[Torch5-bugs] [ torch5-Bugs-1839716 ] BLAS problem with 1-dim tensors?

[SourceForge.net <no...@so...>]

Bugs item #1839716, was opened at 2007-11-27 19:24
Message generated for change (Comment added) made by andresy
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=905598&aid=1839716&group_id=183526

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
>Status: Closed
>Resolution: Fixed
Priority: 5
Private: No
Submitted By: thespermwhale (thespermwhale)
Assigned to: Raoul d'Andresy (andresy)
Summary: BLAS problem with 1-dim tensors?

Initial Comment:
I get this:
lda must be >= MAX(M,1): lda=1 M=2Parameter 7 to routine cblas_dgemv was incorrect
[quits out of lua]
 
when I run this:

require "torch"
require "nn"
mlp= nn.Sequential();
mlp:add(nn.Linear(1,2))
x=torch.Tensor(1)
print(mlp:forward(x))

Without BLAS, it works.


----------------------------------------------------------------------

>Comment By: Raoul d'Andresy (andresy)
Date: 2007-11-27 20:46

Message:
Logged In: YES 
user_id=1651867
Originator: NO

this was because blas relies on the stride of the matrix (in the second
dimension), even if the number of column is 1!! strange, i would say. but
corrected.


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=905598&aid=1839716&group_id=183526