AngelBottomless

@JoanaMarieL Yes, I once tried to attach results after end of training - for example, I assigned the thread to call the inference, but since `accelerator.end_training()` was called before inference...

@JoanaMarieL I'm using the following code in accelerator environment, but seems there is some problem: ``` global run_id if run_id is None: run_id = wandb.run.id # resume if stopped wandb.init(id=run_id,...

@JoanaMarieL Thanks for the response! I successfully resumed the run and logged the metrics with the following code snippet: ```python import wandb wandb.login(key=API_KEY) wandb.init(entity="user_name",project='proj_test',name="test_name") wandb.log({'acc': 0.9, 'loss': 0.2}) #test args...

This is caused because SpawnPlacements does not have Piglin Brutes registry.

@Ghoulboy78 do you want me to resolve conflicts? Or have you heard any plan about thus? :3

@Ghoulboy78 Resolved conflicts! I just tested gradle build, but not sure if it has no NPE :3

@carmeops Of course I can allow it, but usually.... you don't really want to use named items / items with lore.... don't you? 😕 Anyway I'll implement it as option...

@oO0 Seems like you are using 21.5 for warmup step, but it expects int - I'll make it round the value. But if it was unexpected (or using some different...

Its not button, so currently it only applies in real training (no ui change currently) UI will be changed soon too.

Unfortunately I cannot find proper way to update gradio component's value, so its not likely to be implemented. But values are loaded from json and used, unless you see error,...