Arraymancer v0.7.0 Jul. 4 2021 "Memories of Ice"

This release is named after "Memories of Ice" (2001), the third book of Steven Erikson epic dark fantasy masterpiece "The Malazan Book of the Fallen".

Changes :

• Add toUnsafeView as replacement of dataArray to return a ptr UncheckedArray
• Doc generation fixes
• cumsum, cumprod
• Fix least square solver
• Fix mean square error backpropagation
• Adapt to upstream symbol resolution changes
• Basic Graph Convolution Network
• Slicing tutorial revamp
• n-dimensional tensor pretty printing
• Compilation fixes to handle Nim v1.0 to Nim devel

Thanks to @Vindaar for maintaining the repo, the docs, pretty-printing and answering many many questions on Discord while I took a step back. Thanks to @filipeclduarte for the cumsum/cumprod, @Clonkk for updating raw data accesses, @struggle for finding a bug in mean square error backprop, @timotheecour for spreading new upstream requirements downstream and @anon767 for Graph Neural Network.

v0.6.0 only had v0.5.2 in nimble file

Arraymancer v0.6.0 - Windwalkers

This release is named after "Windwalkers" (2004, French "La Horde du Contrevent"), by Alain Damasio.

Changes:

• The symeig proc to compute eigenvectors of a symmetric matrix now accepts an "uplo" char parameter. This allows to fill only the Upper or Lower part of the matrix, the other half is not used in computation.
• Added svd_randomized, a fast and accurate SVD approximation via random sampling. This is the standard driver for large scale SVD applications as SVD on large matrices is very slow.
• pca now uses the randomized SVD instead of computing the covariance matrix. It can now efficiently deal with large scale problems. It now accepts a center, n_oversamples and n_power_iters arguments. Note that pca without centering is equivalent to a truncated SVD.
• LU decomposition has been added
• QR decomposition has been added
• hilbert has been introduced. It creates the famous ill-conditioned Hilbert matrix. The matrix is suitable to stress test decompositions.
• The arange procedure has been introduced. It creates evenly spaced value within a specified range and step
• The ordering of arguments to error functions has been converted to (y_pred, y_target) (from (y_target, y_pred)), enabling the syntax y_pred.accuracy_score(y). All existing error functions in Arraymancer were commutative w.r.t. to arguments so existing code will keep working.
• A solve procedure has been added to solve linear system of equations represented as matrices.
• A softmax layer has been added to the autograd and neural networks complementing the SoftmaxCrossEntropy layer which fused softmax + Negative-loglikelihood.
• The stochastic gradient descent now has a version with Momentum

Bug fixes:

• gemm could crash when the result was column major.
• The automatic fusion of matrix multiplication with matrix addition (A * X) + b could update the b matrix.
• Complex converters do not pollute the global namespace and do not prevent string conversion via $ of number types due to ambiguous call.
• in-place division has been fixed, a typo made it into substraction.
• A conflict between NVIDIA "nanosecond" and Nim times module "nanosecond" preventing CUDA compilation has been fixed

Breaking:

• In symeig, the eigenvectors argument is now called return_eigenvectors.
• In symeig with slice, the new uplo precedes the slice argument.
• pca input "nb_components" has been renamed "n_components".
• pca output tuple used the names (results, components). It has been renamed to (projected, components).
• A pca overload that projected a data matrix on already existing principal axes was removed. Simply multiply the mean-centered data matrix with the components instead.
• Complex converters were removed. This prevents hard to debug and workaround implicit conversion bug in downstream library. If necessary, users can re-implement converters themselves. This also provides a 20% boost in Arraymancer compilation times

Deprecation:

• The syntax gemm(A, B, C) is now deprecated. Use explicit "gemm(1.0, A, B, 0.0, C)" instead. Arguably not zero-ing C could also be a reasonable default.
• The dot in broadcasting and elementwise operators has changed place Use +., *., /., -., ^., +.=, *.=, /.=, -.=, ^.= instead of the previous order .+ and .+=. This allows the broadcasting operators to have the same precedence as the natural operators. This also align Arraymancer with other Nim packages: Manu and NumericalNim

Thanks to @dynalagreen for the SGD with Momentum, @xcokazaki for spotting the in-place division typo, @Vindaar for fixing the automatic matrix multiplication and addition fusion, @Imperator26 for the Softmax layer, @brentp for reviewing and augmenting the SVD and PCA API, @auxym for the linear equation solver and @berquist for the reordering all error functions to the new API. Thanks @b3liever for suggesting the dot change to solve the precedence issue in broadcasting and elementwise operators.

