[StaticQuant] add a linear observer class and test

stack-info: PR: #807, branch: drisspg/stack/8

Author: drisspg

ruff.toml

@@ -8,4 +8,6 @@ include = [
     "torchao/dtypes/nf4tensor.py",
     "test/dtypes/test_nf4.py",
     "torchao/float8/float8_tensor.py",
+    "torchao/quantization/linear_observer_tensor.py",
+    "test/quantization/test_observer.py",
 ]

test/quantization/test_observer.py

@@ -1,20 +1,31 @@
 import torch
+import torch.nn as nn
 from torch.testing._internal.common_utils import TestCase
 from torchao.quantization.observer import (
     AffineQuantizedMinMaxObserver,
     PerTensor,
     PerAxis,
 )
+from torchao.quantization import quantize_
 from torchao.quantization.quant_primitives import (
     MappingType,
 )
+from torchao.quantization.linear_observer_tensor import (
+    insert_observers,
+)
 import unittest
+
 # NOTE: we can copy paste these here if we decide to deprecate them in torch.ao
 from torch.ao.quantization.observer import MinMaxObserver, PerChannelMinMaxObserver
 
+
 class TestQuantFlow(TestCase):
     def _test_obs_helper(self, obs1, obs2):
-        example_inputs = [torch.randn(10, 2048), torch.randn(10, 2048), torch.randn(10, 2048)]
+        example_inputs = [
+            torch.randn(10, 2048),
+            torch.randn(10, 2048),
+            torch.randn(10, 2048),
+        ]
         for example_input in example_inputs:
             obs1(example_input)
             obs2(example_input)
@@ -25,15 +36,98 @@ def _test_obs_helper(self, obs1, obs2):
         self.assertTrue(torch.allclose(zero_point1, zero_point2))
 
     def test_min_max_per_tensor_affine(self):
-        obs = AffineQuantizedMinMaxObserver(MappingType.ASYMMETRIC, torch.uint8, granularity_type=PerTensor(), eps=torch.finfo(torch.float32).eps, scale_dtype=torch.float, zero_point_dtype=torch.int)
+        obs = AffineQuantizedMinMaxObserver(
+            MappingType.ASYMMETRIC,
+            torch.uint8,
+            granularity_type=PerTensor(),
+            eps=torch.finfo(torch.float32).eps,
+            scale_dtype=torch.float,
+            zero_point_dtype=torch.int,
+        )
         ref_obs = MinMaxObserver(dtype=torch.uint8, qscheme=torch.per_tensor_affine)
         self._test_obs_helper(obs, ref_obs)
 
     def test_min_max_per_channel_affine(self):
-        obs = AffineQuantizedMinMaxObserver(MappingType.ASYMMETRIC, torch.uint8, granularity_type=PerAxis(axis=0), eps=torch.finfo(torch.float32).eps, scale_dtype=torch.float, zero_point_dtype=torch.int)
-        ref_obs = PerChannelMinMaxObserver(dtype=torch.uint8, qscheme=torch.per_channel_affine)
+        obs = AffineQuantizedMinMaxObserver(
+            MappingType.ASYMMETRIC,
+            torch.uint8,
+            granularity_type=PerAxis(axis=0),
+            eps=torch.finfo(torch.float32).eps,
+            scale_dtype=torch.float,
+            zero_point_dtype=torch.int,
+        )
+        ref_obs = PerChannelMinMaxObserver(
+            dtype=torch.uint8, qscheme=torch.per_channel_affine
+        )
         self._test_obs_helper(obs, ref_obs)
 
