你的PR提交成功，感谢你对开源项目的贡献!
请关注后续CI自动化测试结果，详情请参考Paddle-CI手册。
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.

Sorry to inform you that 1269701's CIs have passed for more than 7 days. To prevent PR conflicts, you need to re-run all CIs manually.

LGTM

已经为chi2和lkj两个分布新增负样本测试

需要通过流水线

嗯嗯目前还剩下俩，看起来需要同学手动通过

please add link of rfc in description above.

except __init__, shall we need method of expand?

sry,
I'm not quite sure about the role/function of the expand method.
Can you describe it to me? I haven't found a specifically implemented expand method among the distribution.

please add link of rfc in description above.

done.

@cmcamdy 关于expand的问题，comment by @jeff41404

pytorch 中有expand方法，我们是否需要，设计文档中也没有说明，如果不需要，代码转换时如何替代

@cmcamdy 关于expand的问题，comment by @jeff41404

pytorch 中有expand方法，我们是否需要，设计文档中也没有说明，如果不需要，代码转换时如何替代

我明白了，实际上pytorch中的extend是对batch维度的扩展，如gamma中的实现：https://github.com/pytorch/pytorch/blob/e3ac61587aa368c613ef01df1f328a396b64cd5d/torch/distributions/gamma.py#L60
这部分逻辑paddle中并没有专门实现（因此在设计的时候并没有计划实现expand类的方法），但在如果想要等价实现，只需要在初始化的时候传入带batch的参数即可，我们在sample中会考虑多维shape的情况

I see, in PyTorch, the expand functionality is actually used to extend the batch dimension, as exemplified in the Gamma distribution implementation: https://github.com/pytorch/pytorch/blob/e3ac61587aa368c613ef01df1f328a396b64cd5d/torch/distributions/gamma.py#L60
PaddlePaddle doesn't have a dedicated implementation for this specific logic (indicating that it wasn't planned as part of the original design to include an expand-like method). However, to achieve an equivalent functionality in PaddlePaddle, one can simply pass batch-aware parameters during initialization. When sampling, we will take into account multi-dimensional shapes.

@cmcamdy 关于expand的问题，comment by @jeff41404
pytorch 中有expand方法，我们是否需要，设计文档中也没有说明，如果不需要，代码转换时如何替代

我明白了，实际上pytorch中的extend是对batch维度的扩展，如gamma中的实现：https://github.com/pytorch/pytorch/blob/e3ac61587aa368c613ef01df1f328a396b64cd5d/torch/distributions/gamma.py#L60 这部分逻辑paddle中并没有专门实现（因此在设计的时候并没有计划实现expand类的方法），但在如果想要等价实现，只需要在初始化的时候传入带batch的参数即可，我们在sample中会考虑多维shape的情况

I see, in PyTorch, the expand functionality is actually used to extend the batch dimension, as exemplified in the Gamma distribution implementation: https://github.com/pytorch/pytorch/blob/e3ac61587aa368c613ef01df1f328a396b64cd5d/torch/distributions/gamma.py#L60 PaddlePaddle doesn't have a dedicated implementation for this specific logic (indicating that it wasn't planned as part of the original design to include an expand-like method). However, to achieve an equivalent functionality in PaddlePaddle, one can simply pass batch-aware parameters during initialization. When sampling, we will take into account multi-dimensional shapes.

ok, thanks

@luotao1 , 我重启了下APPROVAL的CI，似乎还需要下面几位同学的approve，这个CI里面的报错提示如下：
2024-06-05 11:28:16 0. You must have raindrops2sea or XiaoguangHu01 approval for change 20+ files or add than 1000+ lines of content.
2024-06-05 11:28:16 1. You must have one QA (XieYunshen(Recommend) or chalsliu) approval for setting parameter RUN_TYPE as EXCLUSIVE, DIST, HYBRID, NIGHTLY, EXCLUSIVE:NIGHTLY or DISTNIGHTLY, or setting parameter SERIAL, or setting TIMEOUT properties.

Approval 流水线我们会来操作，请耐心等待 review 即可

Sorry to inform you that fe3edf2's CIs have passed for more than 7 days. To prevent PR conflicts, you need to re-run all CIs manually.

已修改