---
license: apache-2.0
task_categories: [text-generation]
tags: [code, zevo, validation]
configs:
- config_name: mbppplus
  data_files:
  - split: validation
    path: mbppplus/validation-*.parquet
- config_name: codeforces
  data_files:
  - split: validation
    path: codeforces/validation-*.parquet
- config_name: cc_valid
  data_files:
  - split: validation
    path: cc_valid/validation-*.parquet
---

# zero-code L1 independent validation sets

Three **separate** configs. Not mixed into one split. Task test stays `Tommysha/zero-code-l1-lcb-v5-test` (LCB v5 279).

| config | split | rows | source | Zevo independent val |
|---|---|---:|---|---|
| `mbppplus` | validation | 378 | `evalplus/mbppplus` test, as-is plus `id`/`entry_point` | yes (≥200) |
| `codeforces` | validation | 408 | `agentica-org/DeepCoder-Preview-Dataset` `codeforces` test, plus `id` | yes (≥200) |
| `cc_valid` | validation | 117 | `Tommysha/zero-code-l1-cc-valid` / `deepmind/code_contests` valid | Hub copy only. Zevo rejects independent val < 200 |

Gold solutions are not in `mbppplus` (`code` dropped). Each config has its own `evaluator.py` and `sample_submission.csv`.