[Tests] Relax tolerance of flaky failing test (#3755)

relax tolerance slightly

Author: patrickvonplaten

tests/models/test_models_unet_3d_condition.py

@@ -261,7 +261,7 @@ def test_lora_save_load(self):
         with torch.no_grad():
             new_sample = new_model(**inputs_dict, cross_attention_kwargs={"scale": 0.5}).sample
 
-        assert (sample - new_sample).abs().max() < 5e-4
+        assert (sample - new_sample).abs().max() < 1e-3
 
         # LoRA and no LoRA should NOT be the same
         assert (sample - old_sample).abs().max() > 1e-4