use deterministic to get stable result (#11663)

* use deterministic to get stable result

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

* add deterministic for int8 test

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

---------

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

Author: jiqing-feng

tests/quantization/bnb/test_4bit.py

@@ -476,6 +476,7 @@ def test_generate_quality_dequantize(self):
         r"""
         Test that loading the model and unquantize it produce correct results.
         """
+        torch.use_deterministic_algorithms(True)
         self.pipeline_4bit.transformer.dequantize()
         output = self.pipeline_4bit(
             prompt=self.prompt,

tests/quantization/bnb/test_mixed_int8.py

@@ -478,6 +478,7 @@ def test_generate_quality_dequantize(self):
         r"""
         Test that loading the model and unquantize it produce correct results.
         """
+        torch.use_deterministic_algorithms(True)
         self.pipeline_8bit.transformer.dequantize()
         output = self.pipeline_8bit(
             prompt=self.prompt,