Support DeepSeekV3-style block FP8 quantization

src/compressed_tensors/quantization/lifecycle/forward.py

@@ -189,7 +189,50 @@ def _process_quantization(
     q_min, q_max = calculate_range(args, x.device)
     group_size = args.group_size
 
-    if args.strategy in (QuantizationStrategy.GROUP, QuantizationStrategy.TENSOR_GROUP):
+    # blockwise FP8: quantize per 2D block, supports block_structure for static block quant
+    if args.strategy == QuantizationStrategy.BLOCK:
+        # x: [*, H, W] or [H, W]
+        bs = args.block_structure
+        if not (isinstance(bs, (list, tuple)) and len(bs) == 2 and all(isinstance(dim, int) for dim in bs)):
+            raise ValueError(f"Invalid block_structure '{bs}'. Must be a list of two ints [rows, cols].")
+        b_h, b_w = bs
+        rows, cols = x.shape[-2], x.shape[-1]
+        # ensure exact division (assumes shapes match spec)
+        n_rb = ceil(rows / b_h)
+        n_cb = ceil(cols / b_w)
+        # reshape into blocks
+        x_blocks = x.reshape(
+            n_rb,
+            b_h,
+            n_cb,
+            b_w,
+        ).transpose(1, 2)
+        # expand scale/zero_point for blocks
+        sb = scale.unsqueeze(-1).unsqueeze(-1)
+        zb = zero_point.unsqueeze(-1).unsqueeze(-1) if zero_point is not None else None
+        # quantize blocks
+        qb = _quantize(
+            x=x_blocks,
+            scale=sb,
+            zero_point=zb,
+            q_min=q_min,
+            q_max=q_max,
+            args=args,
+            dtype=dtype,
+            global_scale=global_scale,
+        ) if do_quantize else x_blocks
+        # dequantize if needed
+        if do_dequantize:
+            qb = _dequantize(
+                x_q=qb,
+                scale=sb,
+                zero_point=zb,
+                global_scale=global_scale,
+            )
+        # restore original shape
+        out = qb.transpose(1, 2).reshape(rows, cols)
+        return out
+    elif args.strategy in (QuantizationStrategy.GROUP, QuantizationStrategy.TENSOR_GROUP):
         n_dims = x.shape
         if len(n_dims) > 2:
             x = x.squeeze(0)

src/compressed_tensors/quantization/quant_args.py

@@ -14,7 +14,7 @@
 
 import warnings
 from enum import Enum
-from typing import Any, Dict, Optional, Union
+from typing import Any, Dict, List, Optional, Union
 
 import torch
 from compressed_tensors.utils import Aliasable
@@ -153,8 +153,8 @@ class QuantizationArgs(BaseModel, use_enum_values=True):
     :param symmetric: whether or not quantization scale is symmetric about zero-point
     :param strategy: string id determining the scope of scale/zero-point to apply
     :param group_size: group length to use for the group strategy
-    :param block_structure: 2d block structure to use for the block strategy, must be
-    of the format "2x4", "8x16", etc.
+    :param block_structure: 2d block structure to use for the block strategy; must be
+        a list of two ints [rows, cols] like [128, 128].
     :param dynamic: set True to perform dynamic quantization - values will not be
         calibrated during calibration phase, instead during inference new quantization
         ranges will be observed with every sample. Defaults to False for static
@@ -169,7 +169,7 @@ class QuantizationArgs(BaseModel, use_enum_values=True):
     symmetric: bool = True
     group_size: Optional[int] = None
     strategy: Optional[QuantizationStrategy] = None
-    block_structure: Optional[str] = None
+    block_structure: Optional[List[int]] = None
     dynamic: Union[DynamicType, bool] = False
     actorder: Union[ActivationOrdering, bool, None] = None
     observer: Optional[str] = Field(
@@ -207,6 +207,28 @@ def validate_group(cls, value) -> Union[int, None]:
 