Fix packaging bug in ".nimble" and updated readme/changelog

Changes affecting backward compatibility:

• None

Changes:

• 0.20.x compatibility (commit 0921190)
• Complex support
• Einsum
• Naive whitespace tokenizer for NLP
• Preview of Laser backend for matrix multiplication without SIMD autodetection (already 5x faster on integer matrix multiplication)

Fix:

• Fix height/width order when reading an image in tensor

Thanks to @chimez for the complex support and updating for 0.20, @metasyn for the tokenizer, @xcokazaki for the image dimension fix and @Vindaar for the einsum implemention

This release is named after "Sign of the Unicorn" (1975), the third book of Roger Zelazny masterpiece "The Chronicles of Amber".

Changes affecting backward compatibility:

• PCA has been split into 2
  • The old PCA with input pca(x: Tensor, nb_components: int) now returns a tuple of result and principal components tensors in descending order instead of just a result
  • A new PCA pca(x: Tensor, principal_axes: Tensor) will project the input x on the principal axe supplied

Changes:

• Datasets:

  • MNIST is now autodownloaded and cached
  • Added IMDB Movie Reviews dataset

• IO:

  • Numpy file format support
  • Image reading and writing support (jpg, bmp, png, tga)
  • HDF5 reading and writing

• Machine learning

  • Kmeans clustering

• Neural network and autograd:

  • Support substraction, sum and stacking in neural networks
  • Recurrent NN: GRUCell, GRU and Fused Stacked GRU support
  • The NN declarative lang now supports GRU
  • Added Embedding layer with up to 3D input tensors [batch_size, sequence_length, features] or [sequence_length, batch_size, features]. Indexing can be done with any sized integers, byte or chars and enums.
  • Sparse softmax cross-entropy now supports target tensors with indices of type: any size integers, byte, chars or enums.
  • Added ADAM optimiser (Adaptative Moment Estimation)
  • Added Hadamard product backpropagation (Elementwise matrix multiply)
  • Added Xavier Glorot, Kaiming He and Yann Lecun weight initialisations
  • The NN declarative lang automatically initialises weights with the following scheme:
    • Linear and Convolution: Kaiming (suitable for Relu activation)
    • GRU: Xavier (suitable for the internal tanh and sigmoid)
    • Embedding: Not supported in declarative lang at the moment

• Tensors:

  • Add tensor splitting and chunking
  • Fancy indexing via index_select
  • division broadcasting, scalar division and multiplication broadcasting
  • High-dimensional toSeq exports

• End-to-end Examples:

  • Sequence/mini time-series classification example using RNN
  • Training and text generation example with Shakespeare and Jane Austen style. This can be applied to any text-based dataset (including blog posts, Latex papers and code). It should contain at least 700k characters (0.7 MB), this is considered small already.

• Important fixes:

  • Convolution shape inference on non-unit strided convolutions
  • Support the future OpenMP changes from nim#devel
  • GRU: inference was squeezing all singleton dimensions instead of just the "layer" dimension.
  • Autograd: remove pointers to avoid pointing to wrong memory when the garbage collector moves it under pressure. This unfortunately comes at the cost of more GC pressure, this will be addressed in the future.
  • Autograd: remove all methods. They caused issues with generic instantiation and object variants.

Special thanks to @metasyn (MNIST caching, IMDB dataset, Kmeans) and @Vindaar (HDF5 support and the example of using Arraymancer + Plot.ly) for their large contributions on this release.

Ecosystem:

• Using Arraymancer + Plotly for NN training visualisation: https://github.com/Vindaar/NeuralNetworkLiveDemo

• Monocle, proof-of-concept data visualisation in Nim using Vega. Hopefully allowing this kind of visualisation in the future:

  

  and compatibility with the Vega ecosystem, especially the Tableau-like Voyager.

• Agent Smith, reinforcement learning framework. Currently it wraps the Arcade Learning Environment for practicing reinforcement learning on Atari games. In the future it will wrap Starcraft 2 AI bindings and provides a high-level interface and examples to reinforcement learning algorithms.

