[Bugfix] Remove assertion for NVFP4 scale dynamic range (#37465)

Signed-off-by: Michael Goin <mgoin64@gmail.com>
2026-03-18 23:37:49 +01:00
parent 5bc1da147f
commit 9482b0b085
1 changed files with 0 additions and 3 deletions
--- a/vllm/model_executor/layers/quantization/utils/marlin_utils_fp4.py
+++ b/vllm/model_executor/layers/quantization/utils/marlin_utils_fp4.py
@@ -37,9 +37,6 @@ def _nvfp4_compute_scale_factor(marlin_scales: torch.Tensor) -> float:
        min_val = ws_float[nonzero_mask].min()
        if min_val < 2:
            sf = (2 / min_val).log2().ceil().exp2()
-            assert (ws_float[nonzero_mask] * sf <= 448 * (2**7)).all(), (
-                "NVFP4 scale dynamic range too large for rescaling"
-            )
            return sf.item()
    return 1.0