[`Quantization` / `tests` ] Fix bnb MPT test (#27178)

fix bnb mpt test

tests/quantization/bnb/test_mixed_int8.py

@@ -124,7 +124,6 @@ class MixedInt8Test(BaseMixedInt8Test):
         gc.collect()
         torch.cuda.empty_cache()
 
-    @unittest.skip("Un-skip once https://github.com/mosaicml/llm-foundry/issues/703 is resolved")
     def test_get_keys_to_not_convert_trust_remote_code(self):
         r"""
         Test the `get_keys_to_not_convert` function with `trust_remote_code` models.
@@ -135,11 +134,11 @@ class MixedInt8Test(BaseMixedInt8Test):
 
         model_id = "mosaicml/mpt-7b"
         config = AutoConfig.from_pretrained(
-            model_id, trust_remote_code=True, revision="72e5f594ce36f9cabfa2a9fd8f58b491eb467ee7"
+            model_id, trust_remote_code=True, revision="ada218f9a93b5f1c6dce48a4cc9ff01fcba431e7"
         )
         with init_empty_weights():
             model = AutoModelForCausalLM.from_config(
-                config, trust_remote_code=True, code_revision="72e5f594ce36f9cabfa2a9fd8f58b491eb467ee7"
+                config, trust_remote_code=True, code_revision="ada218f9a93b5f1c6dce48a4cc9ff01fcba431e7"
             )
         self.assertEqual(get_keys_to_not_convert(model), ["transformer.wte"])