review comments

Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>

vllm/model_executor/layers/quantization/utils/fp8_utils.py

@@ -81,7 +81,8 @@ def is_dim_blocked(dim, shape, group_shape):
     if is_dim_blocked(0, weight.shape, weight_group_shape[0])\
      and is_dim_blocked(1, weight.shape, weight_group_shape[1]) and\
      input_group_shape == (1, weight_group_shape[1]):
-        return apply_w8a8_block_fp8_linear(input, weight, weight_group_shape,
+        return apply_w8a8_block_fp8_linear(input, weight,
+                                           list(weight_group_shape),
                                            weight_scale)
     else:
         # Despite having linear in the it doesn't conform to
@@ -159,12 +160,18 @@ def block_quant_to_tensor_quant(
 
 
 # Quantize to fp8 assuming once scale per group of elements with shape
-# group_shape
+# group_shape, example group shapes:
+#  * (-1, -1)   for per-tensor quantization
+#  * (1, -1)    for per-row quantization
+#  * (-1, 1)    for per-column quantization
+#  * (128, 128) for 128x128 deepseek style block quantization
+#  * (1, 128)   for deepseek style activation quantization
+#               (i.e. per-token-per-group)
 def fp8_quantize(
     x: torch.Tensor,
     group_shape: Tuple[int, int],
     dtype: Optional[torch.dtype] = None,
-):
+) -> Tuple[torch.Tensor, torch.Tensor]:
     group_shape = _normalize_quant_group_shape(x, group_shape)
     if dtype is None:
         dtype = (torch.float8_e4m3fnuz