Support for setting top_k for HybridRouteLayer?

[andreped]

I see that top_k=5 is hardcoded for the HybridRouteLayer class:

semantic-router/semantic_router/hybrid_layer.py

Line 51 in 1e753d9

results = self._query(text)

Which is possible to set through the _query method:

semantic-router/semantic_router/hybrid_layer.py

Line 124 in 1e753d9

def _query(self, text: str, top_k: int = 5):

I believe being able to set top_k would be of value to achieve better performance in some use cases.

[andreped]

The feature has been merged but has yet to be part of a release. For now, install from source to get feature.

Below is an example on how to set top_k:

from semantic_router.encoders import AzureOpenAIEncoder, TfidfEncoder
from semantic_router.hybrid_layer import HybridRouteLayer

model = HybridRouteLayer(
    encoder=AzureOpenAIEncoder(...),
    sparse_encoder=TfidfEncoder(),
    routes=routes,
    alpha=0.3,  # default = 0.3
    top_k=3,  # default = 5
)