Ze-Yi LIN
Ze-Yi LIN
Must support
人脸检测模型选的是 https://github.com/ipazc/mtcnn ，考虑更换更好的模型
已添加face++模型，在最新的Demo中可体验
让我看看
多图sft，这是什么场景？可以考虑开一个新的教程文档来适配这个场景
你好，感谢你对代码的反馈： 1. 代码是自己写的 2. 在微调时，Qwen-VL格式的数据仍然可以适用于Qwen2-VL，这一块也可以看看[Qwen2-VL-2B-LaTexOCR](https://github.com/datawhalechina/self-llm/blob/master/models/Qwen2-VL/06-Qwen2-VL-2B-Instruct%20Lora%20%E5%BE%AE%E8%B0%83%E6%A1%88%E4%BE%8B%20-%20LaTexOCR.md)，这一块的代码做了更多的精简和改进，微调效果也work 3. 推理实际已经加载了微调后的lora模型，可以看一下这两行，逻辑是先加载原预训练模型，然后再挂载lora训练后的模型。 ```python model = Qwen2VLForConditionalGeneration.from_pretrained( "./Qwen/Qwen2-VL-2B-Instruct", torch_dtype="auto", device_map="auto" ) model = PeftModel.from_pretrained(model, model_id="./output/Qwen2-VL-2B/checkpoint-62", config=config) ```
This is a very nice scene, will pay attention. 这是一个很不错的场景，会关注
I will pay attention to these valuable features you have proposed.
> Any hope on implementing this? Add saturation is ok. Regarding batch processing, perhaps it would be better to do it in the API?
It seems that the CI errors are caused by network factors: ``` RuntimeError: An error occurred while downloading using `hf_transfer`. Consider disabling HF_HUB_ENABLE_HF_TRANSFER for better error handling. ```