DuoAttention is a framework designed to optimize long-context large language models (LLMs) by reducing memory and latency during inference without compromising their long-context abilities. It introduces a concept of Retrieval Heads and Streaming Heads to efficiently manage attention across tokens. By applying a full Key and Value (KV) cache to retrieval heads and a lightweight, constant-length KV cache to streaming heads, DuoAttention achieves significant reductions in memory usage and decoding time for LLMs. The framework uses an optimization-based algorithm with synthetic data to accurately identify retrieval heads, enabling efficient inference with minimal accuracy loss compared to full attention. DuoAttention also supports quantization techniques for further memory optimization, allowing for decoding of up to 3.3 million tokens on a single GPU.

This repository contains the training and inference code needed to replicate the winning solution to the AI Mathematical Olympiad - Progress Prize 1. It consists of fine-tuning DeepSeekMath-Base 7B, high-quality training datasets, a self-consistency decoding algorithm, and carefully chosen validation sets. The training methodology involves Chain of Thought (CoT) and Tool Integrated Reasoning (TIR) training stages. Two datasets, NuminaMath-CoT and NuminaMath-TIR, were used to fine-tune the models. The models were trained using open-source libraries like TRL, PyTorch, vLLM, and DeepSpeed. Post-training quantization to 8-bit precision was done to improve performance on Kaggle's T4 GPUs. The project structure includes scripts for training, quantization, and inference, along with necessary installation instructions and hardware/software specifications.

Atom is an accurate low-bit weight-activation quantization algorithm that combines mixed-precision, fine-grained group quantization, dynamic activation quantization, KV-cache quantization, and efficient CUDA kernels co-design. It introduces a low-bit quantization method, Atom, to maximize Large Language Models (LLMs) serving throughput with negligible accuracy loss. The codebase includes evaluation of perplexity and zero-shot accuracy, kernel benchmarking, and end-to-end evaluation. Atom significantly boosts serving throughput by using low-bit operators and reduces memory consumption via low-bit quantization.

Rig is a Rust library designed for building scalable, modular, and user-friendly applications powered by large language models (LLMs). It provides full support for LLM completion and embedding workflows, offers simple yet powerful abstractions for LLM providers like OpenAI and Cohere, as well as vector stores such as MongoDB and in-memory storage. With Rig, users can easily integrate LLMs into their applications with minimal boilerplate code.

Towards Generalizable Vision-Language Robotic Manipulation: A Benchmark and LLM-guided 3D Policy is a research project focusing on addressing the challenge of generalizing language-conditioned robotic policies to new tasks. The project introduces GemBench, a benchmark to evaluate the generalization capabilities of vision-language robotic manipulation policies. It also presents the 3D-LOTUS approach, which leverages rich 3D information for action prediction conditioned on language. Additionally, the project introduces 3D-LOTUS++, a framework that integrates 3D-LOTUS's motion planning capabilities with the task planning capabilities of LLMs and the object grounding accuracy of VLMs to achieve state-of-the-art performance on novel tasks in robotic manipulation.

Atropos is a robust and scalable framework for Reinforcement Learning Environments with Large Language Models (LLMs). It provides a flexible platform to accelerate LLM-based RL research across diverse interactive settings. Atropos supports multi-turn and asynchronous RL interactions, integrates with various inference APIs, offers a standardized training interface for experimenting with different RL algorithms, and allows for easy scalability by launching more environment instances. The framework manages diverse environment types concurrently for heterogeneous, multi-modal training.

Supallm is a Python library for super resolution of images using deep learning techniques. It provides pre-trained models for enhancing image quality by increasing resolution. The library is easy to use and allows users to upscale images with high fidelity and detail. Supallm is suitable for tasks such as enhancing image quality, improving visual appearance, and increasing the resolution of low-quality images. It is a valuable tool for researchers, photographers, graphic designers, and anyone looking to enhance image quality using AI technology.

A fine-tuning guide for both OpenAI and Open-Source Large Language Models on Azure. Fine-Tuning retrains an existing pre-trained LLM using example data, resulting in a new 'custom' fine-tuned LLM optimized for task-specific examples. Use cases include improving LLM performance on specific tasks and introducing information not well represented by the base LLM model. Suitable for cases where latency is critical, high accuracy is required, and clear evaluation metrics are available. Learning path includes labs for fine-tuning GPT and Llama2 models via Dashboards and Python SDK.

