Thank you very much! I had a similar problem. I just subtracted one from
all the y values and now it works fine.
vineri, 16 august 2013, 22:31:50 UTC+3, Pascal Lamblin a scris:
>
> Hi Petros,
>
> On Fri, Aug 16, 2013, Petros Ypsilantis wrote:
> > I use data from matlab, despite the fact that I have managed to
> transform
> > my data in order to be theano readable and I don't have any problem
> during
> > the pre-training
> >
> > procedure of the dbn.py . During the fine-tuning I take the following
> > massage:
> >
> >
> > ValueError: y_i value out of bounds
>
> It looks like you built a model with two output classes, which would be
> 0 and 1, but some data examples have a different label, hence the "out
> of bounds" message.
>
> Is it possible that your targets are 1 and 2, instead of 0 and 1? I
> think that matlab starts indexing at 1, but Python starts at 0, so that
> might be the issue.
>
> Since the target is not involved in the pre-training, that would explain
> why the issue only appears during fine-tuning.
>
> Hope this helps,
>
> > Apply node that caused the error:
> > CrossentropySoftmaxArgmax1HotWithBias(_dot22.0, b,
> Elemwise{Cast{int32}}.0)
> > Inputs shapes: [(10, 2), (2,), (10,)]
> > Inputs strides: [(16, 8), (8,), (4,)]
> > Inputs types: [TensorType(float64, matrix), TensorType(float64, vector),
> > TensorType(int32, vector)]
> > Use the Theano flag 'exception_verbosity=high' for a debugprint of this
> > apply node.
> >
> >
> >
> > When I run again dbn.py using (THEANO_FLAGS="exception_verbosity=high")
> I
> > take the following error:
> >
> >
> >
> > ValueError: y_i value out of bounds
> > Apply node that caused the error:
> > CrossentropySoftmaxArgmax1HotWithBias(_dot22.0, b,
> Elemwise{Cast{int32}}.0)
> > Inputs shapes: [(10, 2), (2,), (10,)]
> > Inputs strides: [(16, 8), (8,), (4,)]
> > Inputs types: [TensorType(float64, matrix), TensorType(float64, vector),
> > TensorType(int32, vector)]
> > Debugprint of the apply node:
> > CrossentropySoftmaxArgmax1HotWithBias.0 [@A] <TensorType(float64,
> vector)>
> > ''
> > |_dot22 [@B] <TensorType(float64, matrix)> ''
> > | |Elemwise{Composite{[scalar_sigmoid(add(i0, i1))]}}[(0, 0)] [@C]
> > <TensorType(float64, matrix)> ''
> > | | |_dot22 [@D] <TensorType(float64, matrix)> ''
> > | | | |Elemwise{Composite{[scalar_sigmoid(add(i0, i1))]}}[(0, 0)] [@E]
> > <TensorType(float64, matrix)> ''
> > | | | | |_dot22 [@F] <TensorType(float64, matrix)> ''
> > | | | | | |Elemwise{Composite{[scalar_sigmoid(add(i0, i1))]}}[(0, 0)]
> [@G]
> > <TensorType(float64, matrix)> ''
> > | | | | | | |_dot22 [@H] <TensorType(float64, matrix)> ''
> > | | | | | | | |Subtensor{int64:int64:} [@I] <TensorType(float64,
> matrix)>
> > ''
> > | | | | | | | | |<TensorType(float64, matrix)> [@J]
> <TensorType(float64,
> > matrix)>
> > | | | | | | | | |ScalarFromTensor [@K] <int64> ''
> > | | | | | | | | | |Elemwise{mul,no_inplace} [@L] <TensorType(int64,
> > scalar)> ''
> > | | | | | | | | | |TensorConstant{10} [@M] <TensorType(int64,
> scalar)>
> > | | | | | | | | | |index [@N] <TensorType(int64, scalar)>
> > | | | | | | | | |ScalarFromTensor [@O] <int64> ''
> > | | | | | | | | |Elemwise{Composite{[mul(i0, add(i1, i2))]}} [@P]
> > <TensorType(int64, scalar)> ''
> > | | | | | | | | |TensorConstant{10} [@M] <TensorType(int64,
> scalar)>
> > | | | | | | | | |TensorConstant{1} [@Q] <TensorType(int64, scalar)>
> > | | | | | | | | |index [@N] <TensorType(int64, scalar)>
> > | | | | | | | |W [@R] <TensorType(float64, matrix)>
> > | | | | | | |InplaceDimShuffle{x,0} [@S] <TensorType(float64, row)> ''
>
> > | | | | | | |b [@T] <TensorType(float64, vector)>
> > | | | | | |W [@U] <TensorType(float64, matrix)>
> > | | | | |InplaceDimShuffle{x,0} [@V] <TensorType(float64, row)> ''
> > | | | | |b [@W] <TensorType(float64, vector)>
> > | | | |W [@X] <TensorType(float64, matrix)>
> > | | |InplaceDimShuffle{x,0} [@Y] <TensorType(float64, row)> ''
> > | | |b [@Z] <TensorType(float64, vector)>
> > | |W [@BA] <TensorType(float64, matrix)>
> > |b [@BB] <TensorType(float64, vector)>
> > |Elemwise{Cast{int32}} [@BC] <TensorType(int32, vector)> ''
> > |Subtensor{int64:int64:} [@BD] <TensorType(float32, vector)> ''
> > |<TensorType(float32, vector)> [@BE] <TensorType(float32, vector)>
> > |ScalarFromTensor [@K] <int64> ''
> > |ScalarFromTensor [@O] <int64> ''
> > CrossentropySoftmaxArgmax1HotWithBias.1 [@A] <TensorType(float64,
> matrix)>
> > ''
> > CrossentropySoftmaxArgmax1HotWithBias.2 [@A] <TensorType(int32, vector)>
> ''
> >
> >
> >
> >
> > Does anybody have an idea what is the problem?
> >
> >
> > Thanks
> >
> > Petros
> >
> > --
> >
> > ---
> > You received this message because you are subscribed to the Google
> Groups "theano-users" group.
> > To unsubscribe from this group and stop receiving emails from it, send
> an email to [email protected] <javascript:>.
> > For more options, visit https://groups.google.com/groups/opt_out.
>
> --
> Pascal
>
--
---
You received this message because you are subscribed to the Google Groups
"theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
