all 9 languages

README.md

@@ -8,20 +8,23 @@ task_categories:
 
 # LeetCode multilingual benchmark dataset
 
-Post-interface problem rows for the msl-multilingual-self-learning benchmark. Each
-JSON line is one LeetCode problem with the 9-language callable `interfaces`
-(python, cpp, go, java, rust, javascript, typescript, php, ruby), `canonical_tests`
-(a Python check(candidate) oracle), `problem_description`, and `metadata`.
+Post-interface rows for the msl-multilingual-self-learning benchmark, flattened to
+one row per (problem, language). Each row has the single `interface` for its
+`language` (python, cpp, go, java, rust, javascript, typescript, php, ruby), plus
+the shared `canonical_tests` (a Python check(candidate) oracle), `problem_description`,
+and `metadata`. Stored as `data/<split>/<lang>.jsonl` (one JSONL per language, so
+each file stays well under HF's 10 MiB limit); HF globs them all into a single split.
 
 ## Splits
 
-- `train`: 2878 problems (all-9-language universe minus the newfacade test split).
-- `test`: 201 problems (held-out eval).
+- `train`: 2878 problems x 9 languages.
+- `test`: 201 problems x 9 languages = 1809 rows (held-out eval).
 
 ## Load
 
 ```python
 from datasets import load_dataset
-train = load_dataset("neulab/leetcode", split="train")
-test = load_dataset("neulab/leetcode", split="test")
+train = load_dataset("neulab/leetcode", split="train")  # one Dataset, all rows
+test  = load_dataset("neulab/leetcode", split="test")   # one Dataset, 1809 rows
+python_rows = test.filter(lambda r: r["language"] == "python")
 ```