Commits

Commit history

1. Commits on Jul 20, 2025

   • Update README.md

     SketchbookProtest

     authored
     a178b9eView commit details

     Copy full SHA for a178b9e

     Browse repository at this point

   • Add files via upload

     SketchbookProtest

     authored
     171d872View commit details

     Copy full SHA for 171d872

     Browse repository at this point
2. Commits on Jul 19, 2025

   • Update README.md

     SketchbookProtest

     authored
     de7b416View commit details

     Copy full SHA for de7b416

     Browse repository at this point

   • Update README.md

     SketchbookProtest

     authored
     8b01a91View commit details

     Copy full SHA for 8b01a91

     Browse repository at this point

   • Update README.md

     SketchbookProtest

     authored
     e546272View commit details

     Copy full SHA for e546272

     Browse repository at this point

   • Update README.md

     SketchbookProtest

     authored
     334d7c9View commit details

     Copy full SHA for 334d7c9

     Browse repository at this point

   • Update README.md

     SketchbookProtest

     authored
     56a2340View commit details

     Copy full SHA for 56a2340

     Browse repository at this point

   • Update README.md

     SketchbookProtest

     authored
     22332f1View commit details

     Copy full SHA for 22332f1

     Browse repository at this point

   • Add files via upload

     Show description for c292a25
     SketchbookProtest

     authored
     c292a25View commit details

     Copy full SHA for c292a25

     Browse repository at this point
3. Commits on Oct 30, 2024

   • initial alpha build - preprocessing, etc.

     SketchbookProtest

     authored
     46b34a8View commit details

     Copy full SHA for 46b34a8

     Browse repository at this point

   • Update README.md

     SketchbookProtest

     authored
     dbbb9d0View commit details

     Copy full SHA for dbbb9d0

     Browse repository at this point

   • Add files via upload

     SketchbookProtest

     authored
     5afe625View commit details

     Copy full SHA for 5afe625

     Browse repository at this point

   • Update README.md

     SketchbookProtest

     authored
     b2bb4dbView commit details

     Copy full SHA for b2bb4db

     Browse repository at this point

   • Update README.md

     SketchbookProtest

     authored
     eb07acaView commit details

     Copy full SHA for eb07aca

     Browse repository at this point

   • Update README.md

     SketchbookProtest

     authored
     45a14f2View commit details

     Copy full SHA for 45a14f2

     Browse repository at this point

   • Update README.md

     SketchbookProtest

     authored
     a8bf122View commit details

     Copy full SHA for a8bf122

     Browse repository at this point
4. Commits on Sep 12, 2024

   • Attempt #3 to add this directory to the new release

     AngledLuffa

     committed
     539760cView commit details

     Copy full SHA for 539760c

     Browse repository at this point

   • Attempt #3 to add this directory to the new release

     AngledLuffa

     committed
     8dd3cb8View commit details

     Copy full SHA for 8dd3cb8

     Browse repository at this point

   • Update version -> 1.9.1 to fix missing utils/visualization https://github.com/stanfordnlp/stanza/issues/1421

     AngledLuffa

     committed
     174768aView commit details

     Copy full SHA for 174768a

     Browse repository at this point

   • Add an init for stanza/units/visualization

     AngledLuffa

     committed
     c96f52eView commit details

     Copy full SHA for c96f52e

     Browse repository at this point

   • Set a default for num_layers based on the model type - num_layers=2 seems good for the character classifier

     AngledLuffa

     committed
     b999102View commit details

     Copy full SHA for b999102

     Browse repository at this point

   • Add a test of the character_classifier

     AngledLuffa

     committed
     7eb3a50View commit details

     Copy full SHA for 7eb3a50

     Browse repository at this point

   • If the seq2seq doesn't predict any spaces in the MWT, use the original word to avoid it going crazy

     AngledLuffa

     committed
     3211e72View commit details

     Copy full SHA for 3211e72

     Browse repository at this point

   • Allow --eval_file or --gold_file to override the values used by the run_mwt script to produce the final results

     AngledLuffa

     committed
     15e1408View commit details

     Copy full SHA for 15e1408

     Browse repository at this point

   • Allow for loading a save_name for the dev or test sets w/ or w/o the directory

     AngledLuffa

     committed
     c406d43View commit details

     Copy full SHA for c406d43

     Browse repository at this point

   • Save each MWT model based on a flag (useful for debugging individual checkpoints)

     AngledLuffa

     committed
     69d6adfView commit details

     Copy full SHA for 69d6adf

     Browse repository at this point

   • Refactor build_save_each_filename (just a few lines of code) for later reuse in MWT

     AngledLuffa

     committed
     6d7928aView commit details

     Copy full SHA for 6d7928a

     Browse repository at this point

   • Add a separate classifier type which checks exactly where MWT splits should go in the original text, rather than using a seq2seq to rebuild the text. For languages such as English where the splits…

     Show description for 43466ca
     AngledLuffa

     committed
     43466caView commit details

     Copy full SHA for 43466ca

     Browse repository at this point

   • Make it possible for the lemmatizer to return unk characters by reusing the expanded vocab features in the MWT

     AngledLuffa

     committed
     58a1ee0View commit details

     Copy full SHA for 58a1ee0

     Browse repository at this point

   • Refactor the DeltaVocab so as to make it easier for the Lemmatizer to use it

     AngledLuffa

     committed
     ff14d39View commit details

     Copy full SHA for ff14d39

     Browse repository at this point

   • Add a test that the delta vocab is working for the MWT

     AngledLuffa

     committed
     6a7fc14View commit details

     Copy full SHA for 6a7fc14

     Browse repository at this point

   • Add the ability to decode unknown tokens from the encoder side of the seq2seq model by temporarily augmenting the vocab with the new tokens. Helps tasks such as MWT which can receive characters ou…

     Show description for 807987b
     AngledLuffa

     committed
     807987bView commit details

     Copy full SHA for 807987b

     Browse repository at this point

   • Add an option to the seq2seq to never decode unk tokens. This is particularly useful in the MWT where sometimes it was decoding words to look like Bj<UNK>rk<UNK><UNK>ng

     AngledLuffa

     committed
     a8b2754View commit details

     Copy full SHA for a8b2754

     Browse repository at this point
5. Commits on Sep 6, 2024

   • Add a (possibly very noisy) str() for BaseVocab. Actually quite useful in the case of an MWT vocab, for example

     AngledLuffa

     committed
     ddb2caaView commit details

     Copy full SHA for ddb2caa

     Browse repository at this point
6. Commits on Sep 4, 2024

   • Separate out a long log line into pieces to make it easier to update going forward

     AngledLuffa

     committed
     8227bd2View commit details

     Copy full SHA for 8227bd2

     Browse repository at this point