[Bug] ModuleNotFoundError: No module named 'clt_forge.frontend.data' (Missing directory in src/clt_forge)

[lzliangzh]

Description

When attempting to launch the GPT-2 visual interface using torchrun, the application fails immediately due to a missing module. It appears that the source code references a data directory/package within clt_forge (specifically for data loaders) that does not exist in the current repository structure.

Steps to Reproduce

Run the following command from the project root:

poetry run torchrun --nproc_per_node=1 runners/visual_interface/gpt2/launch_interface.py

Error Traceback

Traceback (most recent call last):
  File "/home/cc/CLT-Forge/runners/visual_interface/gpt2/launch_interface.py", line 10, in <module>
    from clt_forge.frontend.app import main
  File "/home/cc/CLT-Forge/src/clt_forge/frontend/app.py", line 6, in <module>
    from .data.loaders import DataLoader # TODO: 找不到 clt_forge.data.loaders
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ModuleNotFoundError: No module named 'clt_forge.frontend.data'

Expected Behavior

The interface should launch or provide a clear configuration error if data is missing.

Actual Behavior

The program crashes because src/clt_forge/frontend/data/ (or the general clt_forge/data utility) is missing from the file tree, despite being imported in app.py.

Environment

• OS: Ubuntu 24.04.1 LTS
• Python version: 3.11

[florentdraye]

Thanks for reporting this, I found the cause.

The issue came from a .gitignore rule (data/) that unintentionally excluded src/clt_forge/frontend/data/, so that package was missing from the repository.

I opened a PR with the fix: #8