Zeyu Li comments

Results 19 comments of


                                            Zeyu Li

Failed to build

@hiGiraffe Hope this will help you ### CMakeLists.txt Line4, Line5 set(GKLIB_PATH METIS/GKlib -> set(GKLIB_PATH metis/GKlib set(METIS_PATH METIS -> set(METIS_PATH metis ### build-parmetis.sh: the absoluted path need to be replaced with...

[Bug] multinode fails with RuntimeError: Gloo connectFullMesh failed with [../third_party/gloo/gloo/transport/tcp/pair.cc:144] no error

I'm also experiencing this issue, I'm using two H100 Nodes with 400Gbps RoCE v2 between the H100s. I have tested nccl and gloo between two nodes using the following program....

unexpected keyword argument 'use_scaled_rope' error when running example script with Meta-Llama-3.1-8B

@nwatab Thanks! I wonder where this official source is

Unable to perform visual simulation when running on a server via SSH

Using X11 forwarding (GPT maybe helpful)

Unable to perform visual simulation when running on a server via SSH

@WYYAHYT I don't get this error, I haven't actually tried ssh, but in my experience Taichi can be forwarded via X11

The same code works on Nvidia CUDA but doesn't work on AMD Vulkan

Can you try to use rocm backend on AMDGPU if you are using linux?

Installing issue with gdrdrv

The compliation of GDRCopy needs nv driver. You may compile this on your compute-node. (this is the cause for me)

Error: can't register atexit after shutdown

> Hi [@rajeshitshoulders](https://github.com/rajeshitshoulders) [@vladandrew](https://github.com/vladandrew) [@Yogaht](https://github.com/Yogaht) [@akaashrp](https://github.com/akaashrp), Vidur now uses `seaborn` (based on `matplotlib`) instead of `plotly` (which uses `kaleido`). This removes dependency on `kaliedo` and hence the original error should...

Update model.py

This gets llama3.1 running, but it doesn't use the scaled rope