Create app.py

app.py

@@ -0,0 +1,51 @@
+import gradio as gr
+from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig, pipeline
+import torch
+
+bnb_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_quant_type="nf4",
+    bnb_4bit_use_double_quant=True,
+)
+
+model_name = "mistralai/Mistral-7B-Instruct-v0.1"
+
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(
+        model_name,
+        load_in_4bit=True,
+        quantization_config=bnb_config,
+        torch_dtype=torch.bfloat16,
+        device_map="auto",
+        trust_remote_code=True,
+    )
+
+device = "cuda"
+
+def greet(input_text):
+  question = input_text
+
+  prompt = f"""<s>[INST] Le contexte est l'assurance maladie en France[/INST]
+    {question}</s>
+    [INST] Rédige un email courtois de réponse en français à la question [/INST]"""
+
+  messages = [
+    {"role": "user", "content": question},
+    {"role": "assistant", "content": "le contexte est l'assurance maladie en France"},
+    {"role": "user", "content": "Rédige un email courtois de réponse en français à la question"}
+]
+
+  encodeds = tokenizer.apply_chat_template(messages, return_tensors="pt")
+
+  model_inputs = encodeds.to(device)
+
+  generated_ids = model.generate(model_inputs, max_new_tokens=1000, do_sample=True)
+  decoded = tokenizer.batch_decode(generated_ids)
+
+  answer = decoded[0].split("[/INST]")[2].replace("</s>", "").replace("[Votre nom]", "").replace("[nom]", "")
+
+  return answer
+
+iface = gr.Interface(fn=greet, inputs=["text"],
+                     outputs="text")
+iface.launch()