Leisure
Leisure
How many steps did you train？ I trained 8w steps and the cross_entropy_loss had been around 2, and it didn't seem to decrease from the beginning to the present.
In stage 1 and 2, do you use pre-trained models in https://github.com/sczhou/CodeFormer/releases/tag/v0.1.0?
多谢你的回复，我补充下测试情况： 模型是Restormer-TLC，GPU是V100 32G，输入是720x1280，一张大概需要3~4秒，这个时间正常吗？
你的输出是正常的么？vllm部署的时候对于短输入会输出一连串的叹号，lmdeploy有这个问题吗？
Yes, the problem is the normalization. In function nondist_validation, I change `sr_img = tensor2img([visuals['result']])` to `sr_img = tensor2img([visuals['result']], rgb2bgr=True, min_max=(-1, 1))`, then the output becomes normal.