YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
train-tui
A lightweight terminal dashboard for monitoring AI training runs. Pure C, no dependencies.
๐ฌ Live demo: https://herrei.github.io/train-tui/
Quick start
git clone https://github.com/HerRei/train-tui.git
cd train-tui
make
./train_tui -p basicsr /path/to/project my_experiment config.yml 12345
Press q to quit.
Arguments
./train_tui [options] <project_root> <exp_name> <config_name> <pid> [log_file]
-p <profile> basicsr, lightning, or hf (default: basicsr)
-c <file> custom profile file
-t <total> total iterations (if no parseable config)
-g <backend> amd, nvidia, none, or auto (default: auto)
Use - for config_name if your framework has no config file.
Frameworks
| Profile | For |
|---|---|
basicsr |
BasicSR, Real-ESRGAN, HAT, SwinIR |
lightning |
PyTorch Lightning |
hf |
HuggingFace Trainer |
custom |
Anything else โ write a key=value profile file (-c), see sample.profile |
GPU backends
| Backend | Source |
|---|---|
amd |
sysfs (amdgpu) |
nvidia |
nvidia-smi |
none |
CPU-only training |
Auto-detected. Force with -g.
Safety
Read-only. Never signals, writes to, or opens the training process/files for writing. Only reads the log, config, GPU stats, and checkpoint listing.
Requirements
- C compiler (gcc or clang)
- Linux
License
MIT
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐ Ask for provider support