Raise error if group_size is passed but wrong strategy (#149)

* add error * Update src/compressed_tensors/quantization/quant_args.py Co-authored-by: Dipika Sikka <[email protected]> * remove extra line * fix invalid configs it tests --------- Co-authored-by: Dipika Sikka <[email protected]>
neuralmagic · Sep 9, 2024 · 4d89141 · 4d89141
1 parent b885229
commit 4d89141
Show file tree

Hide file tree

Showing 4 changed files with 19 additions and 3 deletions.
diff --git a/src/compressed_tensors/quantization/quant_args.py b/src/compressed_tensors/quantization/quant_args.py
@@ -187,6 +187,12 @@ def validate_model_after(model: "QuantizationArgs") -> Dict[str, Any]:
                     f"strategy {strategy} requires group_size to be "
                     "set to a positive value"
                 )
+        if (
+            group_size is not None
+            and group_size > 0
+            and strategy != QuantizationStrategy.GROUP
+        ):
+            raise ValueError("group_size requires strategy to be set to 'group'")
 
         # validate activation ordering and strategy
         if actorder is not None and strategy != QuantizationStrategy.GROUP:

diff --git a/tests/test_compressors/test_fp8_quant.py b/tests/test_compressors/test_fp8_quant.py
@@ -67,7 +67,7 @@ def make_dummy_g_idx(columns: int, group_size: int) -> torch.Tensor:
         ],
         [
             QuantizationStrategy.CHANNEL,
-            128,
+            None,
             torch.rand((512, 1)) * 0.01,
             torch.zeros((512, 1), dtype=torch.int8),
         ],

diff --git a/tests/test_compressors/test_int_quant.py b/tests/test_compressors/test_int_quant.py
@@ -57,7 +57,7 @@ def get_dummy_quant_config(strategy, group_size=None):
         [
             QuantizationStrategy.CHANNEL,
             False,
-            128,
+            None,
             torch.rand((512, 1)) * 0.01,
             ((torch.rand((512, 1)) - 0.5) * 127).to(torch.int8),
         ],
@@ -102,7 +102,7 @@ def test_quant_format(strategy, symmetric, group_size, sc, zp):
         ],
         [
             QuantizationStrategy.CHANNEL,
-            128,
+            None,
             torch.rand((300, 1)) * 0.01,
             torch.zeros((300, 1), dtype=torch.int8),
         ],

diff --git a/tests/test_quantization/test_quant_args.py b/tests/test_quantization/test_quant_args.py
@@ -43,6 +43,16 @@ def test_group():
     with pytest.raises(ValueError):
         QuantizationArgs(strategy=QuantizationStrategy.GROUP, group_size=-1)
 
+    args = QuantizationArgs(group_size=128, strategy="group")
+    assert args.group_size == 128
+    assert args.strategy == "group"
+
+    with pytest.raises(ValueError):
+        QuantizationArgs(strategy=QuantizationStrategy.GROUP)
+
+    with pytest.raises(ValueError):
+        QuantizationArgs(strategy="tensor", group_size=128)
+
 
 def test_block():
     kwargs = {"strategy": "block", "block_structure": "2x4"}