search

OpenGVLab / UniFormerV2

[ICCV2023] UniFormerV2: Spatiotemporal Learning by Arming Image ViTs with Video UniFormer
https://arxiv.org/abs/2211.09552
Apache License 2.0
291 stars 18 forks source link

关于预训练模型 #49

Closed xiezexun closed 1 year ago

xiezexun commented 1 year ago

作者您好，请问vit的预训练模型在哪里下载？ _MODELS = { "ViT-B/16": os.path.join(MODEL_PATH, "ViT-B-16.pth"), "ViT-L/14": os.path.join(MODEL_PATH, "vit_l14.pth"), "ViT-L/14_336": os.path.join(MODEL_PATH, "vit_l14_336.pth"), }

Andy1621 commented 1 year ago

请参考https://github.com/OpenGVLab/UniFormerV2/blob/main/INSTRUCTIONS.md#training

xiezexun commented 1 year ago

不好意思作者，打扰您了，看了您上面的链接，请问是用的open_clip中的预训练模型吗？地址是：https://github.com/mlfoundations/open_clip/releases/tag/v0.2-weights

Andy1621 commented 1 year ago

用的是openai的clip模型https://github.com/openai/CLIP

xiezexun commented 1 year ago

好的，感谢您的回复，我已经知道了

ZachWong123 commented 10 months ago

您好 请问您方便把您提取出的vit预训练权重分享一下么 我这边由于环境问题 没办法进行CLIP库的安装 非常感谢