DeepSeek/DeepSeek-Math
1
0
Fork 0
mirror of https://github.com/deepseek-ai/DeepSeek-Math synced 2025-01-22 10:35:58 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Update summarize_results.py

Browse Source
This commit is contained in:
Zhihong Shao 2024-04-15 15:55:36 +08:00 committed by GitHub
parent 7c34ad4fa4
commit b8b0f8ce09
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 0 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
evaluation/summarize_results.py
View File

@ -87,16 +87,6 @@ def main():
os.makedirs(task_dirname, exist_ok=True)
metric_path = os.path.join(task_dirname, "metrics.json")
pred_path = os.path.join(task_dirname, "predictions.json")
if 'math6' in dataset.lower() and task == 'cot':
data_to_score = []
for pred in task2pred[task]:
item = deepcopy(pred['metadata'])
item['model_answer_turns_1'] = pred['turns'][0]['model_output']
item['model_answer_turns_2'] = pred['turns'][1]['model_output']
data_to_score.append(item)
_metrics = math6_score(data_to_score)
task2metric[task].update(_metrics)
model2dataset2task2metric[model][dataset][task].update(_metrics)
json.dump(task2metric[task], open(metric_path, "w"), indent=4)
json.dump(task2pred[task], open(pred_path, "w"), indent=4)
if 'minif2f' in dataset.lower() and 'isabelle' in dataset.lower() and task2pred[task] and args.eval_atp: