Node v0.13.0

[0.13.0]

• [#1008] Decoder is now a composable trait, but without being backward incompatible
• [#1047, #1051, #1052] Processor is now a composable trait, but without being backward incompatible

Python v0.12.1

[0.12.1]

• [#938] Reverted breaking change. huggingface/transformers#16520

[YANKED] Rust v0.12.0

[0.12.0]

Bump minor version because of a breaking change.

The breaking change was causing more issues upstream in transformers than anticipated:
huggingface/transformers#16537 (comment)

The decision was to rollback on that breaking change, and figure out a different way later to do this modification

• [#938] Breaking change. Decoder trait is modified to be composable. This is only breaking if you are using decoders on their own. tokenizers should be error free.

• [#939] Making the regex in ByteLevel pre_tokenizer optional (necessary for BigScience)

• [#952] Fixed the vocabulary size of UnigramTrainer output (to respect added tokens)

• [#954] Fixed not being able to save vocabularies with holes in vocab (ConvBert). Yell warnings instead, but stop panicking.

• [#961] Added link for Ruby port of tokenizers

• [#960] Feature gate for cli and its clap dependency

[YANKED] Python v0.12.0

[0.12.0]

The breaking change was causing more issues upstream in transformers than anticipated:
huggingface/transformers#16537 (comment)

The decision was to rollback on that breaking change, and figure out a different way later to do this modification

Bump minor version because of a breaking change.

• [#938] Breaking change. Decoder trait is modified to be composable. This is only breaking if you are using decoders on their own. tokenizers should be error free.

• [#939] Making the regex in ByteLevel pre_tokenizer optional (necessary for BigScience)

• [#952] Fixed the vocabulary size of UnigramTrainer output (to respect added tokens)

• [#954] Fixed not being able to save vocabularies with holes in vocab (ConvBert). Yell warnings instead, but stop panicking.

• [#962] Fix tests for python 3.10

• [#961] Added link for Ruby port of tokenizers

[YANKED] Node v0.12.0

[0.12.0]

The breaking change was causing more issues upstream in transformers than anticipated:
huggingface/transformers#16537 (comment)

The decision was to rollback on that breaking change, and figure out a different way later to do this modification

Bump minor version because of a breaking change.
Using 0.12 to match other bindings.

• [#938] Breaking change. Decoder trait is modified to be composable. This is only breaking if you are using decoders on their own. tokenizers should be error free.

• [#939] Making the regex in ByteLevel pre_tokenizer optional (necessary for BigScience)

• [#952] Fixed the vocabulary size of UnigramTrainer output (to respect added tokens)

• [#954] Fixed not being able to save vocabularies with holes in vocab (ConvBert). Yell warnings instead, but stop panicking.

• [#961] Added link for Ruby port of tokenizers

Rust v0.11.2

• [#919] Fixing single_word AddedToken. (regression from 0.11.2)
• [#916] Deserializing faster added_tokens by loading them in batch.

Python v0.11.6

• [#919] Fixing single_word AddedToken. (regression from 0.11.2)
• [#916] Deserializing faster added_tokens by loading them in batch.

Node v0.8.3

node-v0.8.3

Adding rust release CI.

Python v0.11.5

[#895] Add wheel support for Python 3.10

Rust v0.11.1

• [#882] Fixing Punctuation deserialize without argument.
• [#868] Fixing missing direction in TruncationParams
• [#860] Adding TruncationSide to TruncationParams