Package index • sbert

Encoding text

Turn text into Sentence-BERT embeddings.

encode(): Encode Text with Sentence-BERT
pool(): Pool Token Embeddings into Sentence Embeddings

Models and runtime

Install the ONNX Runtime, download revision-pinned models, load custom Hugging Face exports, and manage the on-disk cache.

models(): List the Available Pinned Sentence-BERT Models
model_download(): Download a Pinned Sentence-BERT Model
load_model(): Load a Pinned Sentence-BERT Model
load_custom(): Load an Arbitrary Hugging Face Embedding Model
model_status(): Inspect an Installed Model
model_remove(): Remove an Installed Model from the Cache
install_runtime(): Install ONNX Runtime
cache_dir(): Locate the sbert Model Cache
cache_size(): Measure the sbert Cache

Preparing a corpus

Deduplicate, segment, and context-blend documents before modeling.

dedupe(): Deduplicate a Text Corpus with Frequencies
clean_corpus(): Clean a Text Corpus Before Encoding
strip_list_markers(): Strip Enumeration and List Markers from Text
content_ratio(): Alphabetic Content Ratio of Text
segment(): Segment Text into Sentences, Clauses, or Phrases
abbreviations(): Obtain the Built-in Abbreviation Gazetteer
blend(): Blend Segment Embeddings with Their Document Context
stop_words(): Obtain and Adjust the Topic Stop-Word List

Topic modeling

Fit topic models, sweep the topic count, and merge topics down.

topic_corpus(): Prepare a Corpus Once to Fit Many Topic Models
topics(): Discover Semantic Topics in Documents
select_topics(): Compare Topic Counts Before Committing to One
fitted(<sbert_topic_sweep>): Extract One Fitted Model from a Topic-Count Sweep
reduce_topics(): Reduce a Fitted Topic Model to Fewer Topics

Describing topics

Distinctive terms, representative documents, sizes, and keywords.

terms(<sbert_topic_model>): Topic Terms, Retuned Without Refitting
representatives(): Representative Text Units for Every Topic
topic_sizes(): Topic Sizes on the Distinct and Weighted Scales
keywords(): Extract Keywords from Documents by Embedding Similarity

Topic inference

Assign new documents and recover soft, generative, and mixed memberships.

predict(<sbert_topic_model>): Assign New Documents to Fitted Topics
topic_membership(): Soft Topic Membership Probabilities
topic_gamma(): Document-Topic Distributions from Segment Assignments

Evaluating topics

Intrinsic coherence and topic diversity.

coherence(): Score Topic Coherence
topic_diversity(): Measure Topic Diversity

Visualizing and summarizing

Deterministic base-graphics plots, hierarchies, and reports.

summary(<sbert_topic_model>): Summarize a Semantic Topic Model
plot(<sbert_topic_model>): Plot a Semantic Topic Model
plot(<sbert_topic_sweep>): Plot a Topic-Count Sweep
topic_hierarchy(): Build the Topic Hierarchy of a Fitted Model
topic_similarity(): Compute Cosine Similarity
topic_palette(): Qualitative Colour Palette for Topics

Data

Bundled corpora for offline examples and topic modeling.

feedback_translations: Levebee AI Mathematics Feedback with English Translations
covid: COVID-19 Research Abstracts