Update app.py

app.py

@@ -8,9 +8,9 @@ st.markdown("<h1 style='text-align: center; color: #00BFFF;'>Kaggle Notebooks in
 
 st.markdown("Here you can inspect Kaggle notebooks that were converted to python scripts and deduplicated.")
 @st.cache()
-def load_data(upvote=0):
+def load_data(upvote=0, size=0):
     ds = load_dataset("loubnabnl/subset_kaggle_scripts", split="train")
-    ds = ds.filter(lambda x: x["upvotes"] >= upvote)
+    ds = ds.filter(lambda x: x["upvotes"] >= upvote and x["seq_len"] >= size)
     return ds
 
 def show_extra_info(e):
@@ -32,7 +32,8 @@ def show_extra_info(e):
 
 
 vote = st.sidebar.slider("Minimum notebook ⬆️ upvotes", min_value=0, max_value=100, step=1, value=0)
-samples = load_data(vote)
+size = st.sidebar.slider("Length of the notebook in number of tokens", min_value=0, max_value=15_000, step=1000, value=0)
+samples = load_data(vote, size)
 index_example = st.sidebar.number_input(f"Choose a sample from the existing {len(samples)} notebooks:", min_value=0, max_value=max(0, len(samples)-1), value=0, step=1)
 
 st.markdown(show_extra_info(samples[index_example]), unsafe_allow_html=True)