Philip May
commited on
Commit
•
b5f8a97
1
Parent(s):
823f41f
Update README.md
Browse files
README.md
CHANGED
|
@@ -23,4 +23,5 @@ That is why we suggest to use [DeepSpeed](https://github.com/microsoft/DeepSpeed
|
|
| 23 |
In particular, we recommend the [ZeRO-3 Example](https://huggingface.co/docs/transformers/main_classes/deepspeed#zero3-example) `auto` configuration.
|
| 24 |
|
| 25 |
> ZeRO-Offload pushes the boundary of the maximum model size that can be trained efficiently using minimal GPU resources, by exploiting computational and memory resources on both GPUs and their host CPUs.
|
|
|
|
| 26 |
see [ZeRO-Offload](https://www.deepspeed.ai/features/#zero-offload)
|
|
|
|
| 23 |
In particular, we recommend the [ZeRO-3 Example](https://huggingface.co/docs/transformers/main_classes/deepspeed#zero3-example) `auto` configuration.
|
| 24 |
|
| 25 |
> ZeRO-Offload pushes the boundary of the maximum model size that can be trained efficiently using minimal GPU resources, by exploiting computational and memory resources on both GPUs and their host CPUs.
|
| 26 |
+
|
| 27 |
see [ZeRO-Offload](https://www.deepspeed.ai/features/#zero-offload)
|