search

TencentARC / SmartEdit

Official code of SmartEdit [CVPR-2024 Highlight]
235 stars 8 forks source link

model checkpoint #31

Open ybx193670 opened 5 days ago

ybx193670 commented 5 days ago

Could you please provide the models that fine-tuned these methods (InstructPix2Pix, MagicBrush,) with the same dataset as SmartEdit?

yuzhou914 commented 1 day ago

Thanks for your interest in our work. While these checkpoints are cleaned before I leave Tencent. The training process of InstructPix2Pix and MagicBrush with our data is simple. We adopt the code from here with InstructPix2Pix and MagicBrush initialization with our datasets. You could simply reproduce it if you are interested.

ybx193670 commented 1 day ago

收到，谢谢你的回复

---- Replied Message ---- | From | @.> | | Date | 09/28/2024 13:28 | | To | @.> | | Cc | @.>@.> | | Subject | Re: [TencentARC/SmartEdit] model checkpoint (Issue #31) |

Thanks for your interest in our work. While these checkpoints are cleaned before I leave Tencent. The training process of InstructPix2Pix and MagicBrush with our data is simple. We adopt the code from here[https://github.com/huggingface/diffusers/blob/main/examples/instruct_pix2pix/train_instruct_pix2pix.py] with our datasets. You could simply reproduce it if you are interested.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

ybx193670 commented 1 day ago

请问，在其他数据集上微调也是邮件链接中的代码吗

---- Replied Message ---- | From | @.> | | Date | 09/28/2024 13:28 | | To | @.> | | Cc | @.>@.> | | Subject | Re: [TencentARC/SmartEdit] model checkpoint (Issue #31) |

Thanks for your interest in our work. While these checkpoints are cleaned before I leave Tencent. The training process of InstructPix2Pix and MagicBrush with our data is simple. We adopt the code from here[https://github.com/huggingface/diffusers/blob/main/examples/instruct_pix2pix/train_instruct_pix2pix.py] with our datasets. You could simply reproduce it if you are interested.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

yuzhou914 commented 1 day ago

We use the same training scripts like we train SmartEdit (deepspeed + huggingface trainer), while we adopt the data and model parts from the above link. All experiments are conducted like this.