Closed zhangdongxu closed 8 years ago
Update Theano to the development version.
Le ven. 22 janv. 2016 00:08, zhangdongxu notifications@github.com a écrit :
This error happens when using search mode But program can still run Does it impact the compiling of attention framework?
2016-01-21 17:59:38,114: groundhogtrainerSGD_adadelta: DEBUG: Constructing grad function 2016-01-21 17:59:39,072: groundhogtrainerSGD_adadelta: DEBUG: Compiling grad function ERROR (theanogofopt): SeqOptimizer apply 2016-01-21 18:01:21,598: theanogofopt: ERROR: SeqOptimizer apply ERROR (theanogofopt): Traceback: 2016-01-21 18:01:21,598: theanogofopt: ERROR: Traceback: ERROR (theanogofopt): Traceback (most recent call last): File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python27/site-packages/theano/gof/optpy", line 195, in apply sub_prof = optimizeroptimize(fgraph) File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python27/site-packages/theano/gof/optpy", line 81, in optimize ret = selfapply(fgraph, _args, _kwargs) File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python27/site-packages/theano/tensor/optpy", line 5498, in apply new_outputs = selfoptimizer(node) File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python27/site-packages/theano/tensor/optpy", line 5433, in local_fuse n = OP(C)(_inputs)owner File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python27/site-packages/theano/tensor/elemwisepy", line 496, in init_ scalar_opnout) ValueError: Cannot construct a ufunc with more than 32 operands (requested number were: inputs = 40 and outputs = 1)
2016-01-21 18:01:21,648: theanogofopt: ERROR: Traceback (most recent call last): File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python27/site-packages/theano/gof/optpy", line 195, in apply sub_prof = optimizeroptimize(fgraph) File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python27/site-packages/theano/gof/optpy", line 81, in optimize ret = selfapply(fgraph, _args, _kwargs) File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python27/site-packages/theano/tensor/optpy", line 5498, in apply new_outputs = selfoptimizer(node) File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python27/site-packages/theano/tensor/optpy", line 5433, in local_fuse n = OP(C)(_inputs)owner File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python27/site-packages/theano/tensor/elemwisepy", line 496, in init_ scalar_opnout) ValueError: Cannot construct a ufunc with more than 32 operands (requested number were: inputs = 40 and outputs = 1)
— Reply to this email directly or view it on GitHub https://github.com/lisa-groundhog/GroundHog/issues/44.
Thank you. Development version works.
This error happens when using search mode. But program can still run. Does it impact the compiling of attention framework?
2016-01-21 17:59:38,114: groundhog.trainer.SGD_adadelta: DEBUG: Constructing grad function 2016-01-21 17:59:39,072: groundhog.trainer.SGD_adadelta: DEBUG: Compiling grad function ERROR (theano.gof.opt): SeqOptimizer apply <theano.tensor.opt.FusionOptimizer object at 0x3717f50> 2016-01-21 18:01:21,598: theano.gof.opt: ERROR: SeqOptimizer apply <theano.tensor.opt.FusionOptimizer object at 0x3717f50> ERROR (theano.gof.opt): Traceback: 2016-01-21 18:01:21,598: theano.gof.opt: ERROR: Traceback: ERROR (theano.gof.opt): Traceback (most recent call last): File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python2.7/site-packages/theano/gof/opt.py", line 195, in apply sub_prof = optimizer.optimize(fgraph) File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python2.7/site-packages/theano/gof/opt.py", line 81, in optimize ret = self.apply(fgraph, _args, _kwargs) File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python2.7/site-packages/theano/tensor/opt.py", line 5498, in apply new_outputs = self.optimizer(node) File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python2.7/site-packages/theano/tensor/opt.py", line 5433, in local_fuse n = OP(C)(inputs).owner File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python2.7/site-packages/theano/tensor/elemwise.py", line 496, in init scalar_op.nout) ValueError: Cannot construct a ufunc with more than 32 operands (requested number were: inputs = 40 and outputs = 1)
2016-01-21 18:01:21,648: theano.gof.opt: ERROR: Traceback (most recent call last): File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python2.7/site-packages/theano/gof/opt.py", line 195, in apply sub_prof = optimizer.optimize(fgraph) File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python2.7/site-packages/theano/gof/opt.py", line 81, in optimize ret = self.apply(fgraph, _args, _kwargs) File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python2.7/site-packages/theano/tensor/opt.py", line 5498, in apply new_outputs = self.optimizer(node) File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python2.7/site-packages/theano/tensor/opt.py", line 5433, in local_fuse n = OP(C)(inputs).owner File "/nfs/nlphome/nlp/tools/python27/root/usr/lib64/python2.7/site-packages/theano/tensor/elemwise.py", line 496, in init scalar_op.nout) ValueError: Cannot construct a ufunc with more than 32 operands (requested number were: inputs = 40 and outputs = 1)