384

app.py

@@ -52,7 +52,7 @@ def get_completion(msg):
 
     with torch.no_grad():
         outputs = peft_model.generate(
-            **inputs, max_new_tokens=512, pad_token_id = tokenizer.eos_token_id
+            **inputs, max_new_tokens=384, pad_token_id = tokenizer.eos_token_id
         )
 
     #peft_model.to(cpu_device)