIR, trust_remote_code=True)
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_DIR,
+    torch_dtype=torch.float16,
+    device_map="auto",
+    trust_remote_code=True
+)
+
+# Generation function
+def generate_text(prompt):
+    messages = [{"role": "user", "content": prompt}]
+    inputs = tokenizer.apply_chat_template(
+        messages,
+        return_tensors="pt",
+        add_generation_prompt=True
+    ).to(model.device)
 
-# Modèle de détection de langue
-model_name = "papluca/xlm-roberta-base-language-detection"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForSequenceClassification.from_pretrained(model_name)
+    output = model.generate(
+        **inputs,
+        max_new_tokens=300,
+        temperature=0.7,
+        top_p=0.95,
+        do_sample=True,
+        pad_token_id=tokenizer.eos_token_id
+    )
 
-# Récupération des étiquettes de langue depuis la config
-id2label = model.config.id2label
+    decoded = tokenizer.decode(output[0], skip_special_tokens=True)
+    if prompt in decoded:
+        return decoded.split(prompt)[-1].strip()
+    return decoded.strip()
 
-# Fonction de détection de langue
-def detect_language(text):
-    inputs = tokenizer(text, return_tensors="pt", truncation=True)
-    with torch.no_grad():
-        outputs = model(**inputs)
-        probs = F.softmax(outputs.logits, dim=1)
-        confidence, predicted_class = torch.max(probs, dim=1)
-        label = id2label[predicted_class.item()]
-        return f"{label} ({confidence.item():.2%} confidence)"
+# Gradio interface
+demo = gr.Interface(
+    fn=generate_text,
+    inputs=gr.Textbox(lines=4, label="Enter your message (FR / AR / EN...)"),
+    outputs="text",
+    title="🧠 Mixtral 8x7B Instruct Chat",
+    description="Multilingual response generation with Mistral Mixtral 8x7B Instruct model.",
+)
 
-# Interface Gradio
-demo = gr.Interface(fn=detect_language, inputs="text", outputs="text", title="Language Detection")
+# Launch
 demo.launch()