Corey Morris
commited on
Commit
·
41d7691
1
Parent(s):
e7c50af
Moved moral scenarios information higher on page
Browse files
app.py
CHANGED
|
@@ -332,13 +332,6 @@ st.plotly_chart(fig_radar_top_differences)
|
|
| 332 |
|
| 333 |
st.markdown("## Notable findings and plots")
|
| 334 |
|
| 335 |
-
st.markdown('### Abstract Algebra Performance')
|
| 336 |
-
st.write("Small models showed surprisingly strong performance on the abstract algebra task. A 6 Billion parameter model is tied for the best performance on this task and there are a number of other small models in the top 10.")
|
| 337 |
-
plot_top_n(filtered_data, 'MMLU_abstract_algebra', 10)
|
| 338 |
-
|
| 339 |
-
fig = create_plot(filtered_data, 'Parameters', 'MMLU_abstract_algebra')
|
| 340 |
-
st.plotly_chart(fig)
|
| 341 |
-
|
| 342 |
# Moral scenarios plots
|
| 343 |
st.markdown("### MMLU’s Moral Scenarios Benchmark Doesn’t Measure What You Think it Measures")
|
| 344 |
def show_random_moral_scenarios_question():
|
|
@@ -366,8 +359,12 @@ st.write()
|
|
| 366 |
fig = create_plot(filtered_data, 'MMLU_average', 'MMLU_moral_scenarios')
|
| 367 |
st.plotly_chart(fig)
|
| 368 |
|
|
|
|
|
|
|
|
|
|
| 369 |
|
| 370 |
-
|
|
|
|
| 371 |
|
| 372 |
st.markdown("***Thank you to hugging face for running the evaluations and supplying the data as well as the original authors of the evaluations.***")
|
| 373 |
|
|
|
|
| 332 |
|
| 333 |
st.markdown("## Notable findings and plots")
|
| 334 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 335 |
# Moral scenarios plots
|
| 336 |
st.markdown("### MMLU’s Moral Scenarios Benchmark Doesn’t Measure What You Think it Measures")
|
| 337 |
def show_random_moral_scenarios_question():
|
|
|
|
| 359 |
fig = create_plot(filtered_data, 'MMLU_average', 'MMLU_moral_scenarios')
|
| 360 |
st.plotly_chart(fig)
|
| 361 |
|
| 362 |
+
st.markdown('### Abstract Algebra Performance')
|
| 363 |
+
st.write("Small models showed surprisingly strong performance on the abstract algebra task. A 6 Billion parameter model is tied for the best performance on this task and there are a number of other small models in the top 10.")
|
| 364 |
+
plot_top_n(filtered_data, 'MMLU_abstract_algebra', 10)
|
| 365 |
|
| 366 |
+
fig = create_plot(filtered_data, 'Parameters', 'MMLU_abstract_algebra')
|
| 367 |
+
st.plotly_chart(fig)
|
| 368 |
|
| 369 |
st.markdown("***Thank you to hugging face for running the evaluations and supplying the data as well as the original authors of the evaluations.***")
|
| 370 |
|