Searching for the Right Words: Bringing NLP to Apache solr through ONNX and OpenNLP
Talkat Open AI + Data Forum @ Open Source Summit North America 2022 · on Thursday, 2022-06-23· at 11:10· Room 303/304 (Level 3), JW Marriott Austin, 110 E 2nd Street Austin, TX 78701·
summary
Converts Hugging Face Transformers models to ONNX runtime models, inspects inputs/outputs with Netron, and wires them into Java OpenNLP so Lucene/Solr can call them natively. Prototype supports named entity recognition (token IDs, attention mask, token type IDs) and document classification; example model outputs a float array of length 9 for IOB labels. Implementation runs in a container, produces Solr fields (sentiment/NER), and the repo and Lucene integration are available and working.
AI-written summary · about
chapters (9)
- 0:00Start
- 1:03Search Relevance Engineer
- 6:20Released Open Nlp 2 0
- 7:07In Apache Open Nlp 2 0 the Goal Is To Modernize Apache Open Nlp
- 14:13Named Entity Recognition Model
- 16:46Inputs to the Model
- 18:20Attention Mask
- 25:13Integrate into Apache Opennlp
- 36:15Are Is There any Performance Hit by Converting the Model from Python to Onyx
related talks
20mUsing Apache OpenNLP with OpenSearch K-NN Vector Search
45mLeverage Power of Machine Learning with ONNX
05mONNX Runtime
31mNatural Language Processing & Java
57mMachine Learning with JavaScript
27mNatural Language Processing with Ruby
47mNeural Search Comes to Apache Solr: Approximate Nearest Neighbor, BERT & more
41mLeverage Power of Machine Learning with ONNX