ViT-L transformer

ViT-L是Vision Transformer的一个变体，其中"L"代表"Large"，表示模型的规模较大。根据引用\[2\]中的描述，ViT-L的ViT-Adapter参数数量为23.7M。ViT-L的设计相对简单，它几乎完全复制了Transformer的编码部分，将图像切分成补丁并进行编码，同时添加位置编码以进行分类任务\[3\]。
#### 引用[.reference_title]
- *1* *2* [屠榜语义分割！ViT-Adapter：用于密集预测的视觉Transformer适配器](https://blog.csdn.net/amusi1994/article/details/124938492)[target="_blank" data-report-click={"spm":"1018.2226.3001.9630","extra":{"utm_source":"vip_chatgpt_common_search_pc_result","utm_medium":"distribute.pc_search_result.none-task-cask-2~all~insert_cask~default-1-null.142^v91^insertT0,239^v3^insert_chatgpt"}} ] [.reference_item]
- *3* [Vit-transformers](https://blog.csdn.net/u012193416/article/details/121128715)[target="_blank" data-report-click={"spm":"1018.2226.3001.9630","extra":{"utm_source":"vip_chatgpt_common_search_pc_result","utm_medium":"distribute.pc_search_result.none-task-cask-2~all~insert_cask~default-1-null.142^v91^insertT0,239^v3^insert_chatgpt"}} ] [.reference_item]
[ .reference_list ]