Query engine for LLM-powered data processing using semantic operators
Top 32.2% on sourcepulse
LOTUS is a semantic query engine designed for efficient LLM-powered data processing, targeting data scientists and engineers who need to build complex reasoning pipelines over structured and unstructured data. It offers a declarative, Pandas-like API with semantic operators that leverage natural language expressions for data transformations, simplifying the creation of AI-driven analytics.
How It Works
LOTUS implements a semantic operator model, extending traditional relational operators with natural language predicates. This approach allows users to define data operations (like joins, filters, and aggregations) using high-level, human-readable expressions. The engine then optimizes and executes these operations using various AI-based algorithms, abstracting away the underlying LLM complexities and enabling flexible, composable AI pipelines.
Quick Start & Requirements
pip install lotus-ai
(stable) or pip install git+https://github.com/lotus-data/lotus.git@main
(latest).lotus.settings.configure(lm=lm)
.Highlighted Details
sem_join
, sem_filter
, sem_map
, sem_extract
, sem_agg
, sem_topk
, sem_sim_join
, and sem_search
.Maintenance & Community
5 days ago
1 day