Welcome to
mirror list
, hosted at
ThFree Co
, Russian Federation.
github.com/stanfordnlp/stanza.git - Unnamed repository; edit this file 'description' to name the repository.
index
:
github.com/stanfordnlp/stanza.git
142_resources_patch
NFC
am-notes
aws_sagemaker_tooling
azshan
beam
bert_mix
charlm_cache
charlm_checkpoint
con_add_pattn
con_attn2
con_bigrams
con_checkpoint
con_classifier
con_classifier_reranking
con_focal
con_freeze
con_kbest
con_lattn
con_lattn2
con_mixed_pattn
con_mlp
con_mlp_inputs
con_multitask
con_pattn_lr
con_pattn_replace
con_restart_transitions
con_self_gan
con_shift_tags
con_shift_tags2
con_shift_transitions
con_simple_transformer
con_simple_unary
con_skip
con_trans
con_tree_lstm
con_tree_lstm2
con_tree_lstm3
con_tstack
con_vector_dropout
con_vit
con_warmup
con_warmup_2
con_warmup_lattn
dataloader_local
de_ner
dev
elmo2
elmo_many
fewer_cuda
fix_unit_tests
gh-pages
gh-pages-sent
hebrew_combined
hi-layered-ner
hi-ner-cleaned
hi-shuffle
hi_ner
hi_ner_final
inorder_unary
kazakh_ner
kk_trans
lattn_issue
m1
main
marathi
margin_penalty
masakhane
masks
masks2
masks3
ner_bert
ner_bert_copy
ner_wv
ninf_langid
nner
no_header_pt
numeric_re
ordered_dict
pattn_issue
pos_bert
pos_charlm
ppf_data
pydataloader
refactor_dataloader
refactor_lstm
refactor_tok
refactor_tokenizer
refactor_tokkenizer_2
runner-demo
semgrex_search_visualization
sentence_ids
sentiment
sentiment_charlm
sentiment_lstm
sentiment_trainer
sindhi
spanish_sent
t5
t5b
tagger_mha
thai-sybrnn
thai_ner
tiny_ud
token
tokens
tr_ner
trans_lm
tweet
ug_ner
update_stanza
updated_eval
updated_eval_2
vi_bert_last
visualization
wandb
word_lstm_pattn
wordinput-sentsegmenter
xpos
Unnamed repository; edit this file 'description' to name the repository.
www-data
summary
refs
log
tree
commit
diff
log msg
author
committer
range
Branch
Commit message
Author
Age
142_resources_patch
Put all needed pretrains in default.zip in case the NER uses a different pret...
John Bauer
19 months
NFC
NFC instead of NFD
John Bauer
18 months
am-notes
Update workspace.xml
anwesham-lab
3 years
aws_sagemaker_tooling
Dumping out useless print statements and import statements
SecroLoL
19 months
azshan
testing git controls
unknown
23 months
beam
Beam training loss - rough draft
John Bauer
19 months
bert_mix
Mix with N vectors instead of just 1
John Bauer
18 months
charlm_cache
Add bert to sentiment training. Includes loading it in pipelines and at test...
John Bauer
2 years
charlm_checkpoint
Add a trainer for the charlm - useful for saving and loading everything for c...
John Bauer
21 months
con_add_pattn
Finetune a model that was learned with no pattn to now use pattn
John Bauer
23 months
con_attn2
ATTN method to build larger constituents out of smaller constituents
John Bauer
2 years
con_bigrams
maybe try to avoid the bigram exploding
John Bauer
2 years
con_checkpoint
Always save checkpoints. Always load from a checkpoint if one exists.
John Bauer
20 months
con_classifier
Load a parser as an argument to the parser.
John Bauer
19 months
con_classifier_reranking
Rough draft of using the classifier as a reranker
John Bauer
19 months
con_focal
Use a pip installable focal loss library instead?
John Bauer
19 months
con_freeze
Rather than freezing, just learn reaaaaaly slowly
John Bauer
21 months
con_kbest
Add a second type of reranker (need to make these general)
John Bauer
2 years
con_lattn
Experiment with not doing weight decay at all for pattn
John Bauer
23 months
con_lattn2
Move the LayerNorm outside of the positional encodings. Set the d_model used...
John Bauer
21 months
con_mixed_pattn
Learning the mixing factor rather than hardcoding it to 0.1
John Bauer
20 months
con_mlp
Attempt the MLP again (forgot that it might be more useful in the case of ber...
