issues
search
Mikolaj
/
horde-ad
Higher Order Reverse Derivatives Efficiently - Automatic Differentiation library based on the paper "Provably correct, asymptotically efficient, higher-order reverse-mode automatic differentiation"
BSD 3-Clause "New" or "Revised" License
34
stars
6
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
In approx tests, HUnitFailure is not caught and not pretty-printed
#60
Mikolaj
closed
2 years ago
22
Voice recognition (fixes #58)
#59
Mikolaj
opened
2 years ago
4
Code an example that recognizes a particular person's speech
#58
Mikolaj
opened
2 years ago
2
implement epsilon-based floating point number comparisons in tests
#57
sfindeisen
closed
2 years ago
32
Replace explicit type application for tensor size arguments with custom type values application
#56
Mikolaj
closed
9 months ago
6
Investigate HaskTorch
#55
Mikolaj
opened
2 years ago
1
Convert Deltas to GADT
#54
tomjaguarpaw
closed
2 years ago
7
Update README.md
#53
awf
closed
2 years ago
2
document the core API of the Engine
#52
sfindeisen
opened
2 years ago
1
improve QuickCheck tests
#51
sfindeisen
closed
2 years ago
10
Implement LSTM with both soft and hard attention
#50
Mikolaj
opened
2 years ago
2
test suite parallelization
#49
sfindeisen
closed
2 years ago
15
Rewrite one of shaped tests not to forget type-level shapes of parameters at each gradient descent update
#48
Mikolaj
closed
2 years ago
0
investigate huge floating point errors in (flaky) tests
#47
sfindeisen
opened
2 years ago
1
implement epsilon-based floating point number comparisons in tests
#46
sfindeisen
closed
2 years ago
17
Code an example with audio recordings of speech, e.g., sentiment analysis
#45
Mikolaj
opened
2 years ago
7
Try to get rid of any of the AllowAmbiguousTypes pragmas
#44
Mikolaj
closed
9 months ago
1
[Feature] I can write code with large tensors, and the derivative runs as fast as PyTorch on GPU
#43
Mikolaj
opened
2 years ago
11
Implement bachnorm layer and add it to some tests
#42
Mikolaj
opened
2 years ago
2
Create bindings to libxsmm (in a separate package)
#41
Mikolaj
opened
2 years ago
2
cabal test takes ages to complete and uses only 2 cores
#40
sfindeisen
closed
2 years ago
2
TODOs in HordeAd.Core.DualNumber about defining more derivatives of common functions
#39
sfindeisen
closed
2 years ago
11
Delayed-outlined primitive numeric functions
#38
Mikolaj
closed
2 years ago
7
Weaken IsDualWithScalar for returnLet
#37
tomjaguarpaw
closed
2 years ago
3
Try traces that contain unrolled syntax trees with many primitive functions instead of delta expressions
#36
Mikolaj
closed
2 years ago
0
Argument record example
#35
tomjaguarpaw
closed
2 years ago
0
How can we reduce type parameters further?
#34
tomjaguarpaw
opened
2 years ago
4
Reduce type params
#33
tomjaguarpaw
closed
2 years ago
0
Merge simplify
#32
tomjaguarpaw
closed
2 years ago
6
Port tomjaguarpaw/simplify
#31
tomjaguarpaw
closed
2 years ago
0
Rename scalar to konst
#30
tomjaguarpaw
closed
2 years ago
3
Rename scalar to constant
#29
tomjaguarpaw
closed
2 years ago
1
Define dual number counterparts of most orthotope operations
#28
Mikolaj
opened
2 years ago
2
Improve QuickCheck tests
#27
Mikolaj
closed
2 years ago
3
Eliminate the explosion of spurious IsScalarS constraints
#26
Mikolaj
closed
2 years ago
6
[Blocker] Don't require dozens of IsScalarS sh r constraints, for all shapes sh used anywhere in the code
#25
Mikolaj
closed
2 years ago
9
How to evaluate the delta expression for convolution?
#24
Mikolaj
closed
2 years ago
51
Unpack r in Delta datatypes via backpack, class or otherwise
#23
Mikolaj
opened
2 years ago
1
Remove helper functions needed just for their types
#22
goldfirere
closed
2 years ago
2
Rename dM_MD2 and related functions
#21
awf
closed
9 months ago
1
[Blocker] Performance has broken down after the last two rewrites
#20
Mikolaj
closed
1 year ago
12
Do TODOs in HordeAd.Core.DualNumber about defining more derivatives of common functions
#19
Mikolaj
closed
2 years ago
1
[blocked on #20] Try sparser and not ad-hoc data structures for the scaling factors
#18
Mikolaj
opened
2 years ago
1
Nest HasDual instances, e.g., `Tensor a`, where `a` can be any HasDual instance
#17
Mikolaj
closed
9 months ago
7
Don't use type variables in patterns
#16
goldfirere
closed
2 years ago
10
Implement efficient foward derivatives and generalize DualNumber.hs definitions to work on them
#15
Mikolaj
closed
2 years ago
12
SPECIALIZE needed despite no warnings by -Wall-missed-specialisations
#14
Mikolaj
closed
1 year ago
36
[Mostly solved; postponed] Make the code with type-level magic more readable and simpler
#13
Mikolaj
closed
1 year ago
4
Find a way to implement mapping and other higher order operations on dual numbers efficiently, despite function space not being a differentiable type
#12
Mikolaj
opened
2 years ago
4
Optimize computing the value (as opposed to the gradient) of a function
#11
Mikolaj
closed
2 years ago
2
Previous
Next