---
license: cc-by-sa-4.0
language:
- en
task_categories:
- text-generation
tags:
- text-to-sql
- agents
- reinforcement-learning
- tool-use
- schema-drift
- sqlite
- execution-verified
pretty_name: DriftSQL-Recovery
size_categories:
- 10K<n<100K
configs:
- config_name: tasks
  data_files:
  - split: train
    path: data/tasks/train.parquet
  - split: tune
    path: data/tasks/tune.parquet
- config_name: oracle_next_action
  data_files:
  - split: train
    path: data/oracle_next_action/train.parquet
  - split: tune
    path: data/oracle_next_action/tune.parquet
- config_name: oracle_trajectories
  data_files:
  - split: train
    path: data/oracle_trajectories/train.parquet
  - split: tune
    path: data/oracle_trajectories/tune.parquet
- config_name: rollout_index
  data_files:
  - split: train
    path: data/rollout_index/train.parquet
- config_name: failure_trajectories
  data_files:
  - split: train
    path: data/failure_trajectories/train.parquet
- config_name: recovery_sft
  data_files:
  - split: train
    path: data/recovery_sft/train.parquet
  - split: validation
    path: data/recovery_sft/validation.parquet
- config_name: hard_replay
  data_files:
  - split: train
    path: data/hard_replay/train.parquet
- config_name: sft_mix
  data_files:
  - split: train
    path: data/sft_mix/train.parquet
  - split: validation
    path: data/sft_mix/validation.parquet
- config_name: grpo
  data_files:
  - split: train
    path: data/grpo/train.parquet
  - split: tune
    path: data/grpo/tune.parquet
---

# DriftSQL-Recovery

**DriftSQL-Recovery** is an execution-verified dataset for training and
evaluating SQL agents under schema, business-knowledge, and interaction drift.
It contains database-isolated recovery tasks, canonical seven-tool
trajectories, real on-policy failures, Recovery SFT examples, hard replay, and
full-episode GRPO records.

The accompanying implementation, data factory, sandbox, reward, and evaluation
code are available in [DriftSQL](https://github.com/lixiao69sysu/DriftSQL).

## Release scope

| Configuration | Split | Rows | Description |
|---|---|---:|---|
| `tasks` | train / tune | 2,400 / 432 | Execution-verified clean and drift-recovery tasks |
| `oracle_next_action` | train / tune | 24,285 / 2,254 | Seven-tool next-action supervision |
| `oracle_trajectories` | train / tune | 2,400 / 432 | Canonical execution-verified episodes |
| `rollout_index` | train | 2,400 | Multi-seed on-policy rollout index and outcomes |
| `failure_trajectories` | train | 1,066 | Unique full on-policy failure trajectories |
| `recovery_sft` | train / validation | 1,317 / 397 | State-aware recovery examples around first errors |
| `hard_replay` | train | 1,600 | Balanced real recovery and canonical replay examples |
| `sft_mix` | train / validation | 3,314 / 2,254 | Maintained Recovery SFT + Hard Replay mixture |
| `grpo` | train / tune | 3,200 / 432 | Full-episode VERL/GRPO records |

The public tasks span `clean`, `add_column`, `rename_column`, `rename_table`,
`replace_column`, and `compound` scenarios. Interaction profiles cover
`direct_clean`, `schema_only`, `knowledge_only`, and `must_ask` behavior.

## Deliberately excluded

- **Fresh Blind320 is not published.** Its rows and answers remain sealed and
  were not read while preparing this release.
- **No raw SQLite databases are redistributed.** Database assets must be
  obtained from their upstream sources with the pinned revisions in the
  DriftSQL repository.
- Checkpoints, local filesystem paths, access tokens, and machine-specific
  runtime artifacts are not included.

These exclusions keep the final blind evaluation meaningful and avoid
duplicating roughly 52 GiB of upstream database assets.

## Loading

```python
from datasets import load_dataset

tasks = load_dataset("lxSYSU/DriftSQL-Recovery", "tasks")
failures = load_dataset("lxSYSU/DriftSQL-Recovery", "failure_trajectories")
sft = load_dataset("lxSYSU/DriftSQL-Recovery", "sft_mix")
grpo = load_dataset("lxSYSU/DriftSQL-Recovery", "grpo")
```

Paths that pointed to local upstream assets have been normalized to `hf://` or
`repo://` identifiers. They are provenance references, not bundled database
files.

## Tool protocol

The maintained environment exposes a dynamic subset of seven actions:

```text
get_schema_version
inspect_schema_diff
get_schema
ask_user
get_knowledge_definition
execute_sql
submit_solution
```

SQL targets and canonical trajectories were validated by executing original,
stale, and repaired SQL against isolated SQLite sessions. The runtime enforces
read-only authorization, timeouts, rollback, result-contract validation, and
dynamic action masks.

## Data construction

The task factory derives deterministic schema and interaction changes from
public BIRD-family and SIX-GYM SQLite assets. Splits are assigned by `db_id`,
not by random row:

- Train: 2,400 tasks over 60 databases;
- Tune: 432 tasks over 18 disjoint databases;
- Fresh Blind: 320 tasks over 20 additional databases, withheld from this
  public release.

For on-policy collection, 600 difficult Train tasks were sampled with multiple
seeds, producing 2,400 rollout outcomes and 1,066 unique failures. Failure
mining emphasizes wrong retrieval after schema diff, successful execution
without submission, must-ask mistakes, and compound recovery.

## Source data and attribution

This release derives from the following CC BY-SA 4.0 datasets:

- [BIRD23 Train Filtered](https://huggingface.co/datasets/birdsql/bird23-train-filtered)
- [SIX-GYM-SQLite](https://huggingface.co/datasets/birdsql/six-gym-sqlite)
- [BIRD-Critic-SQLite](https://huggingface.co/datasets/birdsql/bird-critic-1.0-sqlite)
- [Mini-Interact](https://huggingface.co/datasets/birdsql/mini-interact)

Their licenses and terms remain applicable. DriftSQL adds versioned drift
generation, verified tool trajectories, on-policy failure annotations,
training mixtures, and leakage controls. This dataset is therefore released
under **CC BY-SA 4.0** rather than a software license.

## Limitations

- The released execution environment targets SQLite.
- The tasks and tool traces are English; the DriftSQL application provides a
  separate Chinese-to-English input adapter.
- The 2,400-row rollout index contains metadata for every rollout, while full
  trajectories are published for the 1,066 mined failures.
- Tune is intended for checkpoint selection and must not be reported as a
  final blind result.
- SQL execution should always occur in an isolated, read-only sandbox.

## Reproducibility

Dataset revisions, generation scripts, split policies, and validation gates
are maintained in the
[DriftSQL repository](https://github.com/lixiao69sysu/DriftSQL). The release
metadata records that zero Fresh Blind rows and zero raw SQLite files are
included.