Yulong Ao
Yulong Ao
### PR types Others ### PR changes Others ### Describe This pr adds the pybind and python implementation for ProcessMesh and DeviceMesh.
Hi, I can see that GShard can run large models on TPUs, does it also support GPUs now ?
Hi, is there any plan to support more expression syntaxes like this library [clipp](https://github.com/muellan/clipp#examples). Especially, I think an option with multiple arguments is very useful such as `./exe --list=1,2,3,4` or...
Based on the examples, vllm can launch a server with a single model instances. Can vllm serving clients by using multiple model instances? With multiple model instances, the sever will...
For the current implementation, the provided code allows us to monitor all parameters or gradients of a model using the watch function. However, it lacks options to specify particular parameters...
Based on the code, the ordering of parallelization and optimization appears to be: PP → TP → AC → Compile → FSDP/DDP. Is it possible to modify this ordering? If...