Skip to content

Qwen3-Reranker 训练数据集格式到底是啥 #9702

Description

@yvway

Checklist / 检查清单

  • I have searched existing issues, and this is a new bug report. / 我已经搜索过现有的 issues,确认这是一个新的 bug report。

Bug Description / Bug 描述

[WARNING:swift] 👆👆👆There are errors in the dataset, the data will be deleted
[INFO:swift] Traceback (most recent call last):
File "/home/yw/.conda/envs/swift/lib/python3.10/site-packages/swift/dataset/preprocessor/core.py", line 193, in batched_preprocess
self._check_messages(r)
File "/home/yw/.conda/envs/swift/lib/python3.10/site-packages/swift/dataset/preprocessor/core.py", line 71, in _check_messages
assert len(messages) > 0, f'messages: {messages}'
AssertionError: messages: []

[WARNING:swift] 👆👆👆There are errors in the dataset, the data will be deleted
Map (num_proc=8): 100%|█████████████████████████████████████████████████████████████████████████████████████████████████| 3000/3000 [00:00<00:00, 9476.85 examples/s]
[INFO:swift] Dataset filtered, origin length: 3000, filtered dataset length: 0
[rank0]: Traceback (most recent call last):
[rank0]: File "/home/yw/.conda/envs/swift/lib/python3.10/site-packages/swift/cli/sft.py", line 20, in
[rank0]: sft_main()
[rank0]: File "/home/yw/.conda/envs/swift/lib/python3.10/site-packages/swift/pipelines/train/sft.py", line 340, in sft_main
[rank0]: return SwiftSft(args).main()
[rank0]: File "/home/yw/.conda/envs/swift/lib/python3.10/site-packages/swift/pipelines/base.py", line 52, in main
[rank0]: result = self.run()
[rank0]: File "/home/yw/.conda/envs/swift/lib/python3.10/site-packages/swift/ray_utils/base.py", line 168, in wrapper
[rank0]: return func(self, *args, **kwargs)
[rank0]: File "/home/yw/.conda/envs/swift/lib/python3.10/site-packages/swift/pipelines/train/sft.py", line 161, in run
[rank0]: train_dataset, val_dataset = self._prepare_dataset()
[rank0]: File "/home/yw/.conda/envs/swift/lib/python3.10/site-packages/swift/ray_utils/base.py", line 168, in wrapper
[rank0]: return func(self, *args, **kwargs)
[rank0]: File "/home/yw/.conda/envs/swift/lib/python3.10/site-packages/swift/pipelines/train/sft.py", line 107, in _prepare_dataset
[rank0]: train_dataset, val_dataset = self._get_dataset()
[rank0]: File "/home/yw/.conda/envs/swift/lib/python3.10/site-packages/swift/pipelines/train/sft.py", line 81, in _get_dataset
[rank0]: train_dataset, val_dataset = args.load_dataset()
[rank0]: File "/home/yw/.conda/envs/swift/lib/python3.10/site-packages/swift/arguments/base_args/base_args.py", line 351, in load_dataset
[rank0]: train_dataset, val_dataset = load_dataset(
[rank0]: File "/home/yw/.conda/envs/swift/lib/python3.10/site-packages/swift/dataset/loader.py", line 351, in load_dataset
[rank0]: train_dataset, val_dataset = loader.post_process(
[rank0]: File "/home/yw/.conda/envs/swift/lib/python3.10/site-packages/swift/dataset/dataset_meta.py", line 165, in post_process
[rank0]: assert train_sample > 0
[rank0]: AssertionError

文档的格式试了,又来GitHub的Issues中发现不同的格式,试来试去没一种是对的,能不能靠谱一点

How to Reproduce / 如何复现

你们是没人在运维了吗

Additional Information / 补充信息

No response

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type

    Fields

    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions