jittor.nn.Embedding的文档中说有padding_idx，但并没有实现

[PhyllisJi]

Describe the bug

jittor 1.3.7没有实现padding_idx

Full Log

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[5], line 61
     54     y = m(x)
     55     return list(y.shape)
---> 61 go()

Cell In[5], line 53, in go()
     51 jittor.flags.use_cuda = 1
     52 x = jittor.randn([1, 3, 224, 224])
---> 53 m = alexnet()
     54 y = m(x)
     55 return list(y.shape)

Cell In[5], line 27, in alexnet.__init__(self)
     25 self.pool3_mutated = jittor.nn.MaxPool2d(kernel_size=2, stride=4, padding=4, ceil_mode=True, return_indices=False)
     26 self.avgpool_mutated = jittor.nn.ReflectionPad2d(padding=1)
---> 27 self.flatten_mutated = jittor.nn.Embedding(embedding_dim=1, num_embeddings=5, padding_idx=8)

TypeError: __init__() got an unexpected keyword argument 'padding_idx'

Minimal Reproduce

import os
os.environ["disable_lock"] = "1"
import jittor
import jittor.nn as nn
import jittor.optim as optim
import numpy as np
import copy

class alexnet(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1_mutated = jittor.nn.ConvTranspose2d(in_channels=3, kernel_size=11, out_channels=64)
        self.relu1_mutated = jittor.nn.Softmax()
        self.pool1_mutated = jittor.nn.ReplicationPad2d(padding=8)
        self.conv2_mutated = jittor.nn.PixelShuffle(upscale_factor=1)
        self.relu2_mutated = jittor.nn.PReLU()
        self.pool2_mutated = jittor.nn.MaxPool2d(kernel_size=3, stride=2, return_indices=False, ceil_mode=True)
        self.conv3_mutated = jittor.nn.Conv2d(in_channels=64, out_channels=384, kernel_size=3, padding=1, stride=8, groups=1, bias=False, dilation=(1, 1))
        self.relu3_mutated = jittor.nn.ReLU()
        self.conv4_mutated = jittor.nn.Sigmoid()
        self.relu4_mutated = jittor.nn.ReLU6()
        self.conv5_mutated = jittor.nn.AvgPool2d(kernel_size=(7, 1), stride=(2, 4))
        self.relu5_mutated = jittor.nn.ReLU()
        self.pool3_mutated = jittor.nn.MaxPool2d(kernel_size=2, stride=4, padding=4, ceil_mode=True, return_indices=False)
        self.avgpool_mutated = jittor.nn.ReflectionPad2d(padding=1)
        self.flatten_mutated = jittor.nn.Embedding(embedding_dim=1, num_embeddings=5, padding_idx=8)

    def execute(self, x):
        x = self.conv1_mutated(x)
        x = self.relu1_mutated(x)
        x = self.pool1_mutated(x)
        x = self.conv2_mutated(x)
        x = self.relu2_mutated(x)
        x = self.pool2_mutated(x)
        x = self.conv3_mutated(x)
        x = self.relu3_mutated(x)
        x = self.conv4_mutated(x)
        x = self.relu4_mutated(x)
        x = self.conv5_mutated(x)
        x = self.relu5_mutated(x)
        x = self.pool3_mutated(x)
        x = self.avgpool_mutated(x)
        x = self.flatten_mutated(x)
        return x

def go():
    jittor.flags.use_cuda = 1
    x = jittor.randn([1, 3, 224, 224])
    m = alexnet()
    y = m(x)
    return list(y.shape)

go()

[luozhiya]

jittor.nn.Embedding 的 padding_idx 参数需要版本 >=1.3.8.0，官方文档现在是基于 1.3.9.2，建议您升级版本再试一下。

• jittor.nn — Jittor 1.3.9.2 文档
• LLM training tuning · Jittor/jittor@f26fdfe

[PhyllisJi]

jittor.nn.Embedding 的 padding_idx 参数需要版本 >=1.3.8.0，官方文档现在是基于 1.3.9.2，建议您升级版本再试一下。

• jittor.nn — Jittor 1.3.9.2 文档
• LLM training tuning · Jittor/jittor@f26fdfe

由于一些原因我现在无法升级版本，我从哪里可以查看以前版本的文档？

[luozhiya]

@PhyllisJi 官网文档没找到版本切换，如果需要查看以前的文档，也可以看 API 对应代码注释（文档是自动从代码中生成的）

[PhyllisJi]

@PhyllisJi 官网文档没找到版本切换，如果需要查看以前的文档，也可以看 API 对应代码注释（文档是自动从代码中生成的）

再请问一下，为什么文档里没有jittor.nn.AdaptiveMaxPool3d？ https://github.com/Jittor/jittor/issues/484

[luozhiya]

大概是文档生成模板中没有添加 AdaptiveMaxPool3d 引起的

597

https://cg.cs.tsinghua.edu.cn/jittor/assets/docs/_modules/jittor/pool.html#AdaptiveMaxPool3d

class AdaptiveMaxPool3d(Module):
    '''
    对输入进行三维自适应平均池化处理的类。

        参数: 
            - output_size (int, tuple, list) : 期望的输出形状。
            - return_indices (bool): 若为True, 则将最大值索引值和输出一起返回。

        形状:
            - 输入: :math:`[N, C, D, H, W]`
            - 输出: :math:`[N, C, S_0, S_1, S_2]`, 此处  (S_0, S_1, S_2) = ``output_size`` 。

        属性:
            - output_size (int, tuple, list) : 期望的输出形状。
            - return_indices (bool) : 若为True, 则将最大值索引值和输出一起返回。

        代码示例:
            >>> # target output size of 5x7x9
            >>> m = nn.AdaptiveMaxPool3d((5, 7, 9))
            >>> input = jt.randn(1, 64, 8, 9, 10)
            >>> output = m(input)
            >>> # target output size of 7x7x7 (cube)
            >>> m = nn.AdaptiveMaxPool3d(7)
            >>> input = jt.randn(1, 64, 10, 9, 8)
            >>> output = m(input)
            >>> # target output size of 7x9x8
    '''