训练过程出现的问题

[PanTings]

你好，我开始训练之后出现size mismatch的问题，不知道是什么原因，是torch版本不一样吗？我在电脑上装的1.5.0+cpu；或者是什么其他原因？
2020-10-14 11:25:57 Epoch 0...
Traceback (most recent call last):
File "F:\MyProgram\PyCharm\Neu-Review-Rec-master\main.py", line 213, in
fire.Fire()
File "E:\Program Files (x86)\Python-3.6.8\lib\site-packages\fire\core.py", line 138, in Fire
component_trace = _Fire(component, args, parsed_flag_args, context, name)
File "E:\Program Files (x86)\Python-3.6.8\lib\site-packages\fire\core.py", line 468, in _Fire
target=component.name)
File "E:\Program Files (x86)\Python-3.6.8\lib\site-packages\fire\core.py", line 672, in _CallAndUpdateTrace
component = fn(*varargs, **kwargs)
File "F:\MyProgram\PyCharm\Neu-Review-Rec-master\main.py", line 85, in train
output = model(train_datas)
File "E:\Program Files (x86)\Python-3.6.8\lib\site-packages\torch\nn\modules\module.py", line 550, in call
result = self.forward(*input, **kwargs)
File "F:\MyProgram\PyCharm\Neu-Review-Rec-master\framework\models.py", line 42, in forward
output = self.predict_net(ui_feature, uids, iids).squeeze(1)
File "E:\Program Files (x86)\Python-3.6.8\lib\site-packages\torch\nn\modules\module.py", line 550, in call
result = self.forward(*input, **kwargs)
File "F:\MyProgram\PyCharm\Neu-Review-Rec-master\framework\prediction.py", line 32, in forward
return self.model(feature, uid, iid)
File "E:\Program Files (x86)\Python-3.6.8\lib\site-packages\torch\nn\modules\module.py", line 550, in call
result = self.forward(*input, **kwargs)
File "F:\MyProgram\PyCharm\Neu-Review-Rec-master\framework\prediction.py", line 140, in forward
fm_out = self.build_fm(feature)
File "F:\MyProgram\PyCharm\Neu-Review-Rec-master\framework\prediction.py", line 129, in build_fm
fm_linear_part = self.fc(input_vec)
File "E:\Program Files (x86)\Python-3.6.8\lib\site-packages\torch\nn\modules\module.py", line 550, in call
result = self.forward(*input, **kwargs)
File "E:\Program Files (x86)\Python-3.6.8\lib\site-packages\torch\nn\modules\linear.py", line 87, in forward
return F.linear(input, self.weight, self.bias)
File "E:\Program Files (x86)\Python-3.6.8\lib\site-packages\torch\nn\functional.py", line 1610, in linear
ret = torch.addmm(bias, input, weight.t())
RuntimeError: size mismatch, m1: [128 x 64], m2: [128 x 1] at C:\w\b\windows\pytorch\aten\src\TH/generic/THTensorMath.cpp:41

[ShomyLiu]

你好，能否提供下运行的方法和命令？

[PanTings]

谢谢啦
main.py train --model=MPCN --num_fea=2 --output=fm --use_gpu=False --gpu_id=-1

[PanTings]

另外，output设置成mlp，lfm, nfm都会出现类似的问题，都会在prediction.py、使用self.fc()方法的这一行报错

[ShomyLiu]

你好，MPCN模型的num_fea 应该是1，只用了review-level feature， 所以:

python main.py train --model=MPCN --num_fea=1 --output=fm 

设置为2的时候，应该会报错

[PanTings]

啊 可是我设置1就报错这个
ValueError: the num_fea of MPCN is error, please specific --num_fea=2

[ShomyLiu]

MPCN的num_fea是1，确定是最新版吗:
https://github.com/ShomyLiu/Neu-Review-Rec/blob/master/models/mpcn.py#L21

[PanTings]

哇 非常感谢你的耐心回答，真是num_fea的问题

[xiulingque]

你好，我开始训练之后出现维度不匹配的问题，不知道是什么原因，也没有找到有效的解决办法，请问如何解决这个问题。
load npy from dist...

---

user config:
vocab_size => 50002
word_dim => 300
r_max_len => 202
u_max_r => 13
i_max_r => 24
train_data_size => 51764
test_data_size => 6471
val_data_size => 6471
user_num => 5543
item_num => 3570
batch_size => 128
print_step => 100

---

Traceback (most recent call last):
File "main.py", line 212, in
fire.Fire()
File "/root/.local/lib/python3.7/site-packages/fire/core.py", line 141, in Fire
component_trace = _Fire(component, args, parsed_flag_args, context, name)
File "/root/.local/lib/python3.7/site-packages/fire/core.py", line 471, in _Fire
target=component.name)
File "/root/.local/lib/python3.7/site-packages/fire/core.py", line 681, in CallAndUpdateTrace
component = fn(*varargs, **kwargs)
File "main.py", line 45, in train
model = Model(opt, getattr(models, opt.model))
File "/root/Neu-Review-Rec/framework/models.py", line 17, in init
self.net = Net(opt)
File "/root/Neu-Review-Rec/models/mpcn.py", line 39, in init
self.reset_para()
File "/root/Neu-Review-Rec/models/mpcn.py", line 102, in reset_para
self.user_word_embs.weight.data.copy(w2v.cuda())
RuntimeError: The size of tensor a (50002) must match the size of tensor b (24150) at non-singleton dimension 0

[ShomyLiu]

这个应该是预加载的字典与设置的大小不一样的缘故，检查下，要么修改下voca_size要么修改下字典，试试

[xiulingque]

已经解决了，谢谢大佬回复，辛苦

…

------------------ 原始邮件 ------------------ 发件人: HT Liu ***@***.***> 发送时间: 2022年4月2日 09:17 收件人: ShomyLiu/Neu-Review-Rec ***@***.***> 抄送: xiulingque ***@***.***>, Comment ***@***.***> 主题: Re: [ShomyLiu/Neu-Review-Rec] 训练过程出现的问题 (#15) 这个应该是预加载的字典与设置的大小不一样的缘故，检查下，要么修改下voca_size要么修改下字典，试试 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***>