adapter-support

LoRA adapter for the support mode of joduor/adaptive-model.

Status

Not yet trained. Weights will appear here after running the training script.

Train it

ash git clone https://huggingface.co/joduor/adaptive-model cd adaptive-model pip install -r training/requirements.txt huggingface-cli login python training/train_adapter.py --mode support --hub-org joduor

Training takes ~30 min on a T4 (free Colab) with QLoRA enabled.