Spaces:
Runtime error
Runtime error
loading
Browse files
app.py
CHANGED
|
@@ -80,7 +80,7 @@ def data_comparison(df):
|
|
| 80 |
).interactive()
|
| 81 |
|
| 82 |
legend = alt.Chart(df).mark_point().encode(
|
| 83 |
-
y=alt.Y('cluster:N', axis=alt.Axis(orient='
|
| 84 |
x=alt.X("label"),
|
| 85 |
shape=alt.Shape('label', scale=alt.Scale(
|
| 86 |
range=['circle', 'diamond']), legend=None),
|
|
@@ -89,7 +89,7 @@ def data_comparison(df):
|
|
| 89 |
selection
|
| 90 |
)
|
| 91 |
|
| 92 |
-
layered =
|
| 93 |
|
| 94 |
layered = layered.configure_axis(
|
| 95 |
grid=False
|
|
@@ -99,7 +99,7 @@ def data_comparison(df):
|
|
| 99 |
|
| 100 |
return layered
|
| 101 |
|
| 102 |
-
|
| 103 |
def quant_panel(embedding_df):
|
| 104 |
""" Quantitative Panel Layout"""
|
| 105 |
|
|
@@ -112,6 +112,7 @@ def quant_panel(embedding_df):
|
|
| 112 |
st.markdown("* The **shape** of each point reflects the label category -- positive (diamond) or negative sentiment (circle).")
|
| 113 |
st.altair_chart(data_comparison(down_samp(embedding_df)))
|
| 114 |
|
|
|
|
| 115 |
def frequent_tokens(data, tokenizer, loss_quantile=0.95, top_k=200, smoothing=0.005):
|
| 116 |
unique_tokens = []
|
| 117 |
tokens = []
|
|
@@ -171,6 +172,7 @@ def clustering(data,num_clusters):
|
|
| 171 |
|
| 172 |
return data, assigned_clusters
|
| 173 |
|
|
|
|
| 174 |
def kmeans(df, num_clusters=3):
|
| 175 |
data_hl = df.loc[df['slice'] == 'high-loss']
|
| 176 |
data_kmeans,clusters = clustering(data_hl,num_clusters)
|
|
@@ -276,10 +278,11 @@ if __name__ == "__main__":
|
|
| 276 |
st.write(dataframe,width=900, height=300)
|
| 277 |
|
| 278 |
with rcol:
|
| 279 |
-
st.
|
| 280 |
-
|
| 281 |
-
|
| 282 |
-
st.
|
| 283 |
-
|
| 284 |
-
|
| 285 |
-
|
|
|
|
|
|
| 80 |
).interactive()
|
| 81 |
|
| 82 |
legend = alt.Chart(df).mark_point().encode(
|
| 83 |
+
y=alt.Y('cluster:N', axis=alt.Axis(orient='left'), title=""),
|
| 84 |
x=alt.X("label"),
|
| 85 |
shape=alt.Shape('label', scale=alt.Scale(
|
| 86 |
range=['circle', 'diamond']), legend=None),
|
|
|
|
| 89 |
selection
|
| 90 |
)
|
| 91 |
|
| 92 |
+
layered = legend | scatter
|
| 93 |
|
| 94 |
layered = layered.configure_axis(
|
| 95 |
grid=False
|
|
|
|
| 99 |
|
| 100 |
return layered
|
| 101 |
|
| 102 |
+
@st.cache(ttl=600)
|
| 103 |
def quant_panel(embedding_df):
|
| 104 |
""" Quantitative Panel Layout"""
|
| 105 |
|
|
|
|
| 112 |
st.markdown("* The **shape** of each point reflects the label category -- positive (diamond) or negative sentiment (circle).")
|
| 113 |
st.altair_chart(data_comparison(down_samp(embedding_df)))
|
| 114 |
|
| 115 |
+
@st.cache(ttl=600)
|
| 116 |
def frequent_tokens(data, tokenizer, loss_quantile=0.95, top_k=200, smoothing=0.005):
|
| 117 |
unique_tokens = []
|
| 118 |
tokens = []
|
|
|
|
| 172 |
|
| 173 |
return data, assigned_clusters
|
| 174 |
|
| 175 |
+
@st.cache(ttl=600)
|
| 176 |
def kmeans(df, num_clusters=3):
|
| 177 |
data_hl = df.loc[df['slice'] == 'high-loss']
|
| 178 |
data_kmeans,clusters = clustering(data_hl,num_clusters)
|
|
|
|
| 278 |
st.write(dataframe,width=900, height=300)
|
| 279 |
|
| 280 |
with rcol:
|
| 281 |
+
with st.spinner(text='loading...'):
|
| 282 |
+
st.markdown('<h3>Word Distribution in Error Slice</h3>', unsafe_allow_html=True)
|
| 283 |
+
commontokens = frequent_tokens(merged, tokenizer, loss_quantile=loss_quantile)
|
| 284 |
+
with st.expander("How to read the table:"):
|
| 285 |
+
st.markdown("* The table displays the most frequent tokens in error slices, relative to their frequencies in the val set.")
|
| 286 |
+
st.write(commontokens)
|
| 287 |
+
with st.spinner(text='visualizing...'):
|
| 288 |
+
quant_panel(merged)
|
error_analysis/utils/__pycache__/style_hacks.cpython-39.pyc
CHANGED
|
Binary files a/error_analysis/utils/__pycache__/style_hacks.cpython-39.pyc and b/error_analysis/utils/__pycache__/style_hacks.cpython-39.pyc differ
|
|
|