Junfan Zhang comments

Results 434 comments of


                                            Junfan Zhang

[#2090] fix(client): Replace LinkedList with CopyOnWriteArrayList to store the sending status

Why not using the `synchronized` to ensure the safety of race condition ? @xianjingfeng I also concern the performance degression when using the CopyOnWrite

[#2030] fix(client-spark): Remove javax and logging from relocation list to solve NoClassDefFoundError

Could you help review this? @zhengchenyu

[MINOR] fix(client): Fix allocate size by add 9 bytes

Please fix the conflict. After that, I think I can merge this assp. @maobaolong

[Improvement] Independent rust based shuffle server project

cc @jerqi @advancedxy .

[Improvement] Independent rust based shuffle server project

> > However, if you want to move it to a independent project, I think you should sent a discuss thread to the dev mail list and go through the...

[Improvement] Independent rust based shuffle server project

I hope I could get the push/merge privillege for the independent sub-project, that will make me to speed up the commit/develop for me

Fail in running Llama-3.2-1B

I met the similar problems. > Could you try to print `rope_scaling_factor`? It should be a `int`, but it seems that it is a `dict` from the error message Yes....

Fail in running Llama-3.2-1B

Oh, I think this is not simply to get this value directly, just like this issue: https://github.com/alibaba/Pai-Megatron-Patch/issues/361

[#1796] fix(spark): Implicitly unregister map output on fetch failure

Could you help review this? @advancedxy If I understand incorrectly, feel free to point out

[#1796] fix(spark): Implicitly unregister map output on fetch failure

cc @yl09099 Could you help check some write failure logic, I have refactored these part code and make it align with the fetch failure.