How to enable the mysql DB with GPTcache?

[terryweijian]

Could you please introduce how to setup the GPTcache with mysql database? Or example of using mysql with GPTcache.

suppose there should be a config file to store the mysql connection information like host, port, user, password and etc.
or if can use the following method?
cache_base = CacheBase('mysql' host='127.0.0.1', port='3306', user='root', password='XXXX')

Thanks

[SimFG]

you can try the following code snippet:

from gptcache.embedding import Onnx
from gptcache.manager import manager_factory

onnx = Onnx()
data_manager = manager_factory(
    "mysql,faiss",
    scalar_params={
        "sql_url": "mysql+pymysql://root:[email protected]:3306/mysql",
        "table_name": "gptcache",
    },
    vector_params={"dimension": onnx.dimension},
)

more examples: https://github.com/zilliztech/GPTCache/blob/main/examples/data_manager/scalar_store.py

[SimFG]

yes, you are right about the database.

[Coditas-Vikas-Tiwari]

Hey I'm usin the same approach but it's not storing any data into my table.

[SimFG]

you can show me your demo code

[prakashpatidarcoindcx]

Hi I am using this approach here is my code and this is creating all required table starting with gptcache and using cache until process lifespan but the moment I restart the process it does not use cache for all question which is already stored in gptcache_answer.
tables and data is persisted in gptcache table but not sure why it is not able to use cached answer for question which is being asked before process start. here is code base:
self.onnx = Onnx()

    self.data_manager = manager_factory(
        "mysql,faiss",
        scalar_params={
            "sql_url": "mysql+pymysql://root:[email protected]:3306/mysql",
            "table_name": "gptcache",
        },
        vector_params={"dimension": self.onnx.dimension},
    )
    cache.init(
        embedding_func=self.onnx.to_embeddings,
        data_manager=self.data_manager,
        similarity_evaluation=SearchDistanceEvaluation(),
        config=Config(similarity_threshold=0.88,auto_flush=5),
        )
    cache.set_openai_key()

Any help to ensure that caching can be used across process lifespan will be really helpful.
Let me know if you need more context here.

[SimFG]

please give me the code about the cache usage in the llm. like

openai.ChatCompletion.create()