Order overlapping regex targets by specificity

src/compressed_tensors/utils/match.py

@@ -31,6 +31,14 @@
 FusedMappping = Mapping[str, Iterable[str]]
 
 
+def _target_priority_key(target: str) -> tuple:
+    """Sort key for match_targets: exact names, then longer regexes."""
+    if target.startswith("re:"):
+        pattern = target.removeprefix("re:")
+        return (1, -len(pattern), pattern)
+    return (0, 0, target)
+
+
 def match_named_modules(
     model: torch.nn.Module,
     targets: Iterable[str] | None,
@@ -135,10 +143,10 @@ def match_targets(
     # specific to least specific, and this order will be used when merging configs.
     # The entries are sorted in the following order:
     #     1. matches on exact strings
-    #     2. matches on regex patterns
+    #     2. matches on regex patterns (longer patterns before shorter ones)
     #     3. matches on module names (e.g. "Linear")
 
-    targets = sorted(targets, key=lambda x: ("re:" in x, x))
+    targets = sorted(targets, key=_target_priority_key)
     matched_targets = []
     for target in targets:
         if match_name(name, target):

tests/test_utils/test_match.py

@@ -18,6 +18,7 @@
     match_named_modules,
     match_named_parameters,
     match_quantizable_tensors,
+    match_targets,
 )
 from compressed_tensors.utils.match import _match_class
 from transformers import AutoModelForCausalLM
@@ -233,6 +234,35 @@ def test_fused_mapping(self):
         assert is_match("dummy.gate_up_proj", linear, "Linear", fused=mapping)
 
 
+class TestMatchTargets:
+    """Test cases for match_targets priority ordering"""
+
+    def test_regex_specificity_beats_lexicographic_order(self):
+        linear = nn.Linear(10, 20)
+        name = "transformer.layers.0.self_attn.q_proj"
+        matches = match_targets(name, linear, ["re:.*proj", "re:.*q_proj"])
+
+        assert matches[0] == "re:.*q_proj"
+        assert "re:.*proj" in matches
+
+    def test_exact_name_beats_regex(self):
+        linear = nn.Linear(10, 20)
+        name = "model.layers.0.mlp.down_proj"
+        matches = match_targets(
+            name, linear, ["re:.*down_proj", "model.layers.0.mlp.down_proj"]
+        )
+
+        assert matches[0] == "model.layers.0.mlp.down_proj"
+        assert "re:.*down_proj" in matches
+
+    def test_class_match_comes_after_name_matches(self):
+        linear = nn.Linear(10, 20)
+        matches = match_targets("layer1", linear, ["Linear", "layer1"])
+
+        assert matches[0] == "layer1"
+        assert matches[-1] == "Linear"
+
+
 class TestMatchNamedModules:
     """Test cases for match_named_modules function"""