• Laser, the future Arraymancer backend which provides:

  • SIMD intrinsics
  • OpenMP templates with fine-grained control
  • Runtime CPU features detection for ARM and x86
  • A proof-of-concept JIT Assembler
  • A raw minimal tensor type which can work as a view to arbitrary buffers
  • Loop fusion macros for iteration on an arbitrary number of tensors. As far as I know it should provide the fastest multi-threaded iteration scheme on strided tensors all languages and libraries included.
  • Optimized reductions, exponential and logarithm functions reaching 4x to 10x the speed of naively compiled for loops
  • Optimised parallel strided matrix multiplication reaching 98% of OpenBLAS performance
    • This is a generic implementation that can also be used for integers
    • It will support preprocessing (relu_backward, tanh_backward, sigmoid_backward) and epilogue (relu, tanh, sigmoid, bias addition) operation fusion to avoid looping an extra time with a memory bandwidth bound pass.
  • Convolutions will be optimised with a preprocessing pass fused into matrix multiplication. Traditional im2col solutions can only reach 16% of matrix multiplication efficiency on the common deep learning filter sizes
  • State-of-the art random distributions and random sampling implementations for stochastic algorithms, text generation and reinforcement learning.

Future breaking changes.

1. Arraymancer backend will switch to Laser for next version. Impact:

   • At a low-level CPU tensors will become a view on top of a pointer+len fon old data types instead of using the default Nim seqs. This will enable plenty of no-copy use cases and even using memory-mapped tensors for out-of-core processing. However libraries relying on teh very low-level representation of tensors will break. The future type is already implemented in Laser.
   • Tensors of GC-allocated types like seq, string and references will keep using Nim seqs.
   • While it was possible to use the Javascript backend by modifying the iteration scheme this will not be possible at all. Use JS->C FFI or WebAssembly compilation instead.
   • The inline iteration templates map_inline, map2_inline, map3_inline, apply_inline, apply2_inline, apply3_inline, reduce_inline, fold_inline, fold_axis_inline will be removed and replace by forEach and forEachStaged with the following syntax:

   forEach x in a, y in b, z in c:
     x += y * z
   

   Both will work with an arbitrary number of tensors and will generate 2x to 3x more compact code wile being about 30% more efficient for strided iteration. Furthermore forEachStaged will allow precise control of the parallelisation strategy including pre-loop and post-loop synchronisation with thread-local variables, locks, atomics and barriers. The existing higer-order functions map, map2, apply, apply2, fold, reduce will not be impacted. For small inlinable functions it will be recommended to use the forEach macro to remove function call overhead (Yyou can't inline a proc parameter).

2. The neural network domain specific language will use less magic for the forward proc. Currently the neural net domain specific language only allows the type Variable[T] for inputs and the result. This prevents its use with embedding layers which also requires an index input. Furthermore this prevents using tuple[output, hidden: Variable] result type which is very useful to pass RNNs hidden state for generative neural networks (for example text sequence or time-series). So unfortunately the syntax will go from the current forward x, y: shortcut to classic Nim proc forward[T](x, y: Variable[T]): Variable[T]

3. Once CuDNN GRU is implemented, the GRU layer might need some adjustments to give the same results on CPU and Nvidia's GPU and allow using GPU trained weights on CPU and vice-versa.

Thanks:

• metasyn: Datasets and Kmeans clustering
• vindaar: HDF5 support and Plot.ly demo
• bluenote10: toSeq exports
• andreaferetti: Adding axis parameter to Mean layer autograd
• all the contributors of fixes in code and documentation
• the Nim community for the encouragements

This release is named after "The Name of the Wind" (2007), the first book of Patrick Rothfuss masterpiece "The Kingkiller Chronicle".

Changes:

• Core:

  • OpenCL tensors are now available! However Arraymancer will naively select the first backend available. It can be CPU, it can be GPU. They support basic and broadcasted operations (Addition, matrix multiplication, elementwise multiplication, ...)
  • Addition of an argmax and argmax_max procs.

• Datasets:

  • Loading the MNIST dataset from http://yann.lecun.com/exdb/mnist/
  • Reading and writing from CSV

• Linear algebra:

  • Least squares solver
  • Eigenvalues and eigenvectors decomposition for symmetric matrices

• Machine Learning

  • Principal Component Analysis (PCA)

• Statistics

  • Computation of covariance matrices

• Neural network

  • Introduction of a short intuitive syntax to build neural networks! (A blend of Keras and PyTorch).
  • Maxpool2D layer
  • Mean Squared Error loss
  • Tanh and softmax activation functions

• Examples and tutorials

  • Digit recognition using Convolutional Neural Net
  • Teaching Fizzbuzz to a neural network

• Tooling

  • Plotting tensors through Python

Several updates linked to Nim rapid development and several bugfixes.

This release is named after "Wizard's First Rule" (1994), the first book of Terry Goodkind masterpiece "The Sword of Truth".

