Feature Request: add support for fine-tuned (Falcon) models

[ChristianPala]

Hi folks, I'm trying to use a LoRA fine-tuning of Falcon with multilingual support with your pipeline, but it's not natively supported, would be a good addition to the project! Thanks for the attention

[carmocca]

What error are you getting?

LoRA suport for Falcon is landing with #141. You can install that branch to give it a try

[ChristianPala]

Hi! After converting the weights from the repository: OpenAssistant/falcon-7b-sft-top1-696, via hugging face, to the lit-parrot format, when I try generating, using base.py I get the following runtime size mismatch error: RuntimeError: Error(s) in loading state_dict for Parrot: size mismatch for lm_head.weight: copying a param with shape torch.Size([65040, 4544]) from checkpoint, the shape in current model is torch.Size([65024, 4544]). size mismatch for transformer.wte.weight: copying a param with shape torch.Size([65040, 4544]) from checkpoint, the shape in current model is torch.Size([65024, 4544]).