Pretrained weight tags for lucid.models.gpt2_xlarge.
Radford et al. 2019 48-layer trunk (H=1600, A=25, 1.56 B params).
Attributes
WEBTEXTWeightEntryWebText pre-training checkpoint sourced from
transformers/gpt2-xl.DEFAULTWeightEntryAlias for
WEBTEXT.Notes
Reference: Radford, Wu, Child, Luan, Amodei, Sutskever, "Language Models are Unsupervised Multitask Learners", OpenAI 2019.
Examples
>>> from lucid.models import gpt2_xlarge
>>> model = gpt2_xlarge(pretrained=True).eval()