I am very excited to announce the third release of Arraymancer which includes numerous improvements, features and (unfortunately!) breaking changes.

Warning ⚠: ALL deprecated procs will be removed next release due to deprecated spam and to reduce maintenance burden.

Changes:

• Very Breaking

  • Tensors uses reference semantics now: let a = b will share data by default and copies must be made explicitly.
    • There is no need to use unsafe proc to avoid copies especially for slices.
    • Unsafe procs are deprecated and will be removed leading to a smaller and simpler codebase and API/documentation.
    • Tensors and CudaTensors now works the same way.
    • Use clone to do copies.
    • Arraymancer now works like Numpy and Julia, making it easier to port code.
    • Unfortunately it makes it harder to debug unexpected data sharing.

• Breaking (?)

  • The max number of dimensions supported has been reduced from 8 to 7 to reduce cache misses. Note, in deep learning the max number of dimensions needed is 6 for 3D videos: [batch, time, color/feature channels, Depth, Height, Width]

• Documentation

  • Documentation has been completely revamped and is available here: https://mratsim.github.io/Arraymancer/

• Huge performance improvements

  • Use non-initialized seq
  • shape and strides are now stored on the stack
  • optimization via inlining all higher-order functions
    • apply_inline, map_inline, fold_inline and reduce_inline templates are available.
  • all higher order functions are parallelized through OpenMP
  • integer matrix multiplication uses SIMD, loop unrolling, restrict and 64-bit alignment
  • prevent false sharing/cache contention in OpenMP reduction
  • remove temporary copies in several proc
  • runtime checks/exception are now behind unlikely
  • A*B + C and C+=A*B are automatically fused in one operation
  • do not initialized result tensors

• Neural network:

  • Added linear, sigmoid_cross_entropy, softmax_cross_entropy layers
  • Added Convolution layer

• Shapeshifting:

  • Added unsqueeze and stack

• Math:

  • Added min, max, abs, reciprocal, negate and in-place mnegate and mreciprocal

• Statistics:

  • Added variance and standard deviation

• Broadcasting

  • Added .^ (broadcasted exponentiation)

• Cuda:

  • Support for convolution primitives: forward and backward
  • Broadcasting ported to Cuda

• Examples

  • Added perceptron learning xor function example

• Precision

  • Arraymancer uses ln1p (ln(1 + x)) and exp1m procs (exp(1 - x)) where appropriate to avoid catastrophic cancellation

• Deprecated

  • Version 0.3.1 with the ALL deprecated proc removed will be released in a week. Due to issue https://github.com/nim-lang/Nim/issues/6436, even using non-deprecated proc like zeros, ones, newTensor you will get a deprecated warning.
  • newTensor, zeros, ones arguments have been changed from zeros([5, 5], int) to zeros[int]([5, 5])
  • All unsafe proc are now default and deprecated.

This release is named after "The Colour of Magic" (1983), the first book of Terry Pratchett masterpiece "Discworld".

I am very excited to announce the second release of Arraymancer which includes numerous improvements blablabla ...

Without further ado:

• Community

  • There is a Gitter room!

• Breaking

  • shallowCopy is now unsafeView and accepts let arguments
  • Element-wise multiplication is now .* instead of |*|
  • vector dot product is now dot instead of .*

• Deprecated

  • All tensor initialization proc have their Backend parameter deprecated.
  • fmap is now map
  • agg and agg_in_place are now fold and nothing (too bad!)

• Initial support for Cuda !!!

  • All linear algebra operations are supported
  • Slicing (read-only) is supported
  • Transforming a slice to a new contiguous Tensor is supported

• Tensors

  • Introduction of unsafe operations that works without copy: unsafeTranspose, unsafeReshape, unsafebroadcast, unsafeBroadcast2, unsafeContiguous,
  • Implicit broadcasting via .+, .*, ./, .- and their in-place equivalent .+=, .-=, .*=, ./=
  • Several shapeshifting operations: squeeze, at and their unsafe version.
  • New property: size
  • Exporting: export_tensor and toRawSeq
  • Reduce and reduce on axis

• Ecosystem:

  • I express my deep thanks to @edubart for testing Arraymancer, contributing new functions, and improving its overall performance. He built arraymancer-demos and arraymancer-vision,check those out you can load images in Tensor and do logistic regression on those!

This release is named after "Magician: Apprentice" (1982), the first book of Raymond E. Feist masterpiece "The Riftwar Cycle".

First public release.

Include:

• Slicing,
• basic linear algebra operations,
• reshaping, broadcasting, concatenating,
• universal function
• iterators
• ...

See README