Fu

This project is really impressive. where is the DC model ?

OneFormerProcessor

2

加载下载好的模型， oneformer出错，huggingface_hub.utils._validators.HFValidationError: Repo id must be in the form 'repo_name' or 'namespace/repo_name': oneformer_ade20k_processor = OneFormerProcessor.from_pretrained("shi-labs/oneformer_ade20k_swin_large") oneformer_ade20k_model = OneFormerForUniversalSegmentation.from_pretrained("shi-labs/oneformer_ade20k_swin_large").to(rank)

intervl 微调后lora合并推理和直接用原来demo推理结果不一致

4

1. internlvl2.0 swift微调后，然后采用lora merged ， swift python 推理代码结果正常。 2. 用原来推理demo, 推理，结果对不上，代码如下： ``` model = AutoModel.from_pretrained( path, torch_dtype=torch.bfloat16, load_in_8bit=True, low_cpu_mem_usage=True, trust_remote_code=True, device_map=device_map).eval() generation_config = dict( num_beams=1, max_new_tokens=2048, do_sample=True, ) ```

onnx export

1

### Motivation onnx export support ? ### Related resources _No response_ ### Additional context _No response_

demo or code

12

nice work When will the demo or code be provided? thanks

train log ？

2

It is already being reproduced. Can you provide some training logs for reference?

Can you provide the StreetVTON and WildVTON data sets? thanks