[Kernel] Strengthen kernel shape coverage (#29636)

Co-authored-by: Khoa Pham <khoa.pham@radixark.ai>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-authored-by: Mohammad Miadh Angkad <176301910+mmangkad@users.noreply.github.com>

sgl-kernel/tests/test_per_token_quant_fp8.py

@@ -1,6 +1,6 @@
 import itertools
 import sys
-from typing import Optional, Tuple
+from typing import Tuple
 
 import pytest
 import torch
@@ -35,10 +35,16 @@ def sglang_per_token_quant_fp8(
     return output, scale
 
 
-@pytest.mark.parametrize(
-    "num_tokens,hidden_dim",
-    list(itertools.product([128, 256, 512], [512, 1076, 1368, 2048, 4096])),
-)
+PER_TOKEN_QUANT_CASES = list(
+    itertools.product([128, 256, 512], [512, 1076, 1368, 2048, 4096])
+) + [
+    (39, 1536),
+    (1392, 1536),
+    (7807, 1536),
+]
+
+
+@pytest.mark.parametrize("num_tokens,hidden_dim", PER_TOKEN_QUANT_CASES)
 def test_per_token_quant_compare_implementations(
     num_tokens: int,
     hidden_dim: int,