Require dtype argument and vectorizer dtype to match

Author: abrookins

redisvl/extensions/llmcache/semantic.py

@@ -86,12 +86,17 @@ def __init__(
         else:
             prefix = name
 
+        dtype = kwargs.get("dtype")
+
         # Validate a provided vectorizer or set the default
         if vectorizer:
             if not isinstance(vectorizer, BaseVectorizer):
                 raise TypeError("Must provide a valid redisvl.vectorizer class.")
+            if dtype and vectorizer.dtype != dtype:
+                raise ValueError(
+                    f"Provided dtype {dtype} does not match vectorizer dtype {vectorizer.dtype}"
+                )
         else:
-            dtype = kwargs.get("dtype")
             vectorizer_kwargs = {"dtype": dtype} if dtype else {}
 
             vectorizer = HFTextVectorizer(

tests/integration/test_llmcache.py

@@ -884,3 +884,26 @@ def test_bad_dtype_connecting_to_existing_cache(redis_url):
         bad_type = SemanticCache(
             name="float64_cache", dtype="float16", redis_url=redis_url
         )
+
+
+def test_vectorizer_dtype_mismatch():
+    with pytest.raises(ValueError):
+        SemanticCache(
+            name="test_cache",
+            dtype="float32",
+            vectorizer=HFTextVectorizer(dtype="float16"),
+        )
+
+
+def test_invalid_vectorizer():
+    with pytest.raises(TypeError):
+        SemanticCache(
+            name="test_cache",
+            vectorizer="invalid_vectorizer",  # type: ignore
+        )
+
+
+def test_passes_through_dtype_to_default_vectorizer():
+    # The default is float32, so we should see float64 if we pass it in.
+    cache = SemanticCache(name="test_cache", dtype="float64")
+    assert cache._vectorizer.dtype == "float64"