"Word2Vec for Russian text" - my project for course "Scientific Data Computing" in University of Tartu. It was presented as 20-minutes talk on 6th Estonian Digital Humanities Conference at September 2018.

NLP Dataset Creation and Semantic Search Demonstration

Retrieval-Augmented Generation using Azure OpenAI

Universal-Sentence-Encoder-Multilingual-QA is a model developed by researchers at Google mainly for the purpose of question answering. You can use this template to import the model in Inferless.

"BrightPsych" is a holistic mental health platform featuring a supportive chatbot and detail CBT analysis for disorders. Daily Mood Tracking aids emotional well-being, while data analysis unveils student mental health trends. Guided mindfulness contribute to resilience in a nurturing space. Empower, Engage and Elevate through Community Forum.

BGE-M3 is an innovative project known for its versatility, featuring Multi-Functionality, Multi-Linguality, and Multi-Granularity.

MedCPT generates embeddings of biomedical texts that can be used for semantic search (dense retrieval). MedCPT Query Encoder: compute the embeddings of short texts (e.g., questions, search queries, sentences). In this template, we will import the MedCPT Query Encoder on the Inferless Platform.

State-of-the-Art Ember embedding model for retrieval augmented generation

This is a sentence embedding model, initialized from xlm-roberta-large and continually trained on a mixture of multilingual datasets. It supports 100 languages from xlm-roberta, but low-resource languages may see performance degradation.

MS-marco-MiniLM-L-12-v2 model can be used for Information Retrieval: Given a query, encode the query will all possible passages (e.g. retrieved with ElasticSearch). Then sort the passages in a decreasing order.

Julia experimentation using sequence-based NLP models

OpenAI Text Embedding. Clean, process and create vectorize representation of text for indexing and semantic search

jina-embeddings-v2-base-en is an English, monolingual embedding model supporting 8192 sequence length. It is based on a BERT architecture (JinaBERT) that supports the symmetric bidirectional variant of ALiBi to allow longer sequence length. The backbone jina-bert-v2-base-en is pretrained on the C4 dataset.

Functionality of the Bot that fills a particular form on request from the user

Kickstarter project success or failure prediction. Using Word2Vec to train embedding file.

Using the OnionOrNot dataset from kaggle to train a binary classification model the Keras deep learning library.

Some applications of text embedding model, e.g., semantic retrieval and clustering.