codeflash-admin/codeflash-internal
1
0
Fork 0
mirror of https://github.com/codeflash-ai/codeflash-internal.git synced 2026-05-04 18:25:18 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

current model claude

Browse source
This commit is contained in:
Aseem Saxena 2025-04-16 18:02:24 -07:00
parent 6e70818535
commit c9dade08b0
1 changed files with 9 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

13
experiments/metrics_analysis.py
View file

@ -273,10 +273,12 @@ def augment_with_best_correct_speedup_ratio(df: DataFrame) -> DataFrame:
def main() -> None:
df = load_data("cotprompt0424")
df = process_column_pairs(df, "metadata")
df = process_column_pairs(df, "test_framework")
df = process_column_pairs(df, "generated_test")
df = load_data("test_claude_aseem_sanity1")
non_orphan_ids = remove_orphans(df)
df = df.iloc[non_orphan_ids]
#df = process_column_pairs(df, "metadata")
#df = process_column_pairs(df, "test_framework")
#df = process_column_pairs(df, "generated_test")
df = augment_with_best_correct_speedup_ratio(df)
exp0_df = df[df["trace_id"].str.endswith("EXP0")]
exp1_df = df[df["trace_id"].str.endswith("EXP1")]
@ -304,6 +306,9 @@ def main() -> None:
# Output the combined metrics DataFrame
print(metrics_df)
print(paired_performance_metrics)
print(paired_validity_metrics)
print(paired_coverage_metrics)
if __name__ == "__main__":
main()