Update app.py

app.py

@@ -1,3 +1,4 @@
+import os
 from threading import Thread
 from typing import Iterator
 
@@ -7,6 +8,10 @@ import torch
 from huggingface_hub import InferenceClient
 from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 
+MAX_MAX_NEW_TOKENS = 2048
+DEFAULT_MAX_NEW_TOKENS = 1024
+MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))
+
 #Inference API Code
 client = InferenceClient("BenBranyon/zephyr-sumbot-all-songs")