UER-py icon indicating copy to clipboard operation
UER-py copied to clipboard

关于文本生成

Open yuzhang112 opened this issue 4 years ago • 2 comments

你好,我想用自己的语料,对预训练GPT2模型进行微调,然后进行文本生成任务,请问可以用增量预训练的命令进行微调吗?

这是我现在使用的预训练命令: python3 pretrain.py --dataset_path dataset.pt --pretrained_model_path models/cluecorpussmall_gpt2_seq1024_model.bin-250000 --vocab_path models/google_zh_vocab.txt --config_path models/gpt2/distil_config.json --output_model_path models/my_gpt2_model.bin --world_size 1 --gpu_ranks 0 --total_steps 10000 --save_checkpoint_steps 5000 --report_steps 1000 --learning_rate 5e-5 --batch_size 32 --embedding word_pos --remove_embedding_layernorm --encoder transformer --mask causal --layernorm_positioning pre --target lm --tie_weights

yuzhang112 avatar Nov 02 '21 13:11 yuzhang112

可以


发件人: yuzhang112 @.> 发送时间: Tuesday, November 2, 2021 9:59:25 PM 收件人: dbiir/UER-py @.> 抄送: Subscribed @.***> 主题: [dbiir/UER-py] 关于文本生成 (Issue #219)

你好,我想用自己的语料,对预训练GPT2模型进行微调,然后进行文本生成任务,请问可以用增量预训练的命令进行微调吗?

这是我现在使用的预训练命令: python3 pretrain.py --dataset_path dataset.pt --pretrained_model_path models/cluecorpussmall_gpt2_seq1024_model.bin-250000 --vocab_path models/google_zh_vocab.txt --config_path models/gpt2/distil_config.json --output_model_path models/my_gpt2_model.bin --world_size 1 --gpu_ranks 0 --total_steps 10000 --save_checkpoint_steps 5000 --report_steps 1000 --learning_rate 5e-5 --batch_size 32 --embedding word_pos --remove_embedding_layernorm --encoder transformer --mask causal --layernorm_positioning pre --target lm --tie_weights

― You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/dbiir/UER-py/issues/219, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AE3SPV7HQGTVARW2QM6XHJTUJ7VD3ANCNFSM5HGQ7EHA. Triage notifications on the go with GitHub Mobile for iOShttps://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Androidhttps://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

ydli-ai avatar Nov 02 '21 14:11 ydli-ai

好的,感谢回复

yuzhang112 avatar Nov 10 '21 02:11 yuzhang112