SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Update perf_train_gpu_many.md (#31451)

Browse Source 
* Update perf_train_gpu_many.md

* Update docs/source/en/perf_train_gpu_many.md

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Update docs/source/en/perf_train_gpu_many.md

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

---------

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
This commit is contained in:
Rémy Léone
2024-06-18 20:00:26 +02:00
committed by GitHub
parent 280cef51b3
commit 22b41b3f8a
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
docs/source/en/perf_train_gpu_many.md
View File

@@ -56,15 +56,15 @@ impact performance. Here's a breakdown of your options:
If your model can comfortably fit onto a single GPU, you have two primary options: If your model can comfortably fit onto a single GPU, you have two primary options:
1. DDP - Distributed DataParallel 1. DDP - Distributed DataParallel
2. ZeRO - depending on the situation and configuration used, this method may or may not be faster, however, it's worth experimenting with it. 2. [Zero Redundancy Optimizer (ZeRO)](https://arxiv.org/abs/1910.02054) - depending on the situation and configuration used, this method may or may not be faster, however, it's worth experimenting with it.
**Case 2: Your model doesn't fit onto a single GPU:** **Case 2: Your model doesn't fit onto a single GPU:**
If your model is too large for a single GPU, you have several alternatives to consider: If your model is too large for a single GPU, you have several alternatives to consider:
1. PipelineParallel (PP) 1. PipelineParallel (PP)
2. ZeRO 2. [ZeRO](https://arxiv.org/abs/1910.02054)
3. TensorParallel (TP) 3. [TensorParallel](#tensor-parallelism) (TP)
With very fast inter-node connectivity (e.g., NVLINK or NVSwitch) all three strategies (PP, ZeRO, TP) should result in With very fast inter-node connectivity (e.g., NVLINK or NVSwitch) all three strategies (PP, ZeRO, TP) should result in
similar performance. However, without these, PP will be faster than TP or ZeRO. The degree of TP may also similar performance. However, without these, PP will be faster than TP or ZeRO. The degree of TP may also