SmartEdit

SmartEdit copied to clipboard

demo code

3

Can you provide a demo code to edit an image using one instruction?

xmu-xiaoma666

Excellent work! Could you provide the training cost? such as training GPU times and the GPU needed.

Rexzhan

inference issue

1

Thanks so much for the excellent work! Hoever I met some issues in inference. Since the LLaVA-1.1 can not be founded, I replace it with LLAVA-1.5, but it occur the...

shuyansy

testing the editing performance directly using stage 1's ckpt

4

Hi! I have a question, it seems that after training stage 1, the llava+qformer's output is aligned with clip text space. Could we directly use the llava and qformer after...

CharlesGong12

LLaVA version

1

`--LLaVA_model_v1_1_7b_path ./checkpoints/LLaVA-7B-v1 \` For the training workflow, does it work for other LLaVA version? I mean v1.1 seems to be archived.

YerongLi

rahatkutubi

In the scripts `script/MLLMSD_7b.sh` and `script/SmartEdit_7b.sh`, you have specified `--bf16 True`, yet it seems that the corresponding deepspeed configuration in `scripts/zero_mixed.json` is missing the line `"bf16": {"enabled": "auto"}.` As a...

Bilibilee

there are lots of bugs in TrainStage1

2

Thank you for your excellent work, but the open-source code indeed has many minor issues, which makes others hesitant to follow your work. During the TrainStage1 phase, the issues are...

Bilibilee

Qformer mm_projector issue

2

Hello, thanks for ur amazing work! I have a problem when running this code, could u help me to solve it? When I train the script DS_MLLMSD11_train.py, I encountered this...

zjutkarma

Trainstage1 parent dir was not created

2

when I run TrainStage1_7b.sh，because you modified the `_save` method of the `trainer`, it will not create a parent directory for checkpoints like `checkpoint-10` (where the number represents the save steps)....

mo230761