Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot extract the features (columns) for the split 'train' of the config 'default' of the dataset.
Error code:   FeaturesError
Exception:    ArrowInvalid
Message:      Schema at index 2 was different: 
folder: string
archive: string
archive_bytes: int64
archive_sha256: string
file_count: int64
unpacked_bytes: int64
vs
folder: string
source: string
file_count: int64
unpacked_bytes: int64
excluded: list<item: null>
files: list<item: struct<path: string, bytes: int64, sha256: string>>
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/split/first_rows.py", line 244, in compute_first_rows_from_streaming_response
                  iterable_dataset = iterable_dataset._resolve_features()
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 4408, in _resolve_features
                  features = _infer_features_from_batch(self.with_format(None)._head())
                                                        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2679, in _head
                  return next(iter(self.iter(batch_size=n)))
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2861, in iter
                  for key, pa_table in ex_iterable.iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2395, in _iter_arrow
                  yield from self.ex_iterable._iter_arrow()
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 564, in _iter_arrow
                  yield new_key, pa.Table.from_batches(chunks_buffer)
                                 ~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^
                File "pyarrow/table.pxi", line 5040, in pyarrow.lib.Table.from_batches
                File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
                File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
                  raise convert_status(status)
              pyarrow.lib.ArrowInvalid: Schema at index 2 was different: 
              folder: string
              archive: string
              archive_bytes: int64
              archive_sha256: string
              file_count: int64
              unpacked_bytes: int64
              vs
              folder: string
              source: string
              file_count: int64
              unpacked_bytes: int64
              excluded: list<item: null>
              files: list<item: struct<path: string, bytes: int64, sha256: string>>

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

PostTrainBench — AgentBench 环境资源

下载入口:Areyliu/AgentBench_env。

本仓库保存 PostTrainBench 接入 DBBench、WebShop、ALFWorld 所需的三个资源目录。每个目录以 tar.gz 包保存,下载后解压到 PostTrainBench/envs/,即可恢复任务启动脚本使用的目录结构。

资源内容

Hugging Face 目录 包内恢复的目录 内容
agentbenchdbbench/ agentbenchdbbench/data/、run/ MySQL 8.0.28 初始化的干净数据模板;不包含此前测试的 SQL 历史或生成密钥
agentbenchwebshop/ agentbenchwebshop/webshop-src/ WebShop 源码、商品/用户指令数据、100k Lucene 索引和构建资源
agentbenchalfworld/ agentbenchalfworld/data/ ALFWorld 游戏、任务 JSON 和 PDDL 数据

WebShop 完整原始商品文件和 ALFWorld 完整游戏集合保留,便于后续扩充训练数据。当前评测使用的商品子集、索引及题目划分保持不变。WebShop 的 Python 缓存文件不纳入资源包。

各目录的 ARCHIVE.json 记录压缩包 SHA-256、文件数及大小,MANIFEST.json 记录包内每个文件的 SHA-256。根目录 ARCHIVES.json 汇总三个资源包。

下载与解压

以下命令在 Linux 上执行。下载目录应位于 envs/ 外;对已有资源先备份,再将压缩包解压到空的资源目录。

python -m pip install huggingface_hub
hf download Areyliu/AgentBench_env --repo-type dataset \
  --local-dir /absolute/path/to/agentbench-download

cd /absolute/path/to/PostTrainBench
mkdir -p envs
for name in agentbenchdbbench agentbenchwebshop agentbenchalfworld; do
  tar -xzf "/absolute/path/to/agentbench-download/$name/$name.tar.gz" -C envs
done

验证压缩包后再解压:

python - <<'PY'
import hashlib
import json
from pathlib import Path
root = Path('/absolute/path/to/agentbench-download')
for info in json.loads((root / 'ARCHIVES.json').read_text()):
    path = root / info['archive']
    digest = hashlib.sha256()
    with path.open('rb') as stream:
        for chunk in iter(lambda: stream.read(8 * 1024 * 1024), b''):
            digest.update(chunk)
    assert digest.hexdigest() == info['archive_sha256'], path
    print('OK', path.name)
PY

Conda 依赖

资源包不包含 Python/venv、JDK、MySQL 二进制或安装包。共享 Conda 环境须满足 PostTrainBench 中以下文件的依赖:

  • containers/requirements-direct.txt:PTB 原有 Python 依赖;
  • containers/requirements-agentbench.txt:三个任务的补充 Python 依赖;
  • containers/environment-agentbench.yml:Python 3.10、Java 11、MySQL 8.0.28、ICU 70.1。

在 PostTrainBench 根目录创建共享环境:

bash scripts/setup_agentbench_conda.sh --prefix /absolute/path/to/ptb-agentbench

在 PostTrainBench 的 .env 中配置:

PTB_AGENTBENCH_CONDA_PREFIX="/absolute/path/to/ptb-agentbench"

随后照常使用 src/run_task.sh。任务资源默认位于 <repo>/envs,可以用 PTB_AGENTBENCH_ENVS_ROOT 覆盖。模型训练、vLLM 模型服务及模型评测仍使用 PTB 原有 Apptainer SIF。

DBBench 数据模板面向同版本 Linux x86_64 MySQL。也可以不下载 DBBench 模板,直接执行 PTB_AGENTBENCH_CONDA_PREFIX=... bash scripts/bootstrap_agentbench_envs.sh --only dbbench 初始化。

来源

资源随其原始项目的许可使用;源文件中的许可证保留。训练/验证/正式题目切分和 HTTP 服务适配代码位于 PostTrainBench 的三个任务目录中。

Downloads last month
41