Commits

Branch selector

 master

User selector

All users

All time

Commit History

Commits on Oct 19, 2025

• TF is deprecated in Transformers and no longer maintained: https://github.com/hankcs/HanLP/issues/1959

  hankcs

  committed
  942ce93

  Copy full SHA for 942ce93

Commits on Sep 11, 2025

• Test on ubuntu-latest, macos-latest, windows-latest

  hankcs

  committed
  e72db55

  Copy full SHA for e72db55

Commits on Sep 7, 2025

• Fix stdio redirection, fix: https://github.com/hankcs/HanLP/issues/1958

  hankcs

  committed
  5f46aec

  Copy full SHA for 5f46aec

Commits on Jan 15, 2025

• Support reusing existing model and config for finetuning, fix: https://github.com/hankcs/HanLP/issues/1942

  hankcs

  committed
  199f3f3

  Copy full SHA for 199f3f3

Commits on Jan 14, 2025

• Test on ubuntu-20.04, macos-latest, windows-latest

  hankcs

  committed
  81983df

  Copy full SHA for 81983df

• Keep the new hyperparameters when fine-tuning a model, fix: https://github.com/hankcs/HanLP/issues/1942

  hankcs

  committed
  80506cf

  Copy full SHA for 80506cf

Commits on Jan 13, 2025

• Revise document

  hankcs

  committed
  a639dbc

  Copy full SHA for a639dbc

• Release Ancient Chinese models for tokenization, lemmatization, part-of-speech tagging and dependency parsing

  hankcs

  committed
  2eb94f2

  Copy full SHA for 2eb94f2

• Revise documents for Ancient Chinese models

  hankcs

  committed
  0636dc6

  Copy full SHA for 0636dc6

• Release Ancient Chinese MTL model `KYOTO_EVAHAN_TOK_LEM_POS_UDEP_LZH`

  hankcs

  committed
  f717be1

  Copy full SHA for f717be1

Commits on Jan 12, 2025

• Display the main pos by default

  hankcs

  committed
  c1aa002

  Copy full SHA for c1aa002

• Support XPOS for CoNLL-U

  hankcs

  committed
  7aeb0fc

  Copy full SHA for 7aeb0fc

• Implement TransformerTaggingLemmatizer

  hankcs

  committed
  9988be6

  Copy full SHA for 9988be6

• Release Ancient Chinese tokenization model `KYOTO_EVAHAN_TOK_LZH`

  hankcs

  committed
  e8bad3f

  Copy full SHA for e8bad3f

Commits on Jan 11, 2025

• Skip hidden states averaging for tokenization only models

  hankcs

  committed
  c0d5d92

  Copy full SHA for c0d5d92

• Mirror `bert-ancient-chinese`

  hankcs

  committed
  9529587

  Copy full SHA for 9529587

• Automatically detect file type by `.conllu` extension

  hankcs

  committed
  5c6d17b

  Copy full SHA for 5c6d17b

Commits on Jan 7, 2025

• Release `EN_TOK_LEM_POS_NER_SRL_UDEP_SDP_CON_MODERNBERT_LARGE`

  hankcs

  committed
  3185188

  Copy full SHA for 3185188

• Replace `FloatTensor` with `sparse_coo_tensor`

  hankcs

  committed
  7a4b735

  Copy full SHA for 7a4b735

Commits on Dec 29, 2024

• Release `2.1.0`

  hankcs

  committed
  749ecf6

  Copy full SHA for 749ecf6

• Improve logging

  hankcs

  committed
  d4660c3

  Copy full SHA for d4660c3

Commits on Dec 28, 2024

• TransformList supports config

  hankcs

  committed
  3202623

  Copy full SHA for 3202623

• Release an English MTL model with ModernBERT encoder

  hankcs

  committed
  3aba649

  Copy full SHA for 3aba649

• Revise documentation

  hankcs

  committed
  5698f32

  Copy full SHA for 5698f32

• Update eos model config to support new lines

  hankcs

  committed
  75edc60

  Copy full SHA for 75edc60

• Support coarse tokenization of Japanese

  hankcs

  committed
  007f29f

  Copy full SHA for 007f29f

• Implement `PrependSpace`

  hankcs

  committed
  c583768

  Copy full SHA for c583768

• Replace `FloatTensor` with `sparse_coo_tensor`

  hankcs

  committed
  77481bd

  Copy full SHA for 77481bd

• Heuristic to fix lemma for digits

  hankcs

  committed
  36c9e6f

  Copy full SHA for 36c9e6f

• Support ModernBERT as a tokenizer

  hankcs

  committed
  336fd93

  Copy full SHA for 336fd93

Commits on Dec 18, 2024

• Upgrade GitHub Actions to the latest versions

  hankcs

  committed
  d8e6f2a

  Copy full SHA for d8e6f2a

• Revert binding to the immutable full sha1 as it may prevent us from getting fixes: https://github.com/actions/toolkit/blob/main/docs/action-versioning.md#compatibility

  hankcs

  committed
  ae3a910

  Copy full SHA for ae3a910

• fix the risk of permissions and unpinned dependencies in the workflow

  

  gcanlin

  authored and

  hankcs

  committed
  d42012d

  Copy full SHA for d42012d

Commits on Dec 7, 2024

• Fix `dep` key of `to_pretty`

  hankcs

  committed
  c01f4f6

  Copy full SHA for c01f4f6

Commits on Dec 2, 2024

• Fix loading issue of fine-tuned NER models

  hankcs

  committed
  1197741

  Copy full SHA for 1197741