YAML Metadata
Warning:
empty or missing yaml metadata in repo card
(https://huggingface.co/docs/hub/model-cards#model-card-metadata)
To load checkpoint, use the TunedLens
class in the train_tunedlens_mamba.py
file.
import accelerate
lens = TunedLens(model.backbone.layers, d_model_hidden_states).to("cuda")
ckpt = 'tunedlens_34_mamba-130m.safetensors'
lens = accelerate.load_checkpoint_and_dispatch(lens, ckpt)