Similar to these un-answered and closed topics:
I am seeing the same Error on the python client when trying to reproduce cosine similarity script search on text-embeddings as explained in this blogpost:
Query (where query_vector
is a list of floats of length 512):
query = {
"script_score": {
"query": {"match_all": {}},
"script": {
"source": "cosineSimilarity(params.query_vector, doc['my_dense_vector']) + 1.0",
"params": {"query_vector": query_vector}
}
}}
Request:
res = es.search(index=INDEX_NAME, body={"size": QUERY_SIZE,"query": query})
Error:
RequestError: RequestError(400, 'search_phase_execution_exception', 'runtime error')