Skip to contents

Theano is an open source project powered by Assembla

Assembla offers free public and private SVN/Git repositories and project hosting with bug/issue tracking and collaboration tools.

 

Active by Milestone

No Milestone

# Summary Assigned to Status

495

rename inv -> recip
james.bergstra
New

385

storing tensor types in str
 
New

401

conv for seperable kernels
 
New

402

conv unrolling should be automatic (and self-optimizing)
 
New

408

conv with vsipl
 
New

412

compiling graphs of sparse is slow. write a c version of {{{SparseConstantSignature.__eq__}}}
 
New

423

a walkthrough example of how to code something simple in C, using ndarrays.
none
New

438

faster math.h functions with -mveclibabi
 
New

445

n_channels in ConvOp
 
New

456

unflip filters to be more c_contiguous
 
New

458

conv_full_patch_split
none
New

459

make conv op read/write coalesced by making data aligned
none
New

462

conv_rows, implement the nb_rows in one block to higher occupency,
none
New

466

Optimize reduce for long reductions
 
New

470

CopyFromCudaNdarray needs general version
none
New

473

Add typenum, support for all dtypes
none
New

474

recurse on givens argument to function()
 
New

477

document Op.nin and Op.nout
none
New

480

Benchmark: others
 
New

489

note the use of transpose v in benchmark
james.bergstra
New

494

constant shape at build time
none
New

520

add learn rate annealing to deep learning tutorials
 
New

533

Reorganize functions in tensor and methods of _py_tensor_operators
none
New

534

Implement the garbage collector for ProfileMode
 
New

535

DebugMode should implement garbage collection
 
New

562

Review skipped tests
 
New

571

clean all compiledir in the base_compiledir
 
New

582

Follow-up to #573: implement and write tests for io.In(...borrow=True)
 
New

584

CURAND for random streams
none
New

585

CUSPARSE for GPU sparse Ops
none
New

651

Add theano.tensor.not_
 
New

653

make gemv and gemm class and optimization work for complex* type
 
New

704

parallel compilation for multi-cpu machines
 
New

705

are deep learning tutorials importable?
 
New

706

wish: "package mode"
 
New

708

scikits.datasets compatibility
 
New

784

Ticket for testing email alerts
 
New

789

Faster toposort
 
New

798

Lower total memory usage of Theano function
 
New

809

test ticket
 
New

378

refactor nnet.prepend_<X>_to_each_row
 
New

460

Should we unroll the full patch multiplication when the image and kernel fit in shared memory?
none
New

461

all GpuConvOp kernel should allow non c_contiguous memory access
none
New

464

GpuConvOp thread block multiple of 32?
none
New

465

malloc3d
none
New

467

Add Conv logical input shapes
none
New

471

Explore subclass-ability
none
New

492

create a c_code_signature
 
New

497

Break optimization loop in local_mul_canonizer in a cleaner way
 
New

510

use the warnings module
 
New

511

grad of min,max,max_and_argmax with multiple axis not implemented
 
New

515

Finish the Images2Neibs ops
 
New

545

Implement tensor.jacobian
 
Accepted

554

In object should have a shape parameter
nouiz
assigned

555

assert_shape function
nouiz
assigned

556

remove image_shape, and filter_shape from the interface conv2d().
 
New

581

CpuConv optimization by duplicating memory and calling gemm.
 
New

587

Add theano.len to compute a Variable's symbolic length
 
New

589

Update tests so they do not print deprecation warnings
 
New

599

Implement neibs2image support for neib_step and wrap_centered mode
 
New

603

finish the new convolution with fft
 
New

609

Merge the verify_grad and multiple_outputs_numeric_grad
 
New

703

better documentation for "numpy backend"
none
New

440

numeric_grad doesn't work with functions that have scalars as input
 
New
# Summary Assigned to Status

188

describe CodeGeneration (for users)
bergstra
Accepted

222

ban None
lamblin
assigned

230

unit tests for optdb
 
New

251

greedy_distributor confuses rows and vectors
 
New

296

Optimization coverage
 
New

297

Nightly graph optimization coverage
none
New

294

User evaluation script
none
New
# Summary Assigned to Status

167

Track destroy_map and view_map in OpFromGraph
 
New

367

Doc warning boxes should not be on live site
none
New

432

document, test, organize graph optimization.
none
New

121

grad attribute error
 
New

197

Review error messages
 
New

209

Un-hardcode the number of splits in Split
 
New

237

gemm cases
 
New

285

Pickling can sometimes include old code
 
New

289

Verbose mode for compile.function
 
New

304

Refactor OptCheck
 
New

334

complex number operators cause compilation failure.
james.bergstra
New

337

rename Generic -> PythonType
 
New

342

Prettier pprint
 
New

353

Speed up the DestroyHandler
 
New

358

Cache sphinx docs for local builds
none
New

365

Strict mode
 
New

368

Documentation for optimizations and Env
none
New

436

cmodule take too long on NFS to find all old precomputed file
 
New

439

new check in DEBUG_MODE for CLinker.cmodule_key()
 
New

452

The mode FAST_COMPILE should include the stability optimisation
 
New

131

theano.codegen([inputs], [outputs])
 
New

166

OpFromCallable
 
New

199

fuzz testing
 
New

220

Automated testing of graph pickling
 
New

351

DebugMode catch uninitialized storage
 
New

356

Tensor vs. Scalar
 
New

357

Version'ed documentation
none
New

380

Upgrade scalar.test.test_basic
 
New

433

OpWiseCLinker(FAST_RUN) have allow_gc=True, by default, change
 
New