chore: update gpus duration and max tokens default

app.py

@@ -7,8 +7,8 @@ import spaces
 import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 
-MAX_MAX_NEW_TOKENS = 2048
-DEFAULT_MAX_NEW_TOKENS = 1024
+MAX_MAX_NEW_TOKENS = 4096
+DEFAULT_MAX_NEW_TOKENS = 1536
 MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "8192"))
 
 DESCRIPTION = """\
@@ -78,12 +78,12 @@ if torch.cuda.is_available():
     )
 
 
-@spaces.GPU(duration=60)
+@spaces.GPU(duration=120)
 def generate(
     message: str,
     chat_history: list[tuple[str, str]],
     system_prompt: str,
-    max_new_tokens: int = 1024,
+    max_new_tokens: int = 1536,
     temperature: float = 0.4,
     top_p: float = 0.95,
     top_k: int = 50,