intro

benchmark/benchmark.py

@@ -42,8 +42,6 @@ def show_stats(dirnames):
         row = summarize_results(dirname)
         raw_rows.append(row)
 
-    return
-
     repeats = []
     seen = dict()
     rows = []
@@ -71,10 +69,6 @@ def show_stats(dirnames):
         seen[kind] = row.dir_name
         rows.append(vars(row))
 
-    repeats = pd.DataFrame.from_records(repeats)
-    repeat_max = repeats["pass_rate_2"].max()
-    repeat_min = repeats["pass_rate_2"].min()
-
     df = pd.DataFrame.from_records(rows)
     df.sort_values(by=["model", "edit_format"], inplace=True)
 
@@ -135,6 +129,11 @@ def show_stats(dirnames):
             if zorder == 2:
                 ax.bar_label(rects, padding=8, labels=[f"{v:.0f}%" for v in df[fmt]], size=11)
 
+    if repeats:
+        repeats = pd.DataFrame.from_records(repeats)
+        repeat_max = repeats["pass_rate_2"].max()
+        repeat_min = repeats["pass_rate_2"].min()
+
         lo = 44 - repeat_min
         hi = repeat_max - 44
         ax.errorbar(1.4, 44, yerr=[[lo], [hi]], fmt="none", zorder=5, capsize=5)
@@ -428,10 +427,7 @@ def run_test(
             shutil.copy(original_fname, fname)
 
     file_list = " ".join(fname.name for fname in fnames)
-    intro = testdir / ".docs/introduction.md"
-    if intro.exists():
-        instructions = intro.read_text() + "\n\n"
-    else:
+
     instructions = ""
 
     introduction = testdir / ".docs/introduction.md"

docs/benchmarks.md

@@ -74,7 +74,7 @@ their coding skills.
 
 Each exercise has:
 
-  - Some instructions about the exercise, in a markdown file.
+  - Some instructions for the exercise, in markdown files.
   - Stub code for the implementation in a python file, specifying the functions/classes that need to be implemented.
   - Unit tests in a seperate python file.