我如何从头开始训练 gpt 2？
How do I train gpt 2 from scratch?
我想从头开始训练gpt 2，但是在我找到的文章中只有基于预训练模型的微调方法。
我已将此 https://github.com/nshepperd/gpt-2 用于现有模型的训练。我是否应该编辑这些 Python 脚本以从头开始训练？
我在这个回购 https://github.com/nshepperd/gpt-2
的 'issues' 中找到了答案
If you want to not use the released model at all, for instance because
you want to train a model with incompatible hyperparameters, it should
be sufficient to just skip the restore from the released model
checkpoint (around train.py:164-177) on your first run so the
parameters will all be randomly initialized.
我想从头开始训练gpt 2，但是在我找到的文章中只有基于预训练模型的微调方法。 我已将此 https://github.com/nshepperd/gpt-2 用于现有模型的训练。我是否应该编辑这些 Python 脚本以从头开始训练？
我在这个回购 https://github.com/nshepperd/gpt-2
的 'issues' 中找到了答案If you want to not use the released model at all, for instance because you want to train a model with incompatible hyperparameters, it should be sufficient to just skip the restore from the released model checkpoint (around train.py:164-177) on your first run so the parameters will all be randomly initialized.