-
When I run convert.py my CPU usage is over 70% and GPU is only 1%. Is that normal?
Is there a way to make my GPU do the work?
-
### Describe the bug
i noticed that when i add ```align_device_hook``` to module in pipeline manually, then ```load_lora_weights``` function will enable the sequential cpu offload. so i dig deeper a…
-
Hi,
Great product.
In the API section i can see GPU offloading.
After the most recent update, I appear to have lost GPU Offloading from chat.
Love the product and hope you can help.
![ima…
-
It seems like there are (at least) two different Raspberry Pi Zero 2 W revisions on the market, and they differ in terms of which WiFi chip is used.
The BCM43430/1 works as expected, and seems to m…
-
https://github.com/NVIDIA/TransformerEngine/blob/e3bb24e5a347c58353e62307bc84cca856f9e9be/transformer_engine/pytorch/module/linear.py#L405-L407
if the weight.requires_grad set to False, when to cal…
-
Hi,
I am using deepspeed zero-2 with cpu offloading for finetuning LLM model.
I keep getting error like this without any detail error description.
```
[2023-10-26 17:54:44,801] [INFO] [launch.…
-
@HerManNav can you write here the results you obtain?
-
**Is your feature request related to a problem? Please describe.**
Testing the mod I noticed it won't transfer units if there is no HC preset.
Meaning zeus placed units won't be transferred to the …
-
Moving one task from one place to another at code method level.
-
DiffusionKit has an excellent memory optimisation for Flux where it calculates modulation parameters ahead of time
then offloads the adaLN_modulation parameters which for fp16 saves ~ 6.5 GB peak me…