[Question]: When searching vector databases, queries that have no parameters that will not meet the threshold requirements will be excluded.

[union-cmd]

Question Validation

• I have searched both the documentation and discord for an answer.

Question

def rag(self, memory, index, prompt, k=1):
chat_engine = index.as_chat_engine(
chat_mode="condense_plus_context",
memory=memory,
temperature=0.9,
similarity_top_ps=0.9,
similarity_top_k=k,
)
result = self.transform_str(chat_engine.chat(prompt))
return result

this is my code, I want queries that do not meet the threshold requirement to be excluded when searching the index.

[dosubot]

To ensure queries that do not meet a certain threshold requirement are excluded when searching the index, you can use metadata filters for advanced querying. This involves leveraging the SimpleVectorStore from the LlamaIndex framework, which allows for filtering based on metadata. Here's a concise guide on how to implement this:

1. Initialize the SimpleVectorStore and add your data to it. This store will be used to perform queries with filters.

2. Define your query embedding and the number of top results (similarity_top_k) you wish to retrieve.

3. Apply metadata filters to exclude results below a certain threshold. You can use MetadataFilters and ExactMatchFilter to specify these filters. For example, to exclude results below a similarity score of 0.5, you can use ExactMatchFilter(key="similarity_score", value=0.5).

4. Perform the query with the specified filters using VectorStoreQuery. This will automatically exclude queries that do not meet your specified criteria.

Here's a simplified example to illustrate the process:

from llama_index.core.vector_stores import SimpleVectorStore
from llama_index.core.vector_stores.types import (
    ExactMatchFilter,
    MetadataFilters,
    VectorStoreQuery,
)

simple_vector_store = SimpleVectorStore()
# Add your data to the store here

query_embedding = [1.0, 1.0]  # Replace with your actual query embedding
similarity_top_k = 10  # Number of top results to retrieve

filters = MetadataFilters(filters=[
    ExactMatchFilter(key="your_filter_key", value="your_filter_value")
])

query = VectorStoreQuery(
    query_embedding=query_embedding,
    filters=filters,
    similarity_top_k=similarity_top_k
)
result = simple_vector_store.query(query)

print(result.ids)

This approach ensures that only the queries meeting your specified threshold or criteria are included in the search results, effectively filtering out undesired queries.

Details

Sources

• llama-index-core/llama_index/core/indices/vector_store/retrievers/retriever.py
• llama-index-legacy/llama_index/legacy/indices/query/embedding_utils.py
• llama-index-integrations/tools/llama-index-tools-vector-db/llama_index/tools/vector_db/base.py
• llama-index-core/tests/vector_stores/test_simple.py
• llama-index-legacy/llama_index/legacy/vector_stores/simple.py
• llama-index-core/llama_index/core/indices/base.py
• llama-index-legacy/llama_index/legacy/indices/vector_store/retrievers/retriever.py
• llama-index-legacy/llama_index/legacy/indices/query/query_transform/base.py
• llama-index-legacy/tests/vector_stores/test_metadata_filters.py
• llama-index-core/llama_index/core/vector_stores/types.py

About Dosu
This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

To continue the conversation, mention @dosu.