enable/disable compile for quants methods (huggingface#36519)

* disable compile for most quants methods

* fix

* Update src/transformers/generation/configuration_utils.py

Co-authored-by: Matthew Douglas <[email protected]>

* Update tests/quantization/bnb/test_mixed_int8.py

Co-authored-by: Matthew Douglas <[email protected]>

* Update src/transformers/generation/configuration_utils.py

Co-authored-by: Joao Gante <[email protected]>

* changes from joao suggestions

---------

Co-authored-by: Matthew Douglas <[email protected]>
Co-authored-by: Joao Gante <[email protected]>

Author: 3 people

src/transformers/generation/configuration_utils.py

@@ -379,8 +379,7 @@ class GenerationConfig(PushToHubMixin):
             If using a static cache, this controls how `generate` will `compile` the forward pass for performance
             gains.
 
-        disable_compile (`bool`, *optional*): Whether to disable the compilation of the forward pass when using 'statis' cache
-            implementation.
+        disable_compile (`bool`, *optional*): Whether to disable the automatic compilation of the forward pass. Automatic compilation happens when specific criteria are met, including using a compileable cache. Please open an issue if you find the need to use this flag.
 
         > Wild card
 

src/transformers/generation/utils.py

@@ -1613,7 +1613,6 @@ def _prepare_generation_config(
             model_kwargs = generation_config.update(**kwargs)
         else:
             model_kwargs = kwargs
-
         return generation_config, model_kwargs
 
     def _get_initial_cache_position(self, input_ids, model_kwargs):
@@ -3281,7 +3280,9 @@ def _sample(
         model_forward = self.__call__
         if isinstance(model_kwargs.get("past_key_values"), Cache):
             is_compileable = model_kwargs["past_key_values"].is_compileable and self._supports_static_cache
-            is_compileable = is_compileable and not self.generation_config.disable_compile
+            if getattr(self, "hf_quantizer", None) is not None:
+                is_compileable &= self.hf_quantizer.is_compileable
+            is_compileable = is_compileable and not generation_config.disable_compile
             if is_compileable and (
                 self.device.type == "cuda" or generation_config.compile_config._compile_all_devices
             ):

src/transformers/quantizers/base.py

@@ -271,6 +271,11 @@ def is_qat_trainable(self) -> bool:
         """Flag indicating whether the quantized model can carry out quantization aware training"""
         return False
 
+    @property
+    def is_compileable(self) -> bool:
+        """Flag indicating whether the quantized model can be compiled"""
+        return False
+
     @abstractmethod
     def _process_model_before_weight_loading(self, model, **kwargs): ...
 

src/transformers/quantizers/quantizer_torchao.py

@@ -243,3 +243,7 @@ def is_trainable(self):
             "int8_dynamic_activation_int8_weight",
         ]
         return self.quantization_config.quant_type in supported_quant_types_for_training
+
+    @property
+    def is_compileable(self) -> bool:
+        return True

tests/quantization/bnb/test_4bit.py

@@ -771,3 +771,36 @@ def test_set_load_in_8_bit(self):
         quantization_config = BitsAndBytesConfig(load_in_4bit=True)
         with self.assertRaisesRegex(ValueError, "load_in_4bit and load_in_8bit are both True"):
             quantization_config.load_in_8bit = True
+
+
+@require_bitsandbytes
+@require_accelerate
+@require_torch_gpu_if_bnb_not_multi_backend_enabled
+@slow
+@apply_skip_if_not_implemented
+class Bnb4bitCompile(unittest.TestCase):
+    model_name = "hf-internal-testing/tiny-random-LlamaForCausalLM"
+    input_text = "Hello my name is"
+
+    def setUp(self):
+        # Models and tokenizer
+        self.tokenizer = AutoTokenizer.from_pretrained(self.model_name)
+        self.model_4bit = AutoModelForCausalLM.from_pretrained(self.model_name, load_in_4bit=True)
+
+    def test_generate_compile(self):
+        encoded_input = self.tokenizer(self.input_text, return_tensors="pt")
+
+        # if nothing is set, compile will be disabled for bnb
+        self.model_4bit.generate(
+            input_ids=encoded_input["input_ids"].to(self.model_4bit.device),
+            max_new_tokens=10,
+            cache_implementation="static",
+        )
+        with self.assertRaises(Exception):
+            # overwrite property
+            object.__setattr__(self.model_4bit.hf_quantizer, "is_compileable", True)
+            self.model_4bit.generate(
+                input_ids=encoded_input["input_ids"].to(self.model_4bit.device),
+                max_new_tokens=10,
+                cache_implementation="static",
+            )

tests/quantization/bnb/test_mixed_int8.py

@@ -966,3 +966,37 @@ def test_int8_from_pretrained(self):
         output_sequences = model.generate(input_ids=encoded_input["input_ids"].to(torch_device), max_new_tokens=10)
 
         self.assertIn(self.tokenizer.decode(output_sequences[0], skip_special_tokens=True), self.EXPECTED_OUTPUTS)
+
+
+@require_bitsandbytes
+@require_accelerate
+@require_torch
+@require_torch_gpu_if_bnb_not_multi_backend_enabled
+@slow
+@apply_skip_if_not_implemented
+class Bnb8bitCompile(unittest.TestCase):
+    model_name = "hf-internal-testing/tiny-random-LlamaForCausalLM"
+    input_text = "Hello my name is"
+
+    def setUp(self):
+        # Models and tokenizer
+        self.tokenizer = AutoTokenizer.from_pretrained(self.model_name)
+        self.model_8bit = AutoModelForCausalLM.from_pretrained(self.model_name, load_in_8bit=True)
+
+    def test_generate_compile(self):
+        encoded_input = self.tokenizer(self.input_text, return_tensors="pt")
+
+        # if nothing is set, compile will be disabled for bnb
+        self.model_8bit.generate(
+            input_ids=encoded_input["input_ids"].to(self.model_8bit.device),
+            max_new_tokens=10,
+            cache_implementation="static",
+        )
+
+        with self.assertRaises(Exception):
+            object.__setattr__(self.model_8bit.hf_quantizer, "is_compileable", True)
+            self.model_8bit.generate(
+                input_ids=encoded_input["input_ids"].to(self.model_8bit.device),
+                max_new_tokens=10,
+                cache_implementation="static",
+            )