inconsistent tensor size, expected tensor [1 x 736 x 8 x 16] and src [1 x 1248 x 8 x 16] to have the same number of elements

[younglululu]

I am making prediction using the trained models. I used the following command:

python prediction_emb.py 100 param/dia_all_epo20_dim24_conv10/dia_all_epo20_dim24_conv10_filled.pt 10 data/mod.txt

And here is the error I encoutered which is reproducible in both python 2.7 and python 3.7, pytorch version 0.4.1

While copying the parameter named digit_capsules.route_weights, whose dimensions in the model are torch.Size([1, 736, 8, 16]) and whose dimensions in the checkpoint are torch.Size([1, 1248, 8, 16]), ...
Traceback (most recent call last):
File "prediction_emb.py", line 78, in
obse,pred1=pred_from_model(conv1,conv1,round1model,RTtest,15)
File "prediction_emb.py", line 17, in pred_from_model
model.load_state_dict(torch.load(param_path))
File "/home/yanglu/anaconda2/envs/dl_env/lib/python2.7/site-packages/torch/nn/modules/module.py", line 360, in load_state_dict
own_state[name].copy_(param)
RuntimeError: inconsistent tensor size, expected tensor [1 x 736 x 8 x 16] and src [1 x 1248 x 8 x 16] to have the same number of elements, but got 94208 and 159744 elements respectively at /opt/conda/conda-bld/pytorch_1503966894950/work/torch/lib/TH/generic/THTensorCopy.c:86

[horsepurve]

Hi, because here "dia" and "mod" are two different data sets with different maximum peptide lengths (66 vs. 50), please simply change line #16 in config.py to be:

max_length = 66

And while making prediction using another trained model, just change this value to be the max peptide length in that data set (e.g. 50 in "mod" data).

Thank you.

[younglululu]

Dear Chunwei, The program can run now. Thanks for the help! Best regards, Yang Lu

…

On Sun, Nov 11, 2018 at 10:11 AM Chunwei Ma ***@***.***> wrote: Hi, because here "dia" and "mod" are two different data sets with different maximum peptide length (66 vs. 50), please simply change line #16 <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_horsepurve_DeepRTplus_blob_master_config.py-23L16&d=DwMFaQ&c=clK7kQUTWtAVEOVIgvi0NU5BOUHhpN0H8p7CSfnc_gI&r=WpU5ZNHyuHh29_wmb9c4AA&m=RykoWwYV5rhfsUCOP2ptcGEOB6Q6VT3fYV5la4ESu5w&s=_REpg1E-VLRSYD0xP_KNpxxniVHXvw9M5Lj4Pu8wkoE&e=> in config.py to be: max_length = 66 And while making prediction using another trained model, just change this value to be the max peptide length in that data set (e.g. 50 in "mod" data). Thank you. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_horsepurve_DeepRTplus_issues_5-23issuecomment-2D437691691&d=DwMFaQ&c=clK7kQUTWtAVEOVIgvi0NU5BOUHhpN0H8p7CSfnc_gI&r=WpU5ZNHyuHh29_wmb9c4AA&m=RykoWwYV5rhfsUCOP2ptcGEOB6Q6VT3fYV5la4ESu5w&s=dvBI7RnktKUqKBaQMCkJqx859igxS1P9jQsN_WfiTtA&e=>, or mute the thread <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_AOmuWoGFR-2DiYfTaw0V5lFUhmgGf1OSBaks5uuGhjgaJpZM4YYUEj&d=DwMFaQ&c=clK7kQUTWtAVEOVIgvi0NU5BOUHhpN0H8p7CSfnc_gI&r=WpU5ZNHyuHh29_wmb9c4AA&m=RykoWwYV5rhfsUCOP2ptcGEOB6Q6VT3fYV5la4ESu5w&s=ppj0223EspZIgl19ycI_uGO3tfbj-qP_7dmH6h90a5Y&e=> .