         return value
 
+    @field_validator("block_structure", mode="before")
+    def validate_block_structure(cls, value) -> Optional[List[int]]:
+        if value is None:
+            return value
+        # For backward compatibility, allow string format "2x4", "8x16", etc.
+        if isinstance(value, str):
+            try:
+                return [int(x) for x in value.split("x")]
+            except Exception:
+                raise ValueError(
+                    f"Invalid block_structure '{value}'. Must be a list of two ints [rows, cols]."
+                )
+        if isinstance(value, (list, tuple)):
+            if len(value) != 2 or not all(isinstance(v, int) for v in value):
+                raise ValueError(
+                    f"Invalid block_structure '{value}'. Must be a list of two ints [rows, cols]."
+                )
+            return list(value)
+        raise ValueError(
+            f"Invalid block_structure '{value}'. Must be a list of two ints [rows, cols]."
+        )
+
     @field_validator("strategy", mode="before")
     def validate_strategy(cls, value) -> Union[QuantizationStrategy, None]:
         if isinstance(value, str):
@@ -277,14 +299,15 @@ def validate_model_after(model: "QuantizationArgs") -> "QuantizationArgs":
 
         # infer observer w.r.t. dynamic
         if dynamic:
-            if strategy not in (
+            supported_strategies = (
                 QuantizationStrategy.TOKEN,
                 QuantizationStrategy.TENSOR,
                 QuantizationStrategy.TENSOR_GROUP,
-            ):
+                QuantizationStrategy.GROUP,
+            )
+            if strategy not in supported_strategies:
                 raise ValueError(
-                    f"One of {(QuantizationStrategy.TOKEN, QuantizationStrategy.TENSOR, QuantizationStrategy.TENSOR_GROUP)} "
-                    "must be used for dynamic quantization",
+                    f"One of {supported_strategies} must be used for dynamic quantization"
                 )
 
             if (

src/compressed_tensors/quantization/quant_scheme.py

@@ -243,6 +243,29 @@ def is_preset_scheme(name: str) -> bool:
     ),
 )
 
+# Block‐wise FP8 (deepseekv3-style quantization):
+# static 128x128 per‐block weights and 
+# dynamic per‐token‐group activations
+FP8_BLOCK = dict(
+    weights=QuantizationArgs(
+        num_bits=8,
+        type=QuantizationType.FLOAT,
+        strategy=QuantizationStrategy.BLOCK,
+        symmetric=True,
+        dynamic=False,
+        block_structure=[128, 128],
+    ),
+    input_activations=QuantizationArgs(
+        num_bits=8,
+        type=QuantizationType.FLOAT,
+        strategy=QuantizationStrategy.GROUP,
+        symmetric=True,
+        dynamic=True,
+        observer=None,
+        group_size=128,
+    ),
+)
+
 PRESET_SCHEMES = {
     # Unquantized (no-op)
     "UNQUANTIZED": UNQUANTIZED,
@@ -257,6 +280,7 @@ def is_preset_scheme(name: str) -> bool:
     # Float weight and activation schemes
     "FP8": FP8,
     "FP8_DYNAMIC": FP8_DYNAMIC,
+    "FP8_BLOCK": FP8_BLOCK,
     "NVFP4A16": NVFP4A16,
     "NVFP4": NVFP4,
 }

src/compressed_tensors/quantization/utils/helpers.py

@@ -171,7 +171,7 @@ def compute_dynamic_scales_and_zp(
         reduce_dims = tuple(idx for idx in range(value.ndim) if idx not in dim)
     elif args.strategy == QuantizationStrategy.TENSOR:
         reduce_dims = None
-    elif args.strategy == QuantizationStrategy.TENSOR_GROUP:
+    elif args.strategy in (QuantizationStrategy.TENSOR_GROUP, QuantizationStrategy.GROUP):
         if len(value.shape) > 2:
             value = value.squeeze(0)
 
@@ -189,7 +189,12 @@ def compute_dynamic_scales_and_zp(
     else:
         raise ValueError(
             "Dynamic quantization is only supported for ",
-            f"{QuantizationStrategy.TOKEN, QuantizationStrategy.TENSOR, QuantizationStrategy.TENSOR_GROUP}",
+            f"{(
+                QuantizationStrategy.TOKEN,
+                QuantizationStrategy.TENSOR,
+                QuantizationStrategy.TENSOR_GROUP,
+                QuantizationStrategy.GROUP,
+            )}",
         )
 
     if not reduce_dims:

tests/test_examples/test_bitmask_compression_ipynb.py

@@ -12,8 +12,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import nbformat
 import pytest
+nbformat = pytest.importorskip("nbformat")
 from nbconvert.preprocessors import ExecutePreprocessor
 
