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

Age	Commit message (Expand)	Author
2022-05-19	Simplify - can use torch tensors directly rather than first creating np arraysrefactor_dataloader	John Bauer
2022-05-19	Pytorch dataloader	John Bauer
2022-05-19	Start to refactor pieces of the tokenizer dataset into pieces so we can make ...	John Bauer
2022-05-19	numpy instead of torch is slightly faster in the small sentence regime, very ...	John Bauer
2022-05-17	labels is unused in tokenizer predict	John Bauer
2022-05-17	Remove unused import	John Bauer
2022-05-16	Add the skip_newlines test to the file reading version of the tokenizer data ...	John Bauer
2022-05-16	Abstract away labels() rather than having the eval code know the format of th...	John Bauer
2022-05-16	more specific Exception type	John Bauer
2022-05-16	Get rid of the input_data field - was only used for tests, and the tests don'...	John Bauer
2022-05-16	For the MWT test, use the fake tokenizer files rather than putting in the fak...	John Bauer
2022-05-16	Factor out a method to write the input to temp files in a tokenizer test	John Bauer
2022-05-16	Add a tiny bit of doc	John Bauer
2022-05-16	Merge pull request #1029 from stanfordnlp/refactor_tok	John Bauer
2022-05-16	Run some basic tests on the dictionary in the ZH tokenizer	John Bauer
2022-05-16	Rearrange - not necessary for this to be an inner function	John Bauer
2022-05-16	whitespace	John Bauer
2022-05-16	torch tensors instead of lists of numbers	John Bauer
2022-05-15	Separate out the label creation - no need to make a fake string of 0s at runtime	John Bauer
2022-05-15	Add a test which pokes the DataLoader object to make sure it is processing da...	John Bauer
2022-05-15	Separate the next() functionality which advances an unfinished batch into a s...	John Bauer
2022-05-14	whitespace update	John Bauer
2022-05-14	use save_name for consistency	John Bauer
2022-05-14	Use save_name to load if load_name is not set	John Bauer
2022-05-14	zh -> zh-hans to match the names of other models	John Bauer
2022-05-14	Use weighted_f1 by default to pick a best model	John Bauer
2022-05-14	Oops, got the test file name mixed up with train	John Bauer
2022-05-14	Add bert to sentiment training. Includes loading it in pipelines and at test...	John Bauer
2022-05-14	Log average loss even when doing an interim analysis	John Bauer
2022-05-14	Simplify loading of charlm in sentiment	John Bauer
2022-05-14	Remove unused const	John Bauer
2022-05-14	Use save_name and load_name in run_sentiment.py in order to simplify the name...	John Bauer
2022-05-14	Cache charlms in the FoundationCache, including a test of the loading.	John Bauer
2022-05-14	In conparser, load charlm in the load() method	John Bauer
2022-05-14	add pytest as requirement	J38
2022-05-14	update to new conda set up	J38
2022-05-14	write temp directories to test working dir not present dir	J38
2022-05-12	create id_tensor directly on the device needed	John Bauer
2022-05-12	Hindi bert model is missing max_length	John Bauer
2022-05-12	Use the dash replacement for the HI bert as well	John Bauer
2022-05-12	Merge pull request #1026 from stanfordnlp/charlm_refactor	John Bauer
2022-05-12	Simplify opening .xz vs regular text files	John Bauer
2022-05-12	Add a test of save & load charlm	John Bauer
2022-05-12	Move the charlm build code to char_model.py to be near the code it uses. Als...	John Bauer
2022-05-12	Simplify a bit - the phobert notes are now hidden in the bert_embedding module	John Bauer
2022-05-11	Some small notes on running hiner	John Bauer
2022-05-11	bert can be separate from word emb	John Bauer
2022-05-11	Technically this is more correcter -- no list in default parameters	John Bauer
2022-05-11	Add HiNER processing (which is just a conll -> json conversion)	John Bauer
2022-05-10	Add `str()` and `in` methods to multivocab	John Bauer