summaryrefslogtreecommitdiff
path: root/src
Commit message (Collapse)AuthorAge
* Test GMM; it failsTom Smeding2024-11-10
|
* Fix bug in simplifier (effects are hard)Tom Smeding2024-11-10
|
* Complete GMM implementationTom Smeding2024-11-10
|
* Cleanup, more Language operationsTom Smeding2024-11-09
|
* Add Custom to LanguageTom Smeding2024-11-09
|
* Some more primitive operatorsTom Smeding2024-11-09
|
* Maximum/minimumTom Smeding2024-11-09
|
* WIP maximum/minimumTom Smeding2024-11-08
|
* Custom derivativesTom Smeding2024-11-08
|
* WIP custom derivativesTom Smeding2024-11-08
|
* Remove build1Tom Smeding2024-11-07
|
* BenchmarkTom Smeding2024-11-07
|
* Some simplification rulesTom Smeding2024-11-06
|
* Generate EOneHot in D[EIdx]Tom Smeding2024-11-05
| | | | | | This generates a one-hot for the zero-dimensional inner array because indexing one level further to the actual element is too difficult. But this should simplify away fine.
* Support EOneHotTom Smeding2024-11-05
|
* WIP EOneHotTom Smeding2024-11-04
|
* simplifyFixTom Smeding2024-10-29
|
* Store only on tape what's used in the dualTom Smeding2024-10-29
|
* WIP preserve only subset of D0 bindings in dual (...)Tom Smeding2024-10-27
| | | | | | | | | | | | | | | The point of this is to ensure that when an expression occurs in a Build, then the parts of D0 that are only there to make sharing work out for D1 are not laboriously taped in an array and preserved for D2, only for D2 to ignore them. However, while the subtape machinery is a good first step, this is not everything: the current Build translation makes a Build for the (elementwise) tape and separately a build for the primal. Because the primal _does_ generally need the subtaped-away stuff, we can't just not tape those. TODO: figure out how to resolve this / what the next step is.
* Fix buildTom Smeding2024-10-27
|
* Better inline syntax for LanguageTom Smeding2024-10-27
|
* Fix EIdx derivativeTom Smeding2024-10-26
|
* Show accumulator as dummy string, not errorTom Smeding2024-10-26
|
* Fix IIxNil warningTom Smeding2024-10-26
|
* Make interpreter debug printing conditionalTom Smeding2024-10-26
|
* Fix {} usage in pretty-printing of ELetTom Smeding2024-10-26
|
* DebuggingTom Smeding2024-10-26
|
* Fix interpreter bugTom Smeding2024-10-22
|
* Less warningsTom Smeding2024-10-22
|
* Differentiate ReplicateTom Smeding2024-10-22
|
* TestsTom Smeding2024-10-21
|
* WIP testing neuralTom Smeding2024-10-18
|
* More towards test suiteTom Smeding2024-10-14
|
* Towards a test suiteTom Smeding2024-10-07
|
* Reverse-by-forward, and checking neural (it's wrong)Tom Smeding2024-10-01
|
* Dual-numbers forward ADTom Smeding2024-09-28
|
* The size of the empty shape is 1, not 0Tom Smeding2024-09-26
|
* Test neuralTom Smeding2024-09-25
|
* Add some missing casesTom Smeding2024-09-22
|
* Interpreter typechecks, at the cost of compositionality of RepAcTom Smeding2024-09-22
|
* StuffTom Smeding2024-09-22
|
* Some hackingTom Smeding2024-09-22
|
* A little interpreter accum progressTom Smeding2024-09-16
|
* WIP Accum stuffTom Smeding2024-09-15
|
* WIP better zero/plus, fixing Accum (...)Tom Smeding2024-09-13
| | | | | | | | | | | | | The accumulator implementation was wrong because it forgot (in accumAdd) to take into account that values may be variably-sized. Furthermore, it was also complexity-inefficient because it did not build up a sparse value. Thus let's go for the Haskell-interpreter-equivalent of what a real, fast, compiled implementation would do: just a tree with mutable variables. In practice one can decide to indeed flatten parts of that tree, i.e. using a tree representation for nested pairs is bad, but that should have been done _before_ execution and for _all_ occurrences of that type fragment, not live at runtime by the accumulator implementation.
* Towards neuralTom Smeding2024-09-12
|
* Interpreter, some operationsTom Smeding2024-09-12
|
* Finish Accum implementationTom Smeding2024-09-11
|
* More WIP interpreterTom Smeding2024-09-11
|
* WIP interpreterTom Smeding2024-09-06
|
generated by cgit v1.2.3-70-g09d2 (git 2.48.1) at 2025-04-02 09:19:05 +0200