Update app.py

app.py

@@ -6,55 +6,55 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 
 app = FastAPI()
 
-modelo_id = "Qwen/Qwen2.5-Coder-7B-Instruct"
+MODEL = "Qwen/Qwen2.5-Coder-7B-Instruct"
 
-tokenizer = AutoTokenizer.from_pretrained(modelo_id)
+tokenizer = AutoTokenizer.from_pretrained(MODEL)
 
-modelo = None
+model = None
 
 
-class Pedido(BaseModel):
+class Chat(BaseModel):
     model: str
     messages: list
 
 
 @spaces.GPU
-def gerar(prompt):
+def run_ai(prompt):
 
-    global modelo
+    global model
 
-    if modelo is None:
-        modelo = AutoModelForCausalLM.from_pretrained(
-            modelo_id,
+    if model is None:
+        model = AutoModelForCausalLM.from_pretrained(
+            MODEL,
             torch_dtype=torch.float16,
             device_map="auto"
         )
 
-    entrada = tokenizer(
+    inputs = tokenizer(
         prompt,
         return_tensors="pt"
-    ).to(modelo.device)
+    ).to(model.device)
 
-    saida = modelo.generate(
-        **entrada,
+    output = model.generate(
+        **inputs,
         max_new_tokens=1024
     )
 
     return tokenizer.decode(
-        saida[0],
+        output[0],
         skip_special_tokens=True
     )
 
 
 @app.post("/v1/chat/completions")
-def chat(req: Pedido):
+def completions(req: Chat):
 
-    texto = req.messages[-1]["content"]
+    prompt = req.messages[-1]["content"]
 
-    resposta = gerar(texto)
+    resposta = run_ai(prompt)
 
     return {
-        "id": "qwen-local",
+        "id": "qwen-coder",
         "object": "chat.completion",
         "choices": [
             {