Define a cached function:
@st.cache_data
decorator marks the function foo() for caching.
Inside `foo()
perform some expensive computation and return data.
Execute the cached function:
d1 = foo(ref1)
executes foo() with the argument ref1 and caches the result in d1.
d2 = foo(ref1)
retrieves the cached result for the same argument, so foo() is not executed again, and d1 == d2.
d3 = foo(ref2)
calls foo() with a different argument ref2, so the function executes again and caches the result in d3.
Clearing cache entries:
foo.clear()
clears all cached entries for the function foo().
st.cache_data.clear()
clears values from all cached functions, either in-memory or on-disk.
These snippets demonstrate how Streamlit allows you to cache expensive computations or data retrievals to improve performance.
E.g. TensorFlow session, database connection, etc.
@st.cache_resource
def foo(bar):
Create and return a non-data object return session #Executes foo
s1 = foo(ref1)
#Does not execute foo #Returns cached item by reference, s1 == s2
s2 = foo(ref1)
#Different arg, so function foo executes
s3 = foo(ref2)
#Clear all cached entries for this function
foo.clear()
#Clear all global resources from cache
st.cache_resource.clear()
These Streamlit snippets showcase caching with the @st.cache decorator:
Define a cached function:
`@st.cache
decorator marks the function foo() for caching.
Inside `foo()
perform some expensive computation and return data.
Execute the cached function:
`d1 = foo(ref1)` executes `foo()` with the argument `ref1` and caches the result in `d1`.
`d2 = foo(ref1)` retrieves the cached result for the same argument, so `foo()` is not executed again, and `d1 == d2`.
`d3 = foo(ref2)` calls `foo()` with a different argument `ref2`, so the function executes again.
These snippets demonstrate how Streamlit allows you to cache function results to avoid redundant computation and improve performance.