fix kv cache apply

kylesayrs · kylesayrs · commit 2aead0e5e038 · 2025-10-09T16:16:10.000-04:00
Signed-off-by: Kyle Sayers &lt;kylesayrs@gmail.com&gt;
diff --git a/src/compressed_tensors/quantization/lifecycle/apply.py b/src/compressed_tensors/quantization/lifecycle/apply.py
@@ -137,7 +137,7 @@ def apply_quantization_config(
     # because attention quantization is a superset of kv cache quantization
     if config.kv_cache_scheme is not None:
         scheme = QuantizationScheme(
-            targets=".*self_attn$", input_activations=config.kv_cache_scheme
+            targets=[".*self_attn$"], input_activations=config.kv_cache_scheme
         )
         for submodule in model.modules():
             if is_attention_module(submodule):

Original file line number	Diff line number	Diff line change
`@@ -137,7 +137,7 @@ def apply_quantization_config(`
`137`	`137`	`# because attention quantization is a superset of kv cache quantization`
`138`	`138`	`if config.kv_cache_scheme is not None:`
`139`	`139`	`scheme = QuantizationScheme(`
`140`		`- targets=".*self_attn$", input_activations=config.kv_cache_scheme`
	`140`	`+ targets=[".*self_attn$"], input_activations=config.kv_cache_scheme`
`141`	`141`	`)`
`142`	`142`	`for submodule in model.modules():`
`143`	`143`	`if is_attention_module(submodule):`