John Bauer
2 years
con_mlp_inputs
Combine all inputs using MLPs rather than concat
John Bauer
2 years
con_multitask
Begin making it so you can use multiple treebanks and annotation schemes
John Bauer
21 months
con_pattn_lr
low lr and low weight decay for the norms of pattn
John Bauer
23 months
con_pattn_replace
Potentially multiply p into c in the partitioned transformer
John Bauer
2 years
con_restart_transitions
Restart transitions when restarting training
John Bauer
20 months
con_self_gan
Add a loss to make the model try to approximate what the classifier would lik...
John Bauer
2 years
con_shift_tags
Add a flag to control how many tags to use when labeling shift transitions
John Bauer
20 months
con_shift_tags2
Add a flag to control how many tags to use when labeling shift transitions
John Bauer
18 months
con_shift_transitions
Add a label to a Shift
John Bauer
20 months
con_simple_transformer
Add a simple MHA to the model
John Bauer
2 years
con_simple_unary
Remove the specialized unary transform
John Bauer
2 years
con_skip
Use the dynamic LSTM for constituents. Not sure it helps, is definitely slow
John Bauer
2 years
con_trans
use two transitions as input for the LSTM. not sure this will help at all
John Bauer
20 months
con_tree_lstm
Attempt to come up with an initial tree_cx for the TREE_LSTM method
John Bauer
20 months
con_tree_lstm2
Add a TREE_LSTM node combination method.
John Bauer
20 months
con_tree_lstm3
Fix
John Bauer
2 years
con_tstack
Transformer stack. Currently only has one head of attention. TODO: Probably...
John Bauer
20 months
con_vector_dropout
Don't remove pattn... we rely on dropout to prevent pattn from going to 0
John Bauer
20 months
con_vit
IT WORKSgit diffgit diff
John Bauer
2 years
con_warmup
low weight decay for the norms
John Bauer
23 months
con_warmup_2
AdaDelta warmup for the conparser. Motivation: AdaDelta results in
John Bauer
23 months
con_warmup_lattn
Add an option to build the lattn out of the entire input, not just pattn
John Bauer
23 months
dataloader_local
Support variants and pretokenized as before
John Bauer
3 years
de_ner
Attempt to compensate for German BERT tokenizers not handling soft hyphen ver...
John Bauer
2 years
dev
Fix broken unittest
John Bauer
18 months
elmo2
Add support for elmoformanylangs to sentiment
John Bauer
21 months
elmo_many
Allow different size input transforms in the NER
John Bauer
3 years
fewer_cuda
Remove some cuda() calls in favor of getting the device instead. Will make i...
John Bauer
20 months
fix_unit_tests
add pytest as requirement
J38
2 years
gh-pages
Update an obsolete comment
John Bauer
19 months
gh-pages-sent
Add a (not quite complete) sentiment model page
John Bauer
21 months
hebrew_combined
Add the capacity to build he_combined models from UD_Hebrew-IAHLTwiki and a f...
John Bauer
21 months
hi-layered-ner
yeah
anwesham-lab
3 years
hi-ner-cleaned
i
anwesham-lab
3 years
hi-shuffle
Update convert_fire_2013.py
anwesham-lab
3 years
hi_ner
Update convert_fire_2013.py
anwesham-lab
3 years
hi_ner_final
Update convert_fire_2013.py
anwesham-lab
2 years
inorder_unary
will combine PreterminalUnary with CompoundUnary
John Bauer
18 months
kazakh_ner
whitespace
John Bauer
21 months
kk_trans
less noisy, context for the output
John Bauer
21 months
lattn_issue
Replace all inputs with the lattn inputs if it's on. Problem: this does not ...
John Bauer
18 months
m1
fix unittest?
yuhui-zh15
24 months
main
Bump version number to release a few small changes
John Bauer
20 months
marathi
Add a processing for the MR l3cube sentiment
John Bauer
23 months
margin_penalty
Add a margin_loss term
John Bauer
18 months
masakhane
Add a preparation script for Masakhane NER
John Bauer
21 months
masks
fix bert_embedding.py
Kantapong Kotchum
21 months
masks2
Just to compare masked or not-masked for xlnet
John Bauer
21 months
masks3
Another attempt at xlnet masks
John Bauer
21 months
ner_bert
Add bert embeddings to the bottom layer of the NER.
