Wang Yi
Wang Yi
> bfp16可以考虑补充下。 已支持
> 可以加一下 profile 函数，参考这里 https://github.com/Oneflow-Inc/OneTeam/blob/master/tutorial/howto_test_user_op.md#%E6%80%A7%E8%83%BD%E6%B5%8B%E8%AF%95 已添加
可以把 uniform_ 这个方法也在 Python C api 里面绑定到 tensor.uniform_ 上（不过这样的话，那个 assert 就得放在 functor 里面了）
https://github.com/yl-1993/learn-to-cluster/blob/3b834589923baf72523e288cc462e0df591b99c1/utils/adjacency.py#L8-L17 By the way, what is the purpose of row_normalize, and `# if rowsum
文档的截图可以更新一下
可以按照 oneflow/python/oneflow/test/modules/test_loss.py 这里的代码，加一下 nn.functional.l1_loss / mse_loss 的单测；文档中可以加一下参数列表，不然渲染出来文档oneflow.nn.functional.mse_loss后面只跟了一个空括号。其他没啥问题了
Hi, I tried to do the same thing with you, could you please show me your code?
Maybe you can install torch-2.1.2+cu121 manually, by command `pip install torch==2.1.2 torchvision==0.16.2 torchaudio==2.1.2 --index-url https://download.pytorch.org/whl/cu121`. (And if you are in China, the speed of connecting to this pytorch official website...
maybe you can try `--timeout ` param in pip to prevent timeout issues by setting sec to a large value, reference this: https://stackoverflow.com/questions/59796680/disable-pip-install-timeout-for-slow-connections
you can continue use Python3.12, just install pytorch manually by command like `pip install torch==2.2.0+cu121`, and add param ` --skip-python-version-check` when you start sd-webui service.