From 68db64b9ea6a67e3990d030947dd471ef79d02c6 Mon Sep 17 00:00:00 2001
From: mpc <matcoo@ceh.ac.uk>
Date: Fri, 20 Dec 2024 14:12:31 +0000
Subject: [PATCH] outputs evaluation results for all synthetic questions

---
 data/.gitignore     |  1 +
 dvc.yaml            |  3 ++-
 params.yaml         |  1 +
 scripts/evaluate.py | 12 +++++++++---
 4 files changed, 13 insertions(+), 4 deletions(-)

diff --git a/data/.gitignore b/data/.gitignore
index b8dc417..439d928 100644
--- a/data/.gitignore
+++ b/data/.gitignore
@@ -17,3 +17,4 @@
 /eidc_rag_test_set.csv
 /rag-pipeline.yml
 /pipeline.yml
+/results.csv
diff --git a/dvc.yaml b/dvc.yaml
index 179c315..31f1704 100644
--- a/dvc.yaml
+++ b/dvc.yaml
@@ -79,10 +79,11 @@ stages:
     - ${files.eval-set}
     - ${files.pipeline}
   evaluate:
-    cmd: uv run scripts/evaluate.py ${files.eval-set} -m ${files.metrics} -img ${files.eval-plot}
+    cmd: uv run scripts/evaluate.py ${files.eval-set} -m ${files.metrics} -img ${files.eval-plot} -r ${files.results}
     deps:
     - ${files.eval-set}
     - scripts/evaluate.py
     outs:
     - ${files.metrics}
     - ${files.eval-plot}
+    - ${files.results}
diff --git a/params.yaml b/params.yaml
index b3dc785..a387f96 100644
--- a/params.yaml
+++ b/params.yaml
@@ -15,6 +15,7 @@ files:
   test-set: data/eidc_rag_testset.csv
   eval-set: data/evaluation_data.csv
   metrics: data/metrics.json
+  results: data/results.csv
   eval-plot: data/eval.png
   pipeline: data/pipeline.yml
 sub-sample: 0 # sample n datasets for testing (0 will use all datasets)
diff --git a/scripts/evaluate.py b/scripts/evaluate.py
index fbe348f..b4ea981 100644
--- a/scripts/evaluate.py
+++ b/scripts/evaluate.py
@@ -22,7 +22,7 @@
 from ragas.run_config import RunConfig
 
 
-def main(eval_dataset: str, metric_output: str, image_output: str) -> None:
+def main(eval_dataset: str, metric_output: str, image_output: str, results_output: str) -> None:
     nest_asyncio.apply()  # apply the event loop async fix
     df = pd.read_csv(eval_dataset, converters={"contexts": pd.eval})
     eval_dataset = Dataset.from_pandas(df)
@@ -45,7 +45,7 @@ def main(eval_dataset: str, metric_output: str, image_output: str) -> None:
         run_config=RunConfig(max_workers=1),
     )
     result_df = result.to_pandas()
-
+    result_df.to_csv(results_output, index=False)
     Path(metric_output).parent.mkdir(parents=True, exist_ok=True)
     with open(metric_output, "w+") as f:
         json.dump(result, f)
@@ -88,5 +88,11 @@ def main(eval_dataset: str, metric_output: str, image_output: str) -> None:
         help="File to save image plot to.",
         default="data/evaluation.png",
     )
+    parser.add_argument(
+        "-r",
+        "--results",
+        help="File to save evaluation results",
+        default="data/results.csv",
+    )
     args = parser.parse_args()
-    main(args.eval_dataset, args.metrics_output, args.image_output)
+    main(args.eval_dataset, args.metrics_output, args.image_output, args.results)