You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+3-3
Original file line number
Diff line number
Diff line change
@@ -145,7 +145,7 @@ The majority of scripts is licensed under ASL 2.0 (including codes from Diffuser
145
145
- Setting mixed precision to `no` seems to use less memory than `fp16` or `bf16`.
146
146
- Training is possible with a memory usage of about 17GB with a batch size of 1 and fp32. If you specify the `--full_bf16` option, you can further reduce the memory usage (but the accuracy will be lower). With the same memory usage as before, you can increase the batch size.
147
147
- PyTorch 2.1 or later is required because it uses the new API `Tensor.register_post_accumulate_grad_hook(hook)`.
148
-
- Mechanism: Normally, backward -> step is performed for each parameter, so all gradients need to be temporarily stored in memory. "Fuse backward and step" reduces memory usage by performing backward/step for each parameter and reflecting the gradient immediately.
148
+
- Mechanism: Normally, backward -> step is performed for each parameter, so all gradients need to be temporarily stored in memory. "Fuse backward and step" reduces memory usage by performing backward/step for each parameter and reflecting the gradient immediately. The more parameters there are, the greater the effect, so it is not effective in other training scripts (LoRA, etc.) where the memory usage peak is elsewhere, and there are no plans to implement it in those training scripts.
149
149
150
150
- Optimizer groups feature is added to SDXL training. PR [#1319](https://github.com/kohya-ss/sd-scripts/pull/1319)
151
151
- Memory usage is reduced by the same principle as Fused optimizer. The training results and speed are the same as Fused optimizer.
@@ -162,14 +162,14 @@ The majority of scripts is licensed under ASL 2.0 (including codes from Diffuser
0 commit comments