DatabaseError: (databricks.sql.exc.ServerOperationError) [UNBOUND_SQL_PARAMETER]

[nag_kanchan]

Hi,

I am trying to connect my database through LLM and expecting to receive a description of the table and 1st 3 rows from the table. 

from langchain.agents import create_sql_agent
from langchain.agents.agent_toolkits import SQLDatabaseToolkit
from langchain.sql_database import SQLDatabase
from langchain import OpenAI
from langchain.agents.agent_types import AgentType

db = SQLDatabase.from_databricks(catalog="hive_metastore", schema="my_data", host="*****.gcp.databricks.com",api_token="dapi*****")
llm = OpenAI(temperature=0.9)
toolkit = SQLDatabaseToolkit(db=db, llm=llm)
agent = create_sql_agent(llm=llm, toolkit=toolkit, verbose=True)

agent.run("describe silver_tableabc")

The error that I am facing:

DatabaseError: (databricks.sql.exc.ServerOperationError) [UNBOUND_SQL_PARAMETER] Found the unbound parameter: param_1. Please, fix `args` and provide a mapping of the parameter to either a SQL literal or collection constructor functions such as `map()`, `array()`, `struct()`. SQLSTATE: 42P02; line 3 pos 7
[SQL: SELECT silver_tableabc.health,silver_tableabc.cust_id,... FROM silver_tableabc
 LIMIT :param_1]
[parameters: {'param_1': 3}]
(Background on this error at: https://sqlalche.me/e/20/4xp6)

Version:

• Cluster: 14.2ML ( even tried with general purpose 14.2, 13.3)
• Langchain: 0.0.348 (even tried with 0.0.341)
• SQLAlchemy: 2.0.12
• databricks-sql-connector: 3.0.0

  @Retired_mod COuld you please help me how can I resolve this?

[nag_kanchan]

I am getting the same error when executing this notebook: https://docs.databricks.com/en/_extras/notebooks/source/machine-learning/large-language-models/sql-d... 

[MuraliPinninti_]

Hi team,

We tried similar the approach as above with following versions of compatability DBR 14.2

langchain==0.0.348 databricks-vectorsearch==0.22 openai==0.27 langchain-experimental databricks-sql-connector==3.0.0

In the run method there is no way to pass any positional parameters like {'param_1':3}.

Can you please help us how can i modify my below statement using set:param_1 to 3 statement to make it work?

agent.run("what the job title of contact kevin young")

Thanks,

Murali

[MuraliPinninti_ · Accepted solution]

It worked after downgrading databricks-sql-connector to 2.9.3. 

For the same question format:  agent.run("what the job title of contact kevin young")

Thanks,

Murali

[nag_kanchan]

This is not databricks issue but from langchain. A PR has been raised to solve this: 

One workaround that worked is: https://github.com/langchain-ai/langchain/issues/11068 setting sample_rows_in_table_info to 0 when calling SQLDatabase.from_databricks() .