Welcome to mirror list, hosted at ThFree Co, Russian Federation.

github.com/marian-nmt/marian.git - Unnamed repository; edit this file 'description' to name the repository.
summaryrefslogtreecommitdiff
path: root/regression-tests
AgeCommit message (Collapse)Author
2022-02-07update regression tests pointerMarcin Junczys-Dowmunt
2021-11-22Update submodule regression-testsRoman Grundkiewicz
2021-03-19Fix broken links to MNIST data sets (#838)Roman Grundkiewicz
2021-03-09Update regression-testsRoman Grundkiewicz
2021-02-07increase version to 1.10.01.10.0dev-masterMarcin Junczys-Dowmunt
2021-01-26Update regression-testsRoman Grundkiewicz
2021-01-25Update regression-testsRoman Grundkiewicz
2021-01-25Integrate intgemm into marian (#595)Nikolay Bogoychev
Adds intgemm as a module for Marian. Intgemm is @kpu 's 8/16 bit gemm library with support for architectures from SSE2 to AVX512VNNI Removes outdated integer code, related to the --optimize option Co-authored-by: Kenneth Heafield <github@kheafield.com> Co-authored-by: Kenneth Heafield <kpu@users.noreply.github.com> Co-authored-by: Ulrich Germann <ugermann@inf.ed.ac.uk> Co-authored-by: Marcin Junczys-Dowmunt <marcinjd@microsoft.com> Co-authored-by: Roman Grundkiewicz <rgrundkiewicz@gmail.com>
2020-11-11Merged PR 16337: Update sentencepiece to new versionMartin Junczys-Dowmunt
This updates the SentencePiece version in Marian to a much more recent revision. Due to that there is no dependency on Protobuf anymore.
2020-11-10Merged PR 16294: Change stopping criterion for mini-batch-fit binary searchMartin Junczys-Dowmunt
This PR changes the stopping criterion for mini-batch-fit binary search to better maximize batch size.
2020-11-07Merged PR 15925: Training embedder separation with margin lossesMartin Junczys-Dowmunt
* This PR adds training of embedding spaces with better separation based on https://arxiv.org/abs/2007.01852 * We can now train with in-batch negative examples or a handful of hand-constructed negative examples provided in a tsv-file.
2020-10-29Merged PR 15896: Add --after N option to supersede --after-batches and ↵Martin Junczys-Dowmunt
--after-epochs Replace `--after-batches N` and `--after-epochs N` with `--after Nu/Ne` which allows to specify updates, epochs, target labels with units, e.g.: * `--after 30Gt` or `--after 50ku` or `--after 10e` * Can also combine multiple criteria: `--after 30Gt,50ku,10e` and will stop when whichever hits first Changes default `cost-type` from `ce-mean` to `ce-sum` and turns `display-label-counts` on by default.
2020-09-07Merged PR 15233: Sync internal master with public masterMartin Junczys-Dowmunt
Regular sync of public and internal master.
2020-09-04Merged PR 15155: Enable output layer without output biasMartin Junczys-Dowmunt
Adds `--output-omit-bias` option which allows to train an output layer without a bias vector. This is expected to be useful for `--output-approx-knn` during decoding, as the LSH-based k-NN search is then exactly approximating the correct top-K values for decoding. The bias adds a shift otherwise. In first experiments the lack of the output bias does not seem to result in any performance loss.
2020-08-19re-introducing check if file exists if we are not using pipesMarcin Junczys-Dowmunt
2020-08-11disable mkl threading in cosmos.cppMarcin Junczys-Dowmunt
2020-07-29Merged PR 14402: Sync with public marian-dev master 1.9.31Roman Grundkiewicz
This simply pulls the recent updates from the public repo.
2020-05-17Update submodule regression-testsRoman Grundkiewicz
2020-05-14update versionMarcin Junczys-Dowmunt
2020-04-26Update submodule regression-testsRoman Grundkiewicz
2020-04-11Update submodule regression-testsRoman Grundkiewicz
2020-04-11Update submodule regression-testsRoman Grundkiewicz
2020-03-07bump versionMarcin Junczys-Dowmunt
2020-02-26Update the submodule regression-testsRoman Grundkiewicz
2020-01-18Merged PR 11188: Handle empty inputs with batch purgingMartin Junczys-Dowmunt
The previous mechanism to remove empty inputs does not play well with batch purging (removal of finished sentences). Now we reuse the batch purging mechanism to get rid of empty inputs by forcing EOS for all beam entries of a batch entry for the corresponding source batch entry. The purging then takes care of the rest. We set the probability to log(1) = 0.
2020-01-12move regression test pointerMarcin Junczys-Dowmunt
2020-01-12update regression test pointerMarcin Junczys-Dowmunt
2019-12-13ammend changelog and bump versionMarcin Junczys-Dowmunt
2019-11-26regression-tests/Marcin Junczys-Dowmunt
2019-10-31move regression testsMarcin Junczys-Dowmunt
2019-10-27move regression test pointerMarcin Junczys-Dowmunt
2019-10-27bump patch and changelogMarcin Junczys-Dowmunt
2019-10-26move regtest pointerMarcin Junczys-Dowmunt
2019-10-26move pointer to fitting regression testsMarcin Junczys-Dowmunt
2019-10-26merge with public masterMarcin Junczys-Dowmunt
2019-09-16change old error message backMarcin Junczys-Dowmunt
2019-09-16make aliases work with translation and rescoringMarcin Junczys-Dowmunt
2019-09-10set correct regression test versionMarcin Junczys-Dowmunt
2019-09-10update changelog and versionMarcin Junczys-Dowmunt
2019-09-09fix problem with loading averaged parameters for models which overload ↵Marcin Junczys-Dowmunt
Model::load
2019-09-08move regression tests forwardMarcin Junczys-Dowmunt
2019-09-06formattingMarcin Junczys-Dowmunt
2019-09-06bug fix of last commit, directly pushed to master as agreed by MarcinFrank Seide
2019-08-14Update regression testsRoman Grundkiewicz
2019-08-09Update submodule for regression testsRoman Grundkiewicz
2019-05-15updated submodule ref to regression-testsFrank Seide
2019-05-15minor bug fixes, to get last few changes to pass testsFrank Seide
2019-02-12move regression test pointerMarcin Junczys-Dowmunt
2019-02-05move pointer for reg testsMarcin Junczys-Dowmunt
2019-02-05update regression testsMarcin Junczys-Dowmunt
generated by cgit v1.2.3 (git 2.25.1) at 2024-06-21 17:44:08 +0300