optimizing-streamlit-performance

Speed up Streamlit apps by applying caching decorators like @st.cache_data and @st.cache_resource to prevent unnecessary recomputation. You can also isolate heavy UI components using @st.fragment to avoid full script reruns.

@st.cache_data is used for caching computation results like loaded dataframes, while @st.cache_resource caches global resources like database connections. Both support TTL and max_entries to control when cached values expire and limit memory usage.

Use @st.fragment to isolate heavy UI pieces so they execute independently without triggering a full Streamlit script rerun. This allows costlier tasks to run only when needed within that specific fragment boundary.

Handle large datasets in Streamlit by using sample-based loading and @st.cache_resource to manage memory efficiently. Combining these techniques with smart widget choices prevents UI freezing during frequent interactive updates.

Avoid caching in Streamlit dashboards when displaying real-time data that must update instantly on every interaction. If your TTL is too high or max_entries limits are restrictive, users will see stale data instead of live updates.

Yes, Streamlit caching works with forms by using st.form to batch widget inputs and prevent reruns until submission. Combining st.form with @st.cache_data minimizes redundant computations during frequent dashboard interactions.