garychan22

Hi, thanks for your excellent project ! I want to evaluate the model during training by adding the following code in the train.py script, which leads to program stucking. ```...

TypeError: chunk_batch() argument after ** must be a mapping, not NoneType

2

Hi, I have tried to train dreamfusion-if but the following error was encountered. 

[Docs] how to derive the results similar to OpenPose

2

### 📚 The doc issue Hi, after reading the documentation of mmpose, I still have no idea about how to derive the similar results (json and keypoint image) to OpenPose...

why only controlling the weights for the unconditional prediction

1

hi, thanks for your excellent repo and i have much fun with trying it i have a quesion about the attention control on the unconditional prediction only in AttentionControl (p2p_utils.py)...

hi, as described in the title, i have found that the results include multiple boxes of the same object. dose this project provide some tool script to merge these boxes?...

where do we need the image_embeddings_cls in training_step

1

hi, thanks for your excellent work here! I am reading the code and a little bit confused by the image_embeddings_cls in the training_step. The attention layer accurately takes the **image_embeddings**...

any suggestions for training ip adapter with flux

2

hi, i am trying to train an ip adapter for flux with a subset of laion-2b (about 3M). the training loss decreases very fast in hundreds of iterations and stays...

demo script for chain-of-thought

1

Hi, thanks for your great work! I want to perform chain-of-thought with qwen2.5-vl, is there any script provided? thanks

### Describe the bug I have trained a sdxl lora with the script https://github.com/huggingface/diffusers/blob/7bc0a07b1947bcbe0e84bbe9ecf8ae2d234382c4/examples/advanced_diffusion_training/train_dreambooth_lora_sdxl_advanced.py , and a loading error was encountered during the validation in the end of the training...

Qwen3-VL-30B-A3B-Instruct output image captions of all "!"

2

Hi, thanks for your excellent work here. I am using the package vllm to run inference with Qwen3-VL-30B-A3B-Instruct for image captioning, about 20-30% of the generated captions are all "!"....