996_icu

I found that in the latest version (v0.3.2), I can't find the code about data parallelism. Have you removed it? What are the considerations based on this?

[Bug] RuntimeError: Failed to allocate memory for batch_prefill_tmp_v with size 458752000 and alignment 16 in AlignedAllocator

9

### Checklist - [X] 1. I have searched related issues but cannot get the expected help. - [X] 2. The bug has not been fixed in the latest version. -...

test_inter_node.py failed: an illegal memory access was encountered

4

## Description: I test deepep in 2 nodes * 8 H20 mechaine. I can pass the `test_intranode.py`. But fail in `test_internode.py`. Could you please help me with that? ## Error...