search

vearch / vearch

Distributed vector search for AI-native applications
https://vearch.github.io
Apache License 2.0
2.03k stars 325 forks source link

1w左右文档量，召回率50%左右，配置如下 #508

Open codingwangfeng opened 3 years ago

codingwangfeng commented 3 years ago 
{
    "name":"xx",
    "partition_num":1,
    "replica_num":0,
    "properties":{
        "vid":{
            "type":"keyword"
        },
        "vector":{
            "dimension":128,
            "type":"vector"
        }
    },
    "engine":{
        "name":"gamma",
        "index_size": 8192,
        "max_size":10000,
        "retrieval_type":"IVFPQ",
        "retrieval_param":{
            "metric_type":"L2",
            "ncentroids":256,
            "nprobe":25,
            "nsubvector":32
        },
        "id_type":"String"
    }
}
gyd-a commented 3 years ago

数据量较小，建议你使用flat或者ivfflat模型。

codingwangfeng commented 3 years ago

数据量较小，建议你使用flat或者ivfflat模型。

有space配置例子能给一个吗？我在官网没找到 3.2.0