Wang Yi

Results 41 issues of Wang Yi

## Summary https://github.com/Oneflow-Inc/oneflow/pull/9363 这个 PR 中,我用一个 named_tuple 把返回 TensorTuple 的算子结果包了一层,使其能够跟 torch 对齐: ``` In [1]: import torch In [2]: import oneflow as flow In [3]: x = torch.randn(2, 3, 4)...

bug
community

背景:https://github.com/Oneflow-Inc/OneCloud/issues/147#issuecomment-1195307571 概述: 在 functional_api.yaml 中,支持了一种新的写法: ``` "TensorTuple (Tensor input, Int32 dim, Bool keepdim=False) => Max", ``` 来表明返回值是一个 named tuple,第一个元素的 name 是 values,第二个元素的 name 是 indices 实现: 在生成 pybind 接口的时候,判断 tensortuple...

enhancement
system
api

#close https://github.com/Oneflow-Inc/oneflow/issues/9265

enhancement
bug
test
api

TODO: - [x] 实现 helper functions - [x] 最 naive 的 case:native_multi_head_attention 在 functor 层对齐并实现 - [x] 实现 python functional 层的 mha - [x] 实现 nn.Module 层的 mha - [x]...

enhancement
documentation
api
python

enhancement
automerge
op
api
python

![image](https://user-images.githubusercontent.com/53533850/218691474-39fbcc82-70f4-4e42-909d-bf927ff9a433.png) ![image](https://user-images.githubusercontent.com/53533850/218691595-f3d89b58-767f-4e38-a535-76b8487fc5f1.png)

enhancement
op
test
api

```python3 net.load_state_dict(weight) net = net.eval().cuda() x = torch.ones([1,3,112,112]).cuda() trt_net = ttrt.torch2trt(net, [x], max_batch_size=20) trt_net_fp16 = ttrt.torch2trt(net, [x], max_batch_size=20, fp16_mode=True) ori_output = net(imgs) trt_output = trt_net(imgs) trt_fp16_output = trt_net_fp16(imgs) print('without fp16:...

bug

add feature of saving and loading graph, and related UI previous UI: ![image](https://github.com/siliconflow/onediff/assets/53533850/f6453150-5e8c-4062-b8bb-803df5f878cd) current UI: ![image](https://github.com/siliconflow/onediff/assets/53533850/20ce1746-d99d-487e-9c77-a10d051a4859)