Vy
2 years
ner_bert_copy
In situations where we already have a bert_tokenizer loaded, don't load it ag...
John Bauer
2 years
ner_wv
Script to remove unmodified pieces of pretrain from NER model
John Bauer
24 months
ninf_langid
Mask illegal langauges by setting them to -ninf. 0 means that illegal langua...
John Bauer
22 months
nner
add ner_max_depth argument which will be called later
Kantapong Kotchum
21 months
no_header_pt
Add a test of the no-header pretrain
John Bauer
22 months
numeric_re
fix that it takes forever to tokenize a really long non-numeric token
John Bauer
23 months
ordered_dict
DO NOT MERGE - this keeps a large object on the GPU between tests
John Bauer
20 months
pattn_issue
more heads? should run some experiments
John Bauer
18 months
pos_bert
Upgrading POS models to use Bert will require a resources version bump
John Bauer
20 months
pos_charlm
Add a pos-specific charlm map for the medical EN datasets and the one dataset...
John Bauer
22 months
ppf_data
small fix
Pfeiffer_T480s
3 years
pydataloader
Add a note on something that doesn't seem to help
John Bauer
3 years
refactor_dataloader
Simplify - can use torch tensors directly rather than first creating np arrays
John Bauer
24 months
refactor_lstm
Also refactor a constituent_lstm_stack. The unary transitions are a little w...
John Bauer
20 months
refactor_tok
turn the labels into a separate array
John Bauer
24 months
refactor_tokenizer
Separate out the label creation - no need to make a fake string of 0s at runtime
John Bauer
2 years
refactor_tokkenizer_2
Refactor a Dataset & Dataloader... although it is only used to prepare the se...
John Bauer
2 years
runner-demo
demo
J38
24 months
semgrex_search_visualization
Dumping out old and unneeded files from development
SecroLoL
21 months
sentence_ids
Process sentence ids from the corpus, if available.
John Bauer
2 years
sentiment
Switch the VI model to use words tokenized from the stanza tokenizer rather t...
John Bauer
2 years
sentiment_charlm
simplify a bit
John Bauer
2 years
sentiment_lstm
2d conv. Uses the width of a conv feature to rescale the output
John Bauer
21 months
sentiment_trainer
Refactor a Trainer object out of the classifier.py main program. In addition...
John Bauer
21 months
sindhi
Update a few broken tags from the Sindhi NER dataset and add a little documen...
John Bauer
20 months
spanish_sent
Wrapper for converting Spanish TASS2020 for sentiment
John Bauer
21 months
t5
start t5 integration... needs more masks
John Bauer
22 months
t5b
Attempt to integrate T5 into NER/conparse. T5-small doesn't work well at all...
John Bauer
22 months
tagger_mha
Potentially use a MHA layer in the tagger
John Bauer
18 months
thai-sybrnn
Update model.py
Gordon
3 years
thai_ner
Merge branch 'dev' into thai_ner
John Bauer
22 months
tiny_ud
Add expected xpos vocabs for the tiny treebanks
John Bauer
21 months
token
oops
John Bauer
19 months
tokens
Write spaces as their own tokens in Thai? Probably needs a lot of work
John Bauer
22 months
tr_ner
Add more debug to the watch_regex option
John Bauer
2 years
trans_lm
Adjust learning rate, don't print out infinite ppl
John Bauer
18 months
tweet
Copy over the changes from TweebankNLP. Will merge or edit them later
John Bauer
20 months
ug_ner
Aug_17
Arman Aydin
21 months
update_stanza
Updated syllable token + character token model framework
FTdiscovery
3 years
updated_eval
pointless refactoring
John Bauer
21 months
updated_eval_2
Update compose_ete_results.py to allow multiple input files
John Bauer
21 months
vi_bert_last
Use the last word piece instead of the first
John Bauer
18 months
visualization
Removing unnecessary print statements
SecroLoL
22 months
wandb
Generalize the sentiment csv reading code and move it to process_utils
John Bauer
23 months
word_lstm_pattn
Move the pattn & lattn after the word lstm. The position information should ...
John Bauer
18 months
wordinput-sentsegmenter
Update trainer.py
Gordon
3 years
xpos
Automatically determine the vocab type if it isn't already known
John Bauer
22 months