 
+class TestLinearObserver(TestCase):
+    def test_linear_observer_tensor(self):
+        # Create a simple linear layer
+        in_features, out_features = 10, 5
+        linear = nn.Linear(in_features, out_features)
+
+        # Create observers
+        input_observer = AffineQuantizedMinMaxObserver(
+            MappingType.SYMMETRIC,
+            torch.float8_e4m3fn,
+            granularity_type=PerTensor(),
+            eps=torch.finfo(torch.float32).eps,
+            scale_dtype=torch.float,
+            zero_point_dtype=torch.int,
+            zero_point_domain=None,
+        )
+        weight_observer = AffineQuantizedMinMaxObserver(
+            MappingType.SYMMETRIC,
+            torch.float8_e4m3fn,
+            granularity_type=PerTensor(),
+            eps=torch.finfo(torch.float32).eps,
+            scale_dtype=torch.float,
+            zero_point_dtype=torch.int,
+            zero_point_domain=None,
+        )
+
+        # Wrap the weight with LinearObserverTensor
+        quantize_(linear, insert_observers(input_observer, weight_observer))
+
+        # Create some example inputs
+        example_inputs = [torch.randn(5, in_features) for _ in range(3)]
+        max_val = 42.1234
+        min_val = -39.760
+        big_tensor = torch.full((6, in_features), max_val)
+        small_tensor = torch.full((40, in_features), min_val)
+        example_inputs.extend([big_tensor, small_tensor])
+
+        # Run forward passes
+        for example_input in example_inputs:
+            _ = linear(example_input)
+
+        input_observer = linear.weight.input_observer
+        weight_observer = linear.weight.weight_observer
+
+        # Check that the observers have recorded statistics
+        assert input_observer.min_val == min_val
+        assert input_observer.max_val == max_val
+
+        # Calculate qparams and ensure they're not None
+        input_scale, input_zero_point = input_observer.calculate_qparams()
+        weight_scale, weight_zero_point = weight_observer.calculate_qparams()
+
+        max_fp8 = torch.finfo(torch.float8_e4m3fn).max
+        self.assertEqual(
+            input_scale.item(),
+            max_val / max_fp8,
+        )
+        self.assertIsNotNone(input_zero_point)
+        torch.testing.assert_close(
+            weight_scale,
+            torch.max(linear.weight.original_weight_tensor) / max_fp8,
+            atol=5e-5,
+            rtol=0.0,
+        )
+        self.assertIsNotNone(weight_zero_point)
+
+
 if __name__ == "__main__":
     unittest.main()

