✨ New features and improvements NEW: Support multiprocessing in <code>nlp.pipe</code> via the <code>n_process</code> argument (Python 3 only). Base language support for Luxembourgish . Add noun chunks iterator for Swedish . Retrained models for Greek , Norwegian Bokmål and Lithuanian that now correctly support parser-based sentence segmentation. Repackaged models for Greek and German with improved lookup tables via <code>spacy-lookups-data</code> . Add warning in <code>debug-data</code> for low sentences per doc ratio. Improve checks and errors related to ill-formed IOB input in <code>convert</code> and <code>debug-data</code> CLI. Support training dict format as JSONL. Make <code>EntityRuler</code> ID resolution 2× faster and support <code>"id"</code> in patterns to set <code>Token.ent_id</code>. Improve rendering of named entity spans in <code>displacy</code> for RTL languages. Update Thinc to ditch <code>thinc_gpu_ops</code> for simpler GPU install. Support Mish activation in <code>spacy pretrain</code>. Add backwards-compatible support for new <code>Language.disable_pipes</code> API, which will become the default in the future. The method can now also take a list of component names as its first argument (instead of a variable number of arguments). <code class="lang-diff">- disabled = nlp.disable_pipes("tagger", "parser") + disabled = nlp.disable_pipes(["tagger", "parser"]) </code> Add backwards-compatible support for new <code>Matcher.add</code> and <code>PhraseMatcher.add</code> API, which will become the default in the future. The patterns are now the second argument and a list (instead of a variable number of arguments). The <code>on_match</code> callback becomes an optional keyword argument. <code class="lang-diff">patterns = [[