Skip to content

P
PaddleClas

PaddlePaddle / PaddleClas
大约 1 年 前同步成功

通知 115
Star 4999
Fork 1114
P
PaddleClas
提交 3d01ffeb 编写于 作者: C cuicheng01 提交者： cuicheng01
浏览文件
操作

update README.md

上级 412347d5
显示空白变更内容
内联 并排
Showing with 14 addition and 15 deletion
docs/zh_CN/models/ImageNet1k/README.md
浏览文件 @ 3d01ffeb
......@@ -706,21 +706,20 @@ DeiT（Data-efficient Image Transformers）系列模型的精度、速度指标
## NextViT 系列 <sup>[[35](#ref47)]</sup>
关于 NextViT 系列模型的精度、速度指标如下表所示，更多介绍可以参考：[NextViT 系列模型文档](NextViT.md)。
| 模型 | Top-1 Acc | Top-5 Acc | time(ms)<br>bs=1 | time(ms)<br>bs=4 | FLOPs(G) | Params(M) | 预训练模型下载地址 | inference模型下载地>址 |
| ---------- | --------- | --------- | ---------------- | ---------------- | -------- | --------- | ------------------------------------------------------------ | ------------------------------------------------------------ |
| NextViT_small_224 | 0.8248 | 0.9616 | - | - | 5.79 | 31.80 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_small_224_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_small_224_infer.tar) |
| NextViT_base_224 | 0.8324 | 0.9658 | - | - | 8.26 | 44.88 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_base_224_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_base_224_infer.tar) |
| NextViT_large_224 | 0.8363 | 0.9661 | - | - | 10.73 | 57.95 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_large_224_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_large_224_infer.tar) |
| NextViT_small_384 | 0.8401 | 0.9698 | - | - | 17.00 | 31.80 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_small_384_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_small_384_infer.tar) |
| NextViT_base_384 | 0.8465 | 0.9723 | - | - |24.27 | 44.88 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_base_384_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_base_384_infer.tar) |
| NextViT_large_384 | 0.8492 | 0.9728 | - | - | 31.53 | 57.95 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_large_384_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_large_384_infer.tar) |
| NextViT_small_224_ssld | 0.8472 | 0.9734 | - | - | 5.79 | 31.80 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_small_224_ssld_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_small_224_ssld_infer.tar) |
| NextViT_base_224_ssld | 0.8500 | 0.9753 | - | - | 8.26 | 44.88 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_base_224_ssld_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_base_224_ssld_infer.tar) |
| NextViT_large_224_ssld | 0.8536 | 0.9762 | - | - | 10.73 | 57.95 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_large_224_ssld_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_large_224_ssld_infer.tar) |
| NextViT_small_384_ssld | 0.8597 | 0.9790 | - | - | 17.00 | 31.80 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_small_384_ssld_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_small_384_ssld_infer.tar) |
| NextViT_base_384_ssld | 0.8634 | 0.9806 | - | - |24.27 | 44.88 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_base_384_ssld_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_base_384_ssld_infer.tar) |
| NextViT_large_384_ssld | 0.8654 | 0.9814 | - | - | 31.53 | 57.95 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_large_384_ssld_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_large_384_ssld_infer.tar) |
| 模型 | Top-1 Acc | Top-5 Acc | time(ms)<br>bs=1 | time(ms)<br>bs=4 | time(ms)<br/>bs=8 | FLOPs(G) | Params(M) | 预训练模型下载地址 | inference模型下载地址 |
| ---------- | --------- | --------- | ---------------- | ---------------- | -------- | --------- | ------------------------------------------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ |
| NextViT_small_224 | 0.8248 | 0.9616 | - | - | - | 5.79 | 31.80 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_small_224_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_small_224_infer.tar) |
| NextViT_base_224 | 0.8324 | 0.9658 | - | - | - | 8.26 | 44.88 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_base_224_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_base_224_infer.tar) |
| NextViT_large_224 | 0.8363 | 0.9661 | - | - | - | 10.73 | 57.95 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_large_224_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_large_224_infer.tar) |
| NextViT_small_384 | 0.8401 | 0.9698 | - | - | - | 17.00 | 31.80 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_small_384_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_small_384_infer.tar) |
| NextViT_base_384 | 0.8465 | 0.9723 | - | - | - | 24.27 | 44.88 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_base_384_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_base_384_infer.tar) |
| NextViT_large_384 | 0.8492 | 0.9728 | - | - | - | 31.53 | 57.95 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_large_384_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_large_384_infer.tar) |
| NextViT_small_224_ssld | 0.8472 | 0.9734 | - | - | - | 5.79 | 31.80 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_small_224_ssld_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_small_224_ssld_infer.tar) |
| NextViT_base_224_ssld | 0.8500 | 0.9753 | - | - | - | 8.26 | 44.88 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_base_224_ssld_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_base_224_ssld_infer.tar) |
| NextViT_large_224_ssld | 0.8536 | 0.9762 | - | - | - | 10.73 | 57.95 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_large_224_ssld_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_large_224_ssld_infer.tar) |
| NextViT_small_384_ssld | 0.8597 | 0.9790 | - | - | - | 17.00 | 31.80 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_small_384_ssld_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_small_384_ssld_infer.tar) |
| NextViT_base_384_ssld | 0.8634 | 0.9806 | - | - | - | 24.27 | 44.88 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_base_384_ssld_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_base_384_ssld_infer.tar) |
| NextViT_large_384_ssld | 0.8654 | 0.9814 | - | - | - | 31.53 | 57.95 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/NextViT_large_384_ssld_pretrained.pdparams) | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/inference/NextViT_large_384_ssld_infer.tar) |
<a name="Transformer_lite"></a>
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或