# Retrieval-Augmented Generation

TreeHop is a lightweight embedding-level framework designed for efficient query embedding generation and filtering in multi-hop QA, significantly reducing computational overhead.

Qwen-2.5-7B-base-RAG-RL is a large language model with 7B parameters trained from scratch on an unknown dataset, incorporating Retrieval-Augmented Generation (RAG) and Reinforcement Learning (RL) technologies.

A fine-tuned cross-encoder model based on jina-reranker-v2-base-multilingual for text reranking and semantic search

Pleias-RAG-350M is a 350-million-parameter compact reasoning model specifically trained for retrieval-augmented generation (RAG), search, and source summarization tasks.

ColNomic Embed Multimodal 3B is a 3-billion-parameter multimodal embedding model specifically designed for visual document retrieval tasks, supporting unified encoding of multilingual text and images.

This repository contains GGUF format model files of mixedbread-ai/mxbai-rerank-large-v2, quantized by TensorBlock's machine, compatible with llama.cpp.

This model is a GGUF format conversion of BAAI/bge-reranker-v2-m3, designed for text ranking tasks with multilingual support.

The first retriever specifically designed for financial time series forecasting, based on the Retrieval-Augmented Generation (RAG) framework

FinSeer StockLLM is an open-source 1-billion-parameter large language model specifically designed for financial time series forecasting, utilizing a Retrieval-Augmented Generation (RAG) framework.

Gen-8B-R2 is a generation model focused on reducing hallucination issues in RAG systems, particularly suitable for handling retrieval noise and information overload.

Trendyol LLM v4.1.0 is a generative model based on Trendyol LLM base v4.0 (a Qwen2.5 7B version further pre-trained on 13 billion tokens), specializing in e-commerce and Turkish language understanding.

Kanana is a bilingual (Korean/English) language model series developed by Kakao, excelling in Korean tasks while remaining competitive in English tasks, significantly reducing computational costs compared to models of similar scale.

A text chunker based on BertForTokenClassification, suitable for structured and unstructured text, especially optimized for RAG scenarios.

This model is the GGUF quantized version of BAAI/bge-reranker-v2-m3, designed for text reranking tasks with multilingual support.

Fluxi AI is a multimodal intelligent assistant based on Qwen2-VL-7B-Instruct, capable of processing text, images, and videos, with special optimization for Portuguese language support.

The LB Reranker is a model for determining the relevance between queries and text snippets, supporting 95+ languages, suitable for ranking and reranking in retrieval tasks.

Ruri is a general-purpose text embedding model optimized for Japanese, based on the Sentence Transformers architecture, specifically designed for sentence similarity computation and feature extraction tasks.

A high-performance re-ranking model optimized for Arabic document retrieval and sorting

This model is converted from BAAI/bge-reranker-v2-m3 into GGUF format for text reranking tasks, supporting multiple languages.

Vikhr-Nemo is a bilingual large language model optimized based on Mistral-Nemo-Instruct-2407, specifically designed for Russian and English, supporting various tasks such as logical reasoning, text summarization, and code generation.

This is the GGUF format conversion version of the BAAI/bge-reranker-v2-gemma model, compatible with the llama.cpp framework, supporting multilingual text classification tasks.

DataGemma is a series of models fine-tuned based on Gemma 2, specifically designed to help large language models access and integrate reliable public statistical data in Data Commons.

A binary classification model for determining the relevance between reference text and questions, optimized for RAG systems

LLM2Vec is a simple solution for converting decoder-only large language models into text encoders, achieved by enabling bidirectional attention mechanisms, masked next-token prediction, and unsupervised contrastive learning.

Llama3-ChatQA-1.5 is a conversational QA and retrieval-augmented generation model developed based on the Llama-3 foundation model, excelling in conversational QA and RAG tasks.

GTE-base-en-v1.5 is an English sentence transformer model focused on sentence similarity tasks, excelling in multiple text embedding benchmarks.

OpenThaiGPT 70b 1.0.0 is a 70-billion-parameter Thai chat model based on LLaMA v2, specifically fine-tuned for Thai language instructions, significantly enhancing Thai language processing capabilities.

LLM2Vec is a simple solution for transforming decoder-only large language models into text encoders, achieved by enabling bidirectional attention, masked next-token prediction, and unsupervised contrastive learning.

OpenThaiGPT 13b 1.0.0 is a 13-billion-parameter Thai chat model based on LLaMA v2, specifically optimized for Thai instructions

A lightweight language model for query rewriting, capable of parsing user input into structured formats to optimize retrieval effectiveness.

A text ranking model based on Mistral-7B, offering multiple quantized versions for efficient inference.

A text embedding model developed by Salesforce Research, trained on E5-mistral-7b-instruct and Mistral-7B-v0.1, primarily used for text retrieval tasks.

A French large language model optimized from Mistral Open-Hermes 2.5, specifically designed for Retrieval-Augmented Generation (RAG) with source attribution and explainability

Specialized LLM fine-tuned from 13B-parameter Orca-2, focused on biological and bioinspired materials science

WiNGPT2 is a large-scale medical vertical domain model based on the GPT architecture, dedicated to integrating professional medical knowledge, healthcare information, and data to provide intelligent medical Q&A, diagnostic support, and medical knowledge services for the industry.

ClimateGPT is a 7-billion-parameter AI model specifically designed to integrate interdisciplinary research on climate change, developed based on the Llama-2 framework with a focus on climate science.

ClimateGPT is a 70-billion-parameter large language model specialized in climate science, based on continued pre-training and instruction fine-tuning of Llama-2, focusing on interdisciplinary research integration in climate change

stella-base-en-v2 is an English text embedding model based on sentence-transformers, focusing on sentence similarity and feature extraction tasks.

WiNGPT2 is a large medical vertical domain model based on the GPT architecture, dedicated to integrating professional medical knowledge, healthcare information, and data to provide intelligent Q&A, diagnostic support, and medical knowledge services for the industry.

This is a multilingual sentence transformer model that maps sentences and paragraphs into a 768-dimensional dense vector space, supporting multiple languages and suitable for tasks like clustering or semantic search.