StableLlama
StableLlama
PR #353 does this.
PR #353 does the cropping part. And it allows masks by placing rectangle marks on the image, but it doesn't allow real mask painting. The masks are then exported in...
Running again, this time with reduced numbers for the different steps it's failing after 25 more steps. And the only number reduced to 25 is `--eval_steps_interval` - so I guess...
Changing the config to `"--eval_steps_interval": 2500000` (i.e. effectively disabling eval) makes the model train normally, including generation of validation images et al. => This issue is caused by eval
This PR has significantly improved my workflow of preparing images for training. As an example this is how I used it for my test: - Collected images (I stopped with...
@jhc13 this PR is now open for a month. Could you already have a look at it? I'm currently using it in a rather big captioning project and it's helping...
As I've written, I see these additions not as a dataset management tool (for that far too much is missing) but as a workflow tool to prepare the images for...
@Spikhalskiy as I couldn't wait any longer, I've now created a fork, I'll call it "taggui workflow" as it is supporting the image preparation workflow. Next I'll pull a few...
> I think this should be adressed asap. I would preffer this over dataset tool from onetrainer but I am not running anything with virus (even if it woud be...
No, I'm using Python 3.12.3 as this is the version that comes with the latest Ubuntu LTS (24.04)