Repository navigation
Start higher resolution training from a lower resolution checkpoint - #104
Merged
Merged
Conversation
henryruhs
requested changes
Oct 9, 2026
henryruhs
left a comment
henryruhs
left a comment
Member
There was a problem hiding this comment.
the custom load initial weights is off - also where it is located in the config
| loss_embedder_path = .models/arcface.pt | ||
| face_masker_path = .models/face_masker.pt | ||
| face_aligner_path = .models/face_aligner.pt | ||
| initial_path = |
Member
There was a problem hiding this comment.
should not be empty in READM.md - it has to explain how to use it
| loss_embedder_path = | ||
| face_masker_path = | ||
| face_aligner_path = | ||
| initial_path = |
Member
There was a problem hiding this comment.
this are helper models, I doubt initial_path belongs here
| from configparser import ConfigParser | ||
|
|
||
|
|
||
| def create_test_config_parser(output_size : int) -> ConfigParser: |
Member
There was a problem hiding this comment.
unrelated test, also what does it test accept ConfigParser? for me this is pointless
| target_generator = Generator(create_test_config_parser(output_size)) | ||
| source_state = source_generator.state_dict() | ||
| target_state = target_generator.state_dict() | ||
| initial_keys = [] |
Member
There was a problem hiding this comment.
keys seems to be wrong wording, ain't this layers name?
| ) | ||
|
|
||
|
|
||
| def load_initial_weights(hyperswap_trainer : HyperSwapTrainer, initial_path : str) -> None: |
Member
There was a problem hiding this comment.
I do not understand why this exist, initial weights are supported by lightning
model = MyLightningModule.load_from_checkpoint(
checkpoint_path="initial_weights.ckpt",
strict=False # Ignores missing or extra keys automatically
)
or
trainer = Trainer(resume_from_checkpoint="path/to/your/checkpoint.ckpt")
trainer.fit(model)
henryruhs
approved these changes
Oct 9, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds initial_path to [training.model]. When there is no checkpoint to resume, training first loads the generator weights from initial_path. This lets 512 and 1024 training start from a trained 256 model instead of from scratch.