github.com/stanfordnlp/stanza.git - Unnamed repository; edit this file 'description' to name the repository.

Age	Commit message (Expand)	Author
2022-09-25	Add a TREE_LSTM node combination method.con_tree_lstm2	John Bauer
2022-09-22	Transformer stack using MHA instead of LSTM as an option for the transitions ...	John Bauer
2022-09-22	Save the best_epoch as well	John Bauer
2022-09-21	Always use trainer.best_f1 instead of keeping a variable best_f1. Otherwise,...	John Bauer
2022-09-20	Clarify a bit of doc	John Bauer
2022-09-20	remove 'lstm' from transition_stack and constituent_stack, since any updated ...	John Bauer
2022-09-20	Add a split_size flag to make_lm_data.py to accommodate languages with smalle...	John Bauer
2022-09-20	Fix up the broken None args.output feature	John Bauer
2022-09-19	Add a bit of doc to LSTMTreeStack.__init__	John Bauer
2022-09-19	Refactor output() for the lstm_stacks. This makes it easier to refactor a ne...	John Bauer
2022-09-19	Comment on something that can be removed later	John Bauer
2022-09-19	Also refactor a constituent_lstm_stack. The unary transitions are a little w...	John Bauer
2022-09-19	Refactor the LSTM for the Transitions to a separate class. Will make it simp...	John Bauer
2022-09-17	Use the same lstm layer for transitions as well as constituents	John Bauer
2022-09-17	Restrict MultilingualPipeline to only load a few models. Will hopefully save...	John Bauer
2022-09-17	Maybe save a little GPU memory (fragmentation, at least) by reusing some larg...	John Bauer
2022-09-17	Try some option combinations which will hopefully make the models take up les...	John Bauer
2022-09-17	Also test sentence_boundary_vectors==none for the different transition embedd...	John Bauer
2022-09-17	Transition start should be the same as the embedding dim, not the hidden dim ...	John Bauer
2022-09-17	Save a best_f1 with the models so that we only resave a model when restarting...	John Bauer
2022-09-16	Add a small bit of doc	John Bauer
2022-09-16	Three lists appears slightly faster than one list of tuples	John Bauer
2022-09-16	Update tokenizer to use int64 for indices into an Embedding?	John Bauer
2022-09-16	Add a Hebrew charlm	John Bauer
2022-09-16	Each of these arguments should now be baked into every model we released in 1...	John Bauer
2022-09-15	Log notes about missing grad when using watch_regex	John Bauer
2022-09-15	Bump version number to release a few small changesHEAD v1.4.2 main	John Bauer
2022-09-15	normalize and sort dependencies, add transformers extra (#1124)	Nicholas Bollweg
2022-09-15	Hide the imports of SiLU and Mish from older versions of torch. #1120	John Bauer
2022-09-15	3.9 is a supported version of python now	John Bauer
2022-09-15	Stop requiring pytest for all installations. Instead we hopefully can hopefu...	John Bauer
2022-09-15	Switch dict + list to OrderedDict	John Bauer
2022-09-14	Update a couple versions in the README.md to better reflect reality - we supp...	John Bauer
2022-09-14	Squeeze a little bit more - only use depparse in the depparse pipelinev1.4.1	John Bauer
2022-09-14	Turn some multilingual pipeline tests into fixtures. Again, should save memory	John Bauer
2022-09-14	Turn some pipelines getting built over and over into fixtures. Will make the...	John Bauer
2022-09-14	Now there should be POS models which match the PL charlms as well	John Bauer
2022-09-14	Simpler way to have PL charlm specific for NER	John Bauer
2022-09-14	Try to reduce the scope on various pipelines to make the test suite less like...	John Bauer
2022-09-14	Lower log level on some messages we don't want written to the pipeline	John Bauer
2022-09-14	Oops, bugfix. Otherwise you get the whole dictionary for a language/model pa...	John Bauer
2022-09-13	Temporarily don't include charlm in the POS models for PL - those haven't ret...	John Bauer
2022-09-13	Add charlm to the sentiment dependencies when building resources.json	John Bauer
2022-09-13	PL now has an NER model	John Bauer
2022-09-13	Add a couple sentiment models for v1.4.1	John Bauer
2022-09-13	Add a tool to evaluate treebanks that are written out by a parser, such as wh...	John Bauer
2022-09-13	Refactor a little bit. Make it so the scoring interface can handle either sc...	John Bauer
2022-09-13	Default trees written with format _O	John Bauer
2022-09-12	Don't double save_dir if the user gives save_dir as part of the model filename	John Bauer
2022-09-12	Fix remove_optimizer mode	John Bauer