GraphReasoning is a project that utilizes generative AI techniques to transform 1,000 scientific papers into knowledge graphs. Through structured analysis, it computes node degrees, identifies communities and connectivity, evaluates clustering coefficients and betweenness centrality of key nodes, revealing a fascinating knowledge architecture. The graph exhibits scale-free properties and a high degree of interconnectivity, facilitating graph reasoning that leverages transitivity and isomorphism to uncover unprecedented interdisciplinary relationships for answering questions, identifying knowledge gaps, proposing innovative material designs, and predicting material behaviors.

LongCite is an open-source model that trains large language models (LLMs) to produce accurate responses and precise sentence-level citations in long text question-answering scenarios. Its significance lies in enhancing the accuracy and credibility of question-answering systems, allowing users to verify the sources of the output information. LongCite supports context lengths of up to 128K and provides two models: LongCite-glm4-9b and LongCite-llama3.1-8b, trained respectively on GLM-4-9B and Meta-Llama-3.1-8B.

knowledge_graph_maker is a Python library that converts any text into a knowledge graph based on a given ontology. A knowledge graph is a semantic network representing the connections between real-world entities and their relationships. This library assists users in analyzing textual content in depth through graph algorithms and centrality calculations, enabling connectivity analysis between concepts, and enhances communication with text through Graph Retrieval Augmentation (GRAG) techniques.

llm-graph-builder is an application that utilizes large language models (like OpenAI, Gemini, etc.) to extract nodes, relationships, and their attributes from unstructured data (PDFs, DOCS, TXTs, YouTube videos, webpages, etc.) and uses the Langchain framework to create structured knowledge graphs. It supports uploading files from local machines, GCS or S3 buckets, or network resources, selecting an LLM model, and generating knowledge graphs.

Knowledge Graph RAG is an open-source Python library that leverages knowledge graphs and document networks to improve the performance of large language models (LLMs). This library enables users to search and connect information through graph structures, providing richer context for language models. It is widely used in the natural language processing field, particularly in document retrieval and information extraction tasks.

ChatQA is a series of conversational question answering (QA) models that can achieve GPT-4 level accuracy. We propose a two-stage instruction tuning method that can significantly improve the zero-shot conversational QA results of large language models (LLMs). To handle retrieval in dialogue-based QA, we fine-tuned dense retrievers on multi-turn QA datasets, which provides comparable results to using state-of-the-art query rewrite models while significantly reducing deployment costs. Notably, our ChatQA-70B outperforms GPT-4 (54.14 vs. 53.90) on the average score across 10 conversational QA datasets without relying on any synthetic data from the OpenAI GPT model.

KG-RAG is a task-agnostic framework that combines the explicit knowledge from a knowledge graph with the implicit knowledge of a large language model. Here, we utilize a massive biomedical knowledge graph SPOKE as the provider of biomedical context. A key feature of KG-RAG is its ability to extract 'prompt-relevant context' from the SPOKE knowledge graph, defined as the minimal context required to respond to a user prompt.

GraphReasoning is a project that utilizes generative AI techniques to transform 1,000 scientific papers into knowledge graphs. Through structured analysis, it computes node degrees, identifies communities and connectivity, evaluates clustering coefficients and betweenness centrality of key nodes, revealing a fascinating knowledge architecture. The graph exhibits scale-free properties and a high degree of interconnectivity, facilitating graph reasoning that leverages transitivity and isomorphism to uncover unprecedented interdisciplinary relationships for answering questions, identifying knowledge gaps, proposing innovative material designs, and predicting material behaviors.

LongCite is an open-source model that trains large language models (LLMs) to produce accurate responses and precise sentence-level citations in long text question-answering scenarios. Its significance lies in enhancing the accuracy and credibility of question-answering systems, allowing users to verify the sources of the output information. LongCite supports context lengths of up to 128K and provides two models: LongCite-glm4-9b and LongCite-llama3.1-8b, trained respectively on GLM-4-9B and Meta-Llama-3.1-8B.

knowledge_graph_maker is a Python library that converts any text into a knowledge graph based on a given ontology. A knowledge graph is a semantic network representing the connections between real-world entities and their relationships. This library assists users in analyzing textual content in depth through graph algorithms and centrality calculations, enabling connectivity analysis between concepts, and enhances communication with text through Graph Retrieval Augmentation (GRAG) techniques.

llm-graph-builder is an application that utilizes large language models (like OpenAI, Gemini, etc.) to extract nodes, relationships, and their attributes from unstructured data (PDFs, DOCS, TXTs, YouTube videos, webpages, etc.) and uses the Langchain framework to create structured knowledge graphs. It supports uploading files from local machines, GCS or S3 buckets, or network resources, selecting an LLM model, and generating knowledge graphs.

Knowledge Graph RAG is an open-source Python library that leverages knowledge graphs and document networks to improve the performance of large language models (LLMs). This library enables users to search and connect information through graph structures, providing richer context for language models. It is widely used in the natural language processing field, particularly in document retrieval and information extraction tasks.

ChatQA is a series of conversational question answering (QA) models that can achieve GPT-4 level accuracy. We propose a two-stage instruction tuning method that can significantly improve the zero-shot conversational QA results of large language models (LLMs). To handle retrieval in dialogue-based QA, we fine-tuned dense retrievers on multi-turn QA datasets, which provides comparable results to using state-of-the-art query rewrite models while significantly reducing deployment costs. Notably, our ChatQA-70B outperforms GPT-4 (54.14 vs. 53.90) on the average score across 10 conversational QA datasets without relying on any synthetic data from the OpenAI GPT model.

KG-RAG is a task-agnostic framework that combines the explicit knowledge from a knowledge graph with the implicit knowledge of a large language model. Here, we utilize a massive biomedical knowledge graph SPOKE as the provider of biomedical context. A key feature of KG-RAG is its ability to extract 'prompt-relevant context' from the SPOKE knowledge graph, defined as the minimal context required to respond to a user prompt.

Perplexity Copilot leverages powerful AI models like GPT-4 and Claude 2 to provide in-depth answers. It engages in conversations, asks questions, and listens attentively, constantly adjusting and refining search results to meet your specific needs, ultimately delivering answers that are tailored to your requirements.