simcse

experiments of some semantic matching models and comparison of experimental results.

天池阿里灵杰问天引擎电商搜索算法赛非官方 baseline，又名 NLP 从入门到 22/2771。

文本相似度，语义向量，文本向量，text-similarity，similarity, sentence-similarity，BERT，SimCSE，BERT-Whitening，Sentence-BERT, PromCSE, SBERT

Transformer-based models implemented in tensorflow 2.x(using keras).

Exploring Japanese SimCSE

Implementation of "RankCSE: Unsupervised Sentence Representation Learning via Learning to Rank" (ACL 2023)

This is a hands-on for ML beginners to perform SimCSE step-by-step. Implemented both supervised SimCSE and unsupervisied SimCSE, and distributed training is possible with Amazon SageMaker.

Extract Molecular SMILES embeddings from language models pre-trained with various objectives architectures.

A TensorFlow 2 Keras implementation of SimCSE with unsupervised and supervised.

A Keras-based and TensorFlow-backend NLP Models Toolkit.

Reinforcement Calibration SimCSE, combining contrastive learning, artificial potential fields, perceptual loss, and RLHF to achieve improved Semantic Textual Similarity (STS) embeddings. PyTorch-based implementations of PerceptualBERT and ForceBasedInfoNCE, along with fine-tuning capabilities via RLHF and evaluation using SentEval.

Sentence Embeddings using Deep Nerual Networks in PRODUCTION!

CRLT: A Unified Contrastive Learning Toolkit for Unsupervised Text Representation Learning

LLM4uVis是一款针对历史人物不确定性数据，可以进行多轮LLM自动推理，并对回答结果进行综合聚类和可视分析的系统。它针对时间、地点、人物和事件四类核心不确定性数据，对LLM各轮回答之间的相似推理逻辑进行语义聚类，并支持以桑基图形式展现多条推理路径及其逻辑差异，并可以选择一条或多条具体的推理链条进行进一步语义理解或横/纵向对比分析，专家可以对推理过程与结果进行交叉验证与标注。

Data pipelines for both TensorFlow and PyTorch!

Domain adaption for an embedding model using unsupervised and supervised finetuning on scientific texts for the SciFact retrieval task.

This project presents a Graph Neural Network (GNN)-based framework for predicting Drug-Drug Interactions (DDIs) using pretrained SMILES embeddings and Graph Attention Networks (GAT). Given the limitations of clinical studies and traditional computational methods in detecting complex DDIs.

Text Embedding Model Fine-Tuning: Generating Data & Fine-Tuning Methods

Source code for the diploma thesis: Bias Measurement in LLM2Vec embeddings. Explores how transforming Causal Language Models (like Llama & Mistral) into text encoders affects the geometric structure and encoding of gender bias.

Topic clustering of Vietnamese YouTube videos using descriptions with PhoBERT/Sup-SimCSE embeddings, UMAP dimensionality reduction, HDBSCAN, and BERTopic. (Phân cụm chủ đề video YouTube tiếng Việt dựa trên mô tả, sử dụng embedding PhoBERT/Sup-SimCSE, UMAP, HDBSCAN và BERTopic.)