torchao/quantization/linear_observer_tensor.py

@@ -0,0 +1,205 @@
+import torch
+import torch.nn as nn
+from torchao.utils import (
+    _implements,
+    _dispatch__torch_function__,
+    _dispatch__torch_dispatch__,
+)
+from typing import Callable, Optional, Dict
+from torch.utils._python_dispatch import return_and_correct_aliasing
+from torchao.utils import (
+    TorchAOBaseTensor,
+    TORCH_VERSION_AT_LEAST_2_5,
+)
+
+from torchao.quantization.observer import AffineQuantizedObserverBase
+
+__all__ = [
+    "LinearObserverTensor",
+    "to_linear_observer_tensor",
+    "insert_observers",
+]
+
+aten = torch.ops.aten
+Tensor = torch.Tensor
+
+
+class LinearObserverTensor(TorchAOBaseTensor):
+    """
+    This subclass of Tensor is used in conjuction with a static calibration flow.
+    The flow is broken up into 3 parts;
+        1. Insert the LinearObserverTensor subclass into the model's nn.Linear layers
+        2. Run the model with a calibration dataset, the observer will record the min/max of the input and weight
+        3. quantize_ the model to static using the statistics recorded by the observer
+
+    This subclass wraps the original weight tensor on the nn.Linear layer. When forward is called, the observer
+    will first calculat statistics on BOTH the input and weight, and then run the linear op.
+    """
+
+    original_weight_tensor: torch.Tensor
+    input_observer: Optional[AffineQuantizedObserverBase]
+    weight_observer: Optional[AffineQuantizedObserverBase]
+
+    def __new__(
+        cls,
+        original_weight_tensor: torch.Tensor,
+        input_observer: Optional[AffineQuantizedObserverBase] = None,
+        weight_observer: Optional[AffineQuantizedObserverBase] = None,
+    ):
+        kwargs = {}
+        dtype = original_weight_tensor.dtype
+        kwargs["dtype"] = dtype
+        kwargs["requires_grad"] = False
+        kwargs["device"] = original_weight_tensor.device
+        shape = original_weight_tensor.shape
+        return torch.Tensor._make_wrapper_subclass(cls, shape, **kwargs)  # type: ignore[attr-defined]
+
+    def __init__(
+        self,
+        original_weight_tensor: torch.Tensor,
+        input_observer: Optional[AffineQuantizedObserverBase] = None,
+        weight_observer: Optional[AffineQuantizedObserverBase] = None,
+    ):
+        self.original_weight_tensor = original_weight_tensor
+        self.input_observer = input_observer
+        self.weight_observer = weight_observer
+
+    def __repr__(self):
+        return (
+            f"LinearObserverTensor(\n"
+            f"original_weight={self.original_weight_tensor}\n"
+            f"input_observer={self.input_observer.__class__.__name__ if self.input_observer else None}\n"
+            f"weight_observer={self.weight_observer.__class__.__name__ if self.weight_observer else None}\n)"
+        )
+
+    def __tensor_flatten__(self):
+        return ["original_weight_tensor"], [self.input_observer, self.weight_observer]
+
+    @classmethod
+    def __tensor_unflatten__(
+        cls,
+        tensor_data_dict: Dict[str, Tensor],
+        tensor_attributes,
+        outer_size,
+        outer_stride,
+    ):
+        original_weight_tensor = tensor_data_dict["original_weight_tensor"]
+        (input_observer, weight_observer) = tensor_attributes
+        return cls(original_weight_tensor, input_observer, weight_observer)
+
+    @classmethod
+    def from_float(
+        cls,
+        original_weight_tensor: Tensor,
+        input_observer: Optional[AffineQuantizedObserverBase] = None,
+        weight_observer: Optional[AffineQuantizedObserverBase] = None,
+    ):
+        return cls(original_weight_tensor, input_observer, weight_observer)
+
+    def _apply_fn_to_data(self, fn: Callable):
+        """Applies a fn to the tensor component of the LinearObserverTensor"""
+        return self.__class__(
+            fn(self.original_weight_tensor),
+            self.input_observer,
+            self.weight_observer,
+        )
+
+    def to(self, *args, **kwargs):
+        kwargs = self._get_to_kwargs(*args, **kwargs)
+        return self._apply_fn_to_data(lambda x: x.to(**kwargs))
+
+    implements = classmethod(_implements)
+    __torch_function__ = classmethod(_dispatch__torch_function__)
+    __torch_dispatch__ = classmethod(_dispatch__torch_dispatch__)
+
+
+implements = LinearObserverTensor.implements
+
+
+@implements(torch.nn.functional.linear)
+def _(func, types, args, kwargs):
+    input_tensor, weight_tensor, bias = (
+        args[0],
+        args[1],
+        args[2] if len(args) > 2 else None,
+    )
+    if weight_tensor.input_observer is not None:
+        input_tensor = weight_tensor.input_observer(input_tensor)
+    if weight_tensor.weight_observer is not None:
+        weight_tensor = weight_tensor.weight_observer(
+            weight_tensor.original_weight_tensor
+        )
+
+    return torch.nn.functional.linear(input_tensor, weight_tensor, bias)
+
+
+@implements(aten.detach.default)
+def _(func, types, args, kwargs):
+    return return_and_correct_aliasing(
+        func, args, kwargs, args[0]._apply_fn_to_data(torch.detach)
+    )
+
+
+@implements(aten.clone.default)
+def _(func, types, args, kwargs):
+    return return_and_correct_aliasing(
+        func, args, kwargs, args[0]._apply_fn_to_data(torch.clone)
+    )
+
+
+@implements(aten._to_copy.default)
+def _(func, types, args, kwargs):
+    return return_and_correct_aliasing(
+        func,
+        args,
+        kwargs,
+        args[0].to(*args[1:], **kwargs)._apply_fn_to_data(torch.clone),
+    )
+
+
+to_linear_observer_tensor = LinearObserverTensor.from_float
+
+if TORCH_VERSION_AT_LEAST_2_5:
+    # Allow a model with LinearActivationQuantizedTensor weights to be loaded with `weights_only=True`
+    torch.serialization.add_safe_globals([LinearObserverTensor])
+
+
+def insert_observers(
+    input_observer: Optional[AffineQuantizedObserverBase],
+    weight_observer: Optional[AffineQuantizedObserverBase],
+) -> Callable:
+    """
+    Converts the weight of a linear module to a LinearObserverTensor.
+
+    This function wraps the weight of the given linear module with a LinearObserverTensor,
+    which enables observation of both input and weight tensors during forward passes.
+    The wrapped weight is then re-wrapped as a nn.Parameter to maintain compatibility
+    with PyTorch's module system.
+
+    Usaage:
+    ```
+    linear_module = nn.Linear(10, 20)
+    quantize_(linear_module, convert_linear_weight_to_observer))
+
+    Args:
+        linear_module (nn.Linear): The linear module to be converted.
+        input_observer (Optional[AffineQuantizedObserverBase]): Observer for input tensor.
+        weight_observer (Optional[AffineQuantizedObserverBase]): Observer for weight tensor.
+
+    Returns:
+        nn.Linear: The modified linear module with its weight wrapped in a LinearObserverTensor.
+    """
+
+    def convert_to_linear_observer(linear_module: nn.Linear):
+        # Wrap the weight with LinearObserverTensor and then with nn.Parameter
+        linear_module.weight = nn.Parameter(
+            to_linear_observer_tensor(
+                linear_module.weight,
+                input_observer=input_observer,
+                weight_observer=weight_observer,
+            ),
+            requires_grad=linear_module.weight.requires_grad,
+        )
+        return linear_module
+
+    return convert_to_linear_observer