use st.columns instead of st.beta_columns

This is not beta any more since streamlit 0.86.0

Author: dasmy

CHANGELOG.md

@@ -14,6 +14,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Changed
 - Fix internal TODOs to improve code quality
 - Fix partial reconstruction in plain numpy backend
+- Fix deprecation warning about st.beta_columns
 
 ## [0.1.1] - 2021-07-22
 ### Added

demos/demo_image.py

@@ -66,7 +66,7 @@ def st_visualize_results(V: np.ndarray,
     if verbose:
         st.caption('''The visualization below shows a **comparison between the input signal and its
         reconstruction** obtained through the learned factorization model.''')
-    cols = st.beta_columns(2)
+    cols = st.columns(2)
     for col, X, Xc, title in zip(cols, [V, R], [Vc, Rc], ['Input', 'Reconstruction']):
         with col:
             for x, xc in zip(X, cycle(Xc)):
@@ -100,7 +100,7 @@ def plot(X, title):
         plt.close(fig)
 
     for i_atom in range(n_atoms):
-        col1, col2, col3 = st.beta_columns((1, 3, 3))
+        col1, col2, col3 = st.columns((1, 3, 3))
         with col1:
             plot(np.moveaxis(nmf.W[i_atom], 0, -1), f'Atom {i_atom}')
         with col2:

requirements.txt

@@ -4,7 +4,7 @@ more-itertools==8.6.0
 numpy==1.19.4
 opt-einsum==3.3.0
 scipy==1.5.4
-streamlit==0.84.0
+streamlit==0.87.0
 torch==1.8.1
 
 # Development and Tests

tnmf/utils/demo.py

@@ -238,7 +238,7 @@ def st_compare_signals(cls, V: np.ndarray, R: np.ndarray, verbose: bool = True):
             ''')
 
         # show the input, its reconstruction, and the reconstruction error as images next to each other
-        cols = st.beta_columns(3)
+        cols = st.columns(3)
         for col, X, title in zip(cols, [V, R, V-R], ['Input', 'Reconstruction', 'Error']):
             with col:
                 fig = plt.figure()
@@ -298,7 +298,7 @@ def st_plot_partial_reconstructions(cls, V: np.ndarray, nmf: TransformInvariantN
         i_signal = st.slider('Signal number', 1, V.shape[0], key='i_signal_partial') - 1 if V.shape[0] > 1 else 0
         for i_atom in range(nmf.n_atoms):
             R_atom = nmf.R_partial(i_atom)
-            cols = st.beta_columns(2)
+            cols = st.columns(2)
             for col, signals, signal_opts, opts in zip(
                     cols,
                     [[nmf.W[i_atom]], [R_atom[i_signal], V[i_signal]]],
@@ -497,7 +497,7 @@ def generate_signal(cls, signal_params: dict) -> Tuple[np.ndarray, np.ndarray]:
 
     @classmethod
     def _st_compare_individual_signals(cls, V_i: np.ndarray, R_i: np.ndarray):
-        cols = st.beta_columns(2)
+        cols = st.columns(2)
         for col, X, title in zip(cols, [V_i, R_i], ['Input', 'Reconstruction']):
             with col:
                 cls.plot_signals([X], opts={'title': title})