We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
1 parent 638c20f commit f27352fCopy full SHA for f27352f
src/lemonade/tools/mmlu.py
@@ -210,6 +210,14 @@ def run(
210
state.save_stat(stat_units_name, "%")
211
self.status_stats.append(stat_name)
212
213
+ # Calculate average of mmlu accuracy
214
+ # And display in the CLI
215
+ acc_avg = np.mean([accuracy_data["Accuracy"] for accuracy_data in summary_data])
216
+ avg_stat_name = "avg_accuracy"
217
+ state.save_stat(avg_stat_name, float(acc_avg) * 100)
218
+ state.save_stat("accuracy_units", "%")
219
+ self.status_stats.append(avg_stat_name)
220
+
221
# Save accuracy results to CSV file
222
summary_df = pd.DataFrame(summary_data)
223
summary_df.to_csv(
0 commit comments