RecursiveAI.OneNlp.Neural
1.0.0
Prefix Reserved
dotnet add package RecursiveAI.OneNlp.Neural --version 1.0.0
NuGet\Install-Package RecursiveAI.OneNlp.Neural -Version 1.0.0
<PackageReference Include="RecursiveAI.OneNlp.Neural" Version="1.0.0" />
<PackageVersion Include="RecursiveAI.OneNlp.Neural" Version="1.0.0" />
<PackageReference Include="RecursiveAI.OneNlp.Neural" />
paket add RecursiveAI.OneNlp.Neural --version 1.0.0
#r "nuget: RecursiveAI.OneNlp.Neural, 1.0.0"
#:package RecursiveAI.OneNlp.Neural@1.0.0
#addin nuget:?package=RecursiveAI.OneNlp.Neural&version=1.0.0
#tool nuget:?package=RecursiveAI.OneNlp.Neural&version=1.0.0
RecursiveAI.OneNLP.Neural
spaCy-class NLP for .NET. The neural layer of RecursiveAI.OneNlp, running on OneModel with no native code. Unaffiliated with other projects named OneNLP.
- Encoders behind one seam: the fast hash-embedding encoder, and transformer encoders (ModernBERT and BERT/XLM-R families) loaded from exported ONNX weights, which you can fine-tune.
- Heads on any encoder: tagger, morphologizer, edit-tree lemmatizer, biaffine dependency parser, entity recognizer, text classifier (multi-class and multi-label), span categorizer, multiword splitter and word segmenter.
Train: the API the shipped models are trained with. AdamW, early stopping, atomic checkpoints that resume to the same weights, JSON Lines and CoNLL-U readers. The recipes in the repository are small console projects over it, so "retrain our entity recognizer on your data" means running our recipe.PipelineFilesaves and loads a whole pipeline as one file, andSentenceEmbedderruns sentence-embedding models.
using RecursiveAI.OneNlp;
Pipeline nlp = Pipeline.Blank(Language.English)
.Add(new TextClassifier(FastEncoder.English(7)) { OwnsEncoder = true });
nlp.Train(examples, new TrainOptions { Epochs = 10, Seed = 7 }); // examples: from JsonlReader or ConllUReader
PipelineFile.Save(nlp, "model.onenlp");
Models come in separate packages, one per language (RecursiveAI.OneNlp.Models.English, .French, .German, .Spanish, .Portuguese, .Dutch, .Italian, .Chinese, .Turkish, .Hindi), the nine-language distilled accurate tier (.Multilingual), the 84-language base-size accurate tier (.Multilingual.Large, 19 with entities) and for embeddings (.Embeddings, .Embeddings.Multilingual).
- Training and recipes: https://github.com/recursive-ai-oss/onenlp-csharp/blob/main/docs/training.md
- Measured results: https://github.com/recursive-ai-oss/onenlp-csharp/blob/main/docs/results-table.md
- Source: https://github.com/recursive-ai-oss/onenlp-csharp
- Home: https://recursiveai.org/onenlp/
Licensed MIT OR Apache-2.0. Copyright (c) Recursive AI, LLC.
| Product | Versions Compatible and additional computed target framework versions. |
|---|---|
| .NET | net10.0 is compatible. net10.0-android was computed. net10.0-browser was computed. net10.0-ios was computed. net10.0-maccatalyst was computed. net10.0-macos was computed. net10.0-tvos was computed. net10.0-windows was computed. |
-
net10.0
- RecursiveAI.OneModel (>= 1.2.0)
- RecursiveAI.OneModel.Training (>= 1.2.0)
- RecursiveAI.OneNlp (>= 1.0.0)
NuGet packages (15)
Showing the top 5 NuGet packages that depend on RecursiveAI.OneNlp.Neural:
| Package | Downloads |
|---|---|
|
RecursiveAI.OneNlp.Models.Multilingual.Large
spaCy-class NLP for .NET: the base-size accurate-tier model (fine-tuned mmBERT-base) for 84 languages. UD annotations and entity heads for 19 languages for the Nlp verbs; the Nlp verbs use it for these languages when no fast-tier package is referenced. About 3.5 GB, split over several packages. No downloads. Unaffiliated with other projects named OneNLP. |
|
|
RecursiveAI.OneNlp.Models.Multilingual.FrozenEncoder
spaCy-class NLP for .NET: the frozen distilled multilingual encoder (a six-layer, width-384 student of mmBERT-base, 142 MB) with its tokenizer, loadable as an encoder with FrozenStudentEncoder.Create(). No downloads. Unaffiliated with other projects named OneNLP. |
|
|
RecursiveAI.OneNlp.Models.English
spaCy-class NLP for .NET: the English fast-tier model. Tagging, morphology, lemmas, dependency parse and PER/ORG/LOC entities for the Nlp verbs, trained on Universal Dependencies and Universal NER English-EWT (CC BY-SA 4.0). No downloads. Unaffiliated with other projects named OneNLP. |
|
|
RecursiveAI.OneNlp.Models.Hindi
spaCy-class NLP for .NET: the Hindi fast-tier model. UPOS, morphology, dependency parse and PER/LOC/ORG entities for the Nlp verbs, trained on Hindi web text labelled by the accurate tier (there is no allowed Hindi treebank). No lemmatizer: the token text is the lemma. No downloads. Unaffiliated with other projects named OneNLP. |
|
|
RecursiveAI.OneNlp.Models.German
spaCy-class NLP for .NET: the German fast-tier model. Tagging, morphology, lemmas, dependency parse and PER/LOC/ORG/MISC entities for the Nlp verbs, trained on Universal Dependencies German-GSD (CC BY-SA 4.0) and GermEval 2014 (CC BY 4.0). No downloads. Unaffiliated with other projects named OneNLP. |
GitHub repositories
This package is not used by any popular GitHub repositories.
| Version | Downloads | Last Updated |
|---|---|---|
| 1.0.0 | 267 | 11/8/2026 |
See CHANGELOG.md in the repository.