# BridgeData V2 (/docs/evaluation/benchmark-hub/bridgedata-v2)



## About [#about]

BridgeData V2 external benchmark task manifest for real-robot dataset and offline policy result normalization.

## Metrics [#metrics]

BridgeData V2 external benchmark task manifest for real-robot dataset and offline policy result normalization.

### Task YAML [#task-yaml]

`worldfoundry/data/benchmarks/tasks/external/bridgedata-v2.yaml`

### Primary [#primary]

* `success_rate`

| Metric            | Focus                                                                                          |
| ----------------- | ---------------------------------------------------------------------------------------------- |
| `success_rate`    | Share of episodes or tasks that complete successfully.                                         |
| `task_success`    | Share of episodes or tasks that complete successfully.                                         |
| `goal_success`    | Share of episodes or tasks that complete successfully.                                         |
| `action_accuracy` | Correctness of predicted actions, states, or outcomes against the benchmark success criterion. |
| `rollout_success` | Share of episodes or tasks that complete successfully.                                         |

## Environment setup [#environment-setup]

### Runtime environment [#runtime-environment]

* **Conda env:** `worldplay-core`.
* **Separate setup:** defaults to the unified WorldFoundry env (`needs_new_env: false`); a benchmark-only conda env is usually not required.
* BridgeData V2 需要已物化的评测 split 和 policy-result import；当前只配置通用 Python 环境。

### Evaluation data [#evaluation-data]

* Candidate model outputs: set `WORLDFOUNDRY_BRIDGEDATA_V2_RESULTS_PATH` to the generated-video or rollout artifact root.
* `${WORLDFOUNDRY_DATA_DIR}/datasets/bridgedata-v2`
* `${WORLDFOUNDRY_CACHE_DIR}/repos/rail-berkeley--bridge_data_v2`

### Checkpoints & assets [#checkpoints--assets]

* This benchmark is primarily rollout- or official-result-driven; see catalog `requires` and the task YAML for policy/simulator checkpoints.

### Key environment variables [#key-environment-variables]

* `WORLDFOUNDRY_BRIDGEDATA_V2_DATASET_ROOT`
* `WORLDFOUNDRY_BRIDGEDATA_V2_RESULTS_PATH`

### Verify setup [#verify-setup]

```bash
PYTHONPATH=${WORLDFOUNDRY_CACHE_DIR}/repos/rail-berkeley--bridge_data_v2:src ${WORLDFOUNDRY_CONDA_ROOT}/worldplay/bin/python -m worldfoundry.evaluation.tasks.execution.orchestration.runtime_preflight --profile bridgedata-v2 --manifest worldfoundry/data/benchmarks/runtime_profiles/official/bridgedata-v2.yaml --output-dir tmp/worldfoundry_preflight/bridgedata-v2 --json
```

## Run Evaluation [#run-evaluation]

After preparing the assets and candidate artifacts listed on this page, use the public run entry points below.

### Set Candidate Artifacts [#set-candidate-artifacts]

```bash
export WORLDFOUNDRY_GENERATED_ARTIFACT_DIR=/path/to/generated/artifacts
```

### Import Official Results [#import-official-results]

```bash
worldfoundry-eval zoo benchmark-run \
  --benchmark-id bridgedata-v2 \
  --mode normalizer \
  --official-results-path /path/to/official/results-or-report \
  --generated-artifact-dir "${WORLDFOUNDRY_GENERATED_ARTIFACT_DIR}" \
  --output-dir tmp/bridgedata-v2/normalizer \
  --json
```

### Score Generated Artifacts [#score-generated-artifacts]

```bash
worldfoundry-eval zoo benchmark-run \
  --benchmark-id bridgedata-v2 \
  --mode official-run \
  --generated-artifact-dir "${WORLDFOUNDRY_GENERATED_ARTIFACT_DIR}" \
  --output-dir tmp/bridgedata-v2/official-run \
  --json
```

### Direct In-Tree Runner [#direct-in-tree-runner]

```bash
worldfoundry-eval zoo benchmark-run --benchmark-id bridgedata-v2 --mode official-validation --official-results-path <official_results.json> --output-dir <out> --json
```

## Requirements [#requirements]

### Inputs [#inputs]

* BridgeData V2 dataset materialization
* Eval split definition
* Policy checkpoint or offline prediction dump
* Action normalization and task-goal mapping

### Environment [#environment]

* `WORLDFOUNDRY_BRIDGEDATA_V2_RESULTS_PATH`

### Outputs [#outputs]

* `scorecard.json`
* `raw_results.jsonl`

[← Benchmark Hub](/docs/evaluation/benchmark-hub)