bytedance / bytedance/1d-tokenizer

maskgit_vqgan‘s vq_loss is incorrect

Open
#95 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
Jupyter Notebook
Stars
1.2k
Forks
70
PR merge metrics
No merged PRs in 30d

Description

In the code, the vq_loss is incorrect, and the commitment_cost is placed in the wrong position.

https://github.com/bytedance/1d-tokenizer/blob/bdec006fb7226b309aaf3511956d0aafac30125a/modeling/modules/maskgit_vqgan.py#L296-L299

After correction:
In the code, the vq_loss is fixed, and the commitment_cost is placed in the correct position.

```
if return_loss:
loss = self.commitment_cost * torch.mean((z_q.detach() - hidden_states) ** 2) + torch.mean(
(z_q - hidden_states.detach()) ** 2
)
```

Contributor guide

No contributing guide indexed for this repository

Research direction

Open modeling/modules/maskgit_vqgan.py at lines 296-299 and inspect how vq_loss and commitment_cost are currently applied. Compare the calculation with the corrected expression in the issue, then run the repository's relevant checks; done means the loss uses the stated placement and weighting without unrelated changes.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, pytorch
Domain
machine-learning
Issue type
Bug
Difficulty
1/5
Estimated time
Under an hour
Activity status
Stale
Clarity
Clearly specified
Newbie friendliness
55/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.