ReLiK: Retrieve and LinK, Fast and Accurate Entity Linking and Relation Extraction on an Academic Budget
Paper
•
2408.00103
•
Published
•
18
ReLiK: Retrieve, Read and LinK
A blazing fast and lightweight Information Extraction model for Entity Linking and Relation Extraction.
-
-
Running on Zero47🏃
Information Extraction
sapienzanlp/relik-entity-linking-large
Updated • 244 • 10Note Pretrained configuration of ReLiK-large for Entity Linking
sapienzanlp/relik-entity-linking-base
Updated • 44 • 1Note Pretrained configuration for ReLiK-base for Entity Linking
-
sapienzanlp/relik-entity-linking-small
Updated • 123 • 2
relik-ie/relik-relation-extraction-small
Updated • 325 • 1Note Pretrained configuration for ReLiK-small for Relation Extraction.
relik-ie/relik-cie-small
Updated • 296 • 5Note Pretrained configuration for ReLiK-small for Relation Extraction and Entity Linking (aka closed Information Extraction). It will load one Retriever per task, and a shared Reader.
relik-ie/relik-cie-tiny
Updated • 38 • 1Note Note Pretrained configuration for ReLiK-tiny for Relation Extraction and Entity Linking (aka closed Information Extraction). It will load one Retriever (both small) per task, and a shared Reader (small).
sapienzanlp/relik-relation-extraction-nyt-large
Updated • 414 • 4Note Pretrained configuration for ReLiK-large for Relation Extraction trained on the NYT corpus
sapienzanlp/relik-retriever-e5-base-v2-aida-blink-encoder
Feature Extraction • Updated • 382Note Retriever Encoder fine-tuned on AIDA for Entity Linking
sapienzanlp/relik-retriever-e5-base-v2-aida-blink-wikipedia-index
Updated • 213 • 1Note Wikipedia index computed with sapienzanlp/relik-retriever-e5-base-v2-aida-blink-encoder
sapienzanlp/relik-retriever-e5-base-v2-aida-blink-wikipedia-faiss-flat-index
Updated • 4Note Wikipedia FAISS index computed with sapienzanlp/relik-retriever-e5-base-v2-aida-blink-encoder
sapienzanlp/relik-reader-deberta-v3-large-aida
Feature Extraction • Updated • 258 • 1Note Entity Linking pretrained Large Reader Encoder
sapienzanlp/relik-reader-deberta-v3-base-aida
Feature Extraction • Updated • 195Note Entity Linking pretrained Base Reader Encoder
sapienzanlp/relik-retriever-e5-base-v2-blink-first1M-encoder
Feature Extraction • Updated • 5Note Retriever Encoder pretrained on BLINK
sapienzanlp/relik-reader-deberta-v3-large-nyt
Feature Extraction • Updated • 797Note Relation Extraction pretrained Large Reader Encoder
sapienzanlp/relik-retriever-small-nyt-document-index
Updated • 424Note Wikipedia index computed with sapienzanlp/relik-retriever-small-nyt-question-encoder
sapienzanlp/relik-retriever-small-nyt-question-encoder
Feature Extraction • Updated • 797Note Retriever Encoder fine-tuned on NYT
sapienzanlp/relik-retriever-e5-base-v2-aida-blink-wikipedia-2Mpopular-index
Updated • 7Note Wikipedia index computed with 2M entities filtered by popularity with sapienzanlp/relik-retriever-e5-base-v2-aida-blink-encoder. Can be used instead of sapienzanlp/relik-retriever-e5-base-v2-aida-blink-wikipedia-index for more lightweight inference and up to date Wikipedia titles, if performance on GERBIL is not the main objective.
sapienzanlp/relik-retriever-e5-base-v2-aida-blink-wikipedia-index-updated
Updated • 5Note Wikipedia index computed with sapienzanlp/relik-retriever-e5-base-v2-aida-blink-encoder. Can be used instead of sapienzanlp/relik-retriever-e5-base-v2-aida-blink-wikipedia-index for more up to date Wikipedia titles, if performance on GERBIL is not the main objective.