Use `config.embedding_dimensions` (if specified) in the RedisVectorStore instead of embedding dummy text.

[AlirezaRoshanzamir]

Thank you for your excellent library.

Making API calls or performing time-consuming tasks during module import or class initialization is a bad idea. You've already considered this by providing the embedding_dimensions argument and using a dummy embedding as a fallback method:

langchain-redis/libs/redis/langchain_redis/vectorstores.py

Lines 273 to 278 in 2bdf379

	if self.config.embedding_dimensions is None:
	self.config.embedding_dimensions = len(
	self._embeddings.embed_query(
	"The quick brown fox jumps over the lazy dog"
	)
	)

However, a few lines later, you didn't use the provided variable and embedded dummy text again:

langchain-redis/libs/redis/langchain_redis/vectorstores.py

Lines 326 to 335 in 2bdf379

	"attrs": {
	"dims": len(
	self._embeddings.embed_query(
	"The quick brown fox jumps over the lazy dog"
	)
	),
	"distance_metric": self.config.distance_metric,
	"algorithm": self.config.indexing_algorithm,
	"datatype": self.config.vector_datatype,
	},

[bsbodden]

Thanks for catching that! Will be addressed in the next release. Cheers.