Update app.py

app.py

@@ -1,20 +1,24 @@
+import os
 import gradio as gr
 from transformers import AutoModelForCausalLM, AutoTokenizer
 import torch
 
+# Inserir a chave da API da Hugging Face
+os.environ["HF_API_TOKEN"] = "sua_chave_da_api_aqui"
+
 # Carregar os modelos disponíveis
 models = {
     "GPT-MUX (Code)": {
-        "tokenizer": AutoTokenizer.from_pretrained("EleutherAI/gpt-neox-20b"),
-        "model": AutoModelForCausalLM.from_pretrained("EleutherAI/gpt-neox-20b", torch_dtype=torch.float16, device_map="auto")
+        "tokenizer": AutoTokenizer.from_pretrained("EleutherAI/gpt-neox-20b", use_auth_token=True),
+        "model": AutoModelForCausalLM.from_pretrained("EleutherAI/gpt-neox-20b", torch_dtype=torch.float16, device_map="auto", use_auth_token=True)
     },
     "Mistral": {
-        "tokenizer": AutoTokenizer.from_pretrained("mistral_model_endpoint"),
-        "model": AutoModelForCausalLM.from_pretrained("mistral_model_endpoint")
+        "tokenizer": AutoTokenizer.from_pretrained("mistral_model_endpoint", use_auth_token=True),
+        "model": AutoModelForCausalLM.from_pretrained("mistral_model_endpoint", use_auth_token=True)
     },
     "CodeLlama": {
-        "tokenizer": AutoTokenizer.from_pretrained("codellama_model_endpoint"),
-        "model": AutoModelForCausalLM.from_pretrained("codellama_model_endpoint")
+        "tokenizer": AutoTokenizer.from_pretrained("codellama_model_endpoint", use_auth_token=True),
+        "model": AutoModelForCausalLM.from_pretrained("codellama_model_endpoint", use_auth_token=True)
     }
     # Outros modelos podem ser adicionados aqui futuramente
 }