[docs] refactor docs for easier info parsing #175
Conversation
| @@ -128,540 +125,24 @@ eval $cmd | |||
| echo -ne "-------------------- Finished executing script --------------------\n\n" | |||
| ``` | |||
|
|
|||
| ### Inference: | |||
There was a problem hiding this comment.
Went to specific model docs.
| > [!NOTE] | ||
| > To lower memory requirements: | ||
| > - Use a DeepSpeed config to launch training (refer to [`accelerate_configs/deepspeed.yaml`](./accelerate_configs/deepspeed.yaml) as an example). | ||
| > - Pass `--precompute_conditions` when launching training. | ||
| > - Pass `--gradient_checkpointing` when launching training. | ||
| > - Pass `--use_8bit_bnb` when launching training. Note that this is only applicable to Adam and AdamW optimizers. | ||
| > - Do not perform validation/testing. This saves a significant amount of memory, which can be used to focus solely on training if you're on smaller VRAM GPUs. |
There was a problem hiding this comment.
In docs/training/optimizations.md.
| > - Pass `--use_8bit_bnb` when launching training. Note that this is only applicable to Adam and AdamW optimizers. | ||
| > - Do not perform validation/testing. This saves a significant amount of memory, which can be used to focus solely on training if you're on smaller VRAM GPUs. | ||
|
|
||
| ## Memory requirements |
There was a problem hiding this comment.
We don't need as have it here already:
https://github.com/a-r-r-o-w/finetrainers/tree/main/training
| ).frames[0] | ||
| export_to_video(output, "output.mp4", fps=15) | ||
| ``` | ||
| | Model Name | Tasks | Ckpts Tested | Min. GPU<br>VRAM | Comments | |
There was a problem hiding this comment.
Would remove the comments column here tbh. All models are supported with multi-resolution and multi-frames so far.
LTX is fast to train but it is extremely hard to teach it new styles tbh. I have had very limited success getting a good lora. Have not figured out the best settings yet but continuing to try.
There was a problem hiding this comment.
Oh I kept the comments column for including anything we should note that couldn't be fit in the other columns. Anything that's out of the ordinary and is absolutely important for the users to know.
WDYT?
Co-authored-by: a-r-r-o-w <contact.aryanvs@gmail.com>
Co-authored-by: a-r-r-o-w <contact.aryanvs@gmail.com>
|
@a-r-r-o-w thanks for the reviews! In the latest commits:
|
Approach taken: