LLM

LLM-powered helpers for enhancing predictive modeling work

Enrich

Enrich structured datasets with completions from LLMs, chains, or tools. PROTOTYPE

df['is_seasonal'] = df.apply(enrich('Is {product} demand highly seasonal?', llm=llm), axis=1)
df_with_embeddings = df.join(df.apply(embed("Product: {product} - {description}"), axis=1))

Data Dictionary Generation

Infer a data dictionary using a DataRobot project PROTOTYPE

chain = DataDictChain()
data_dict = chain(context='Telco churn prediction', project_id='643ee357653d09185b081823')

Guard

Monitor and guard the usage of an LLM based on your defined criteria using DataRobot MLOps PROTOTYPE

@aguard(monitor_config, guardrail_config)
async def llm_proxy(prompt):
    completion = call_llm(prompt)
    return completion