 

tests/test_quantization/lifecycle/test_forward.py

@@ -30,6 +30,10 @@
 )
 from compressed_tensors.quantization.quant_config import QuantizationStatus
 from torch.nn import Linear
+import math
+
+from compressed_tensors.quantization.lifecycle.forward import _process_quantization
+from compressed_tensors.quantization.utils.helpers import calculate_range
 
 
 def make_dummy_g_idx(columns: int, group_size: int) -> torch.Tensor:
@@ -203,3 +207,49 @@ def test_dequantize(num_bits, type, strategy, group_size, scale, zero_point, g_i
         dtype=None,
         g_idx=g_idx,
     )
+
+
+def test_process_quantization_block_static():
+    """
+    Static block quantization (QuantizationStrategy.BLOCK) should split a 2D tensor
+    into blocks, quantize each block, and reassemble without changing shape.
+    """
+    rows, cols = 8, 8
+    bh, bw = 2, 4
+    x = torch.randn(rows, cols)
+    args = QuantizationArgs(
+        num_bits=8,
+        type="float",
+        strategy=QuantizationStrategy.BLOCK,
+        symmetric=True,
+        dynamic=False,
+        block_structure=[bh, bw],
+    )
+    num_rb = math.ceil(rows / bh)
+    num_cb = math.ceil(cols / bw)
+    scale = torch.rand(num_rb, num_cb) + 0.1
+    zp = torch.zeros_like(scale)
+    q_min, q_max = calculate_range(args, x.device)
+    out = _process_quantization(
+        x=x,
+        scale=scale,
+        zero_point=zp,
+        args=args,
+        do_quantize=True,
+        do_dequantize=False,
+        dtype=None,
+        global_scale=None,
+    )
+    assert out.shape == x.shape
+    # full fake-quantize roundtrip
+    out2 = _process_quantization(
+        x=x,
+        scale=scale,
+        zero_point=zp,
+        args=args,
+        do_quantize=True,
+        do_dequantize=True,
+        dtype=None,
+        global_scale=None,
+    )
+    assert out2.shape == x.shape

tests/test_quantization/test_quant_args.py

@@ -59,7 +59,7 @@ def test_block():
 
     block = QuantizationArgs(**kwargs)
     assert block.strategy == QuantizationStrategy.BLOCK
-    assert block.block_structure == kwargs["block_structure"]
+    assert block.block_structure == [2, 4]
 
 
 def test_infer_strategy():

tests/test_quantization/test_utils/test_helpers.py

@@ -20,7 +20,7 @@
     QuantizationArgs,
     QuantizationStrategy,
 )
-from compressed_tensors.quantization.utils import calculate_qparams, generate_gparam
+from compressed_tensors.quantization.utils import calculate_qparams, compute_dynamic_scales_and_zp, generate_gparam
 
 
 @pytest.mark.parametrize(
@@ -73,3 +73,25 @@ def test_fused_global_scales():
     assert max_tensor_value.item() == pytest.approx(
         FP4_E2M1_DATA.max * FP8_E4M3_DATA.max / global_scale, abs=0.001
     )
+
+@pytest.mark.parametrize(
+    "shape,group_size,exp_shape",
+    [
+        # Only batch size =1 is supported for dynamic GROUP quantization
+        ((1, 4, 8), 4, torch.Size([4, 2])),
+    ],
+)
+def test_compute_dynamic_scales_and_zp_group(shape, group_size, exp_shape):
+    """
+    Dynamic group quantization should reduce activations in groups, producing
+    scales and zero points of shape [batch, num_groups].
+    """
+    value = torch.randn(*shape)
+    args = QuantizationArgs(
+        strategy=QuantizationStrategy.GROUP,
+        group_size=group_size,
+        dynamic=True,
+    )
+    scale, zp = compute_dynamic_scales_and_zp(value, args, module=torch.nn.Module())
+    assert scale.shape == exp_shape
+    assert zp.shape == exp_shape