From b2f7f733d934ad31870b6ec3041c52607cc5cec9 Mon Sep 17 00:00:00 2001 From: Rita Aleksziev Date: Wed, 22 Jan 2025 10:58:44 +0100 Subject: [PATCH] create output dir if doesnt exist --- evals/eval_on_hotpot.py | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/evals/eval_on_hotpot.py b/evals/eval_on_hotpot.py index 29c61f97e..b0591a1eb 100644 --- a/evals/eval_on_hotpot.py +++ b/evals/eval_on_hotpot.py @@ -109,10 +109,15 @@ async def eval_on_QA_dataset( dataset = load_qa_dataset(dataset_name_or_filename) context_provider = qa_context_providers[context_provider_name] eval_metrics = get_metrics(metric_name_list) + + out_path = Path(out_path) + if not out_path.exists(): + out_path.mkdir(parents=True, exist_ok=True) + random.seed(42) instances = dataset if not num_samples else random.sample(dataset, num_samples) - contexts_filename = Path(out_path) / Path( + contexts_filename = out_path / Path( f"contexts_{dataset_name_or_filename.split('.')[0]}_{context_provider_name}.json" ) if "promptfoo_metrics" in eval_metrics: @@ -122,7 +127,7 @@ async def eval_on_QA_dataset( else: promptfoo_results = {} - answers_filename = Path(out_path) / Path( + answers_filename = out_path / Path( f"answers_{dataset_name_or_filename.split('.')[0]}_{context_provider_name}.json" ) deepeval_results = await deepeval_on_instances(