Model2Vec is a technique to turn any sentence transformer into a really small static model, reducing model size by 15x and making the models up to 500x faster, with a small drop in performance. It outperforms other static embedding models like GLoVe and BPEmb, is lightweight with only `numpy` as a major dependency, offers fast inference, dataset-free distillation, and is integrated into Sentence Transformers, txtai, and Chonkie. Model2Vec creates powerful models by passing a vocabulary through a sentence transformer model, reducing dimensionality using PCA, and weighting embeddings using zipf weighting. Users can distill their own models or use pre-trained models from the HuggingFace hub. Evaluation can be done using the provided evaluation package. Model2Vec is licensed under MIT.

Arch is an intelligent Layer 7 gateway designed to protect, observe, and personalize AI agents with APIs. It handles tasks related to prompts, including detecting jailbreak attempts, calling backend APIs, routing between LLMs, and managing observability. Built on Envoy Proxy, it offers features like function calling, prompt guardrails, traffic management, and observability. Users can build fast, observable, and personalized AI agents using Arch to improve speed, security, and personalization of GenAI apps.

OpenAdapt is an open-source software adapter between Large Multimodal Models (LMMs) and traditional desktop and web Graphical User Interfaces (GUIs). It aims to automate repetitive GUI workflows by leveraging the power of LMMs. OpenAdapt records user input and screenshots, converts them into tokenized format, and generates synthetic input via transformer model completions. It also analyzes recordings to generate task trees and replay synthetic input to complete tasks. OpenAdapt is model agnostic and generates prompts automatically by learning from human demonstration, ensuring that agents are grounded in existing processes and mitigating hallucinations. It works with all types of desktop GUIs, including virtualized and web, and is open source under the MIT license.

Hyper-Pretrained Transformers (HPT) is a novel multimodal LLM framework from HyperGAI, trained for vision-language models capable of understanding both textual and visual inputs. The repository contains the open-source implementation of inference code to reproduce the evaluation results of HPT Air on different benchmarks. HPT has achieved competitive results with state-of-the-art models on various multimodal LLM benchmarks. It offers models like HPT 1.5 Air and HPT 1.0 Air, providing efficient solutions for vision-and-language tasks.

Revornix is an information management tool designed for the AI era. It allows users to conveniently integrate all visible information and generates comprehensive reports at specific times. The tool offers cross-platform availability, all-in-one content aggregation, document transformation & vectorized storage, native multi-tenancy, localization & open-source features, smart assistant & built-in MCP, seamless LLM integration, and multilingual & responsive experience for users.

LLM-Drop is an official implementation of the paper 'What Matters in Transformers? Not All Attention is Needed'. The tool investigates redundancy in transformer-based Large Language Models (LLMs) by analyzing the architecture of Blocks, Attention layers, and MLP layers. It reveals that dropping certain Attention layers can enhance computational and memory efficiency without compromising performance. The tool provides a pipeline for Block Drop and Layer Drop based on LLaMA-Factory, and implements quantization using AutoAWQ and AutoGPTQ.

Codellm-devkit (CLDK) is a Python library that serves as a multilingual program analysis framework bridging traditional static analysis tools and Large Language Models (LLMs) specialized for code (CodeLLMs). It simplifies the process of analyzing codebases across multiple programming languages, enabling the extraction of meaningful insights and facilitating LLM-based code analysis. The library provides a unified interface for integrating outputs from various analysis tools and preparing them for effective use by CodeLLMs. Codellm-devkit aims to enable the development and experimentation of robust analysis pipelines that combine traditional program analysis tools and CodeLLMs, reducing friction in multi-language code analysis and ensuring compatibility across different tools and LLM platforms. It is designed to seamlessly integrate with popular analysis tools like WALA, Tree-sitter, LLVM, and CodeQL, acting as a crucial intermediary layer for efficient communication between these tools and CodeLLMs. The project is continuously evolving to include new tools and frameworks, maintaining its versatility for code analysis and LLM integration.

Dispider is an implementation enabling real-time interactions with streaming videos, providing continuous feedback in live scenarios. It separates perception, decision-making, and reaction into asynchronous modules, ensuring timely interactions. Dispider outperforms VideoLLM-online on benchmarks like StreamingBench and excels in temporal reasoning. The tool requires CUDA 11.8 and specific library versions for optimal performance.

DuoAttention is a framework designed to optimize long-context large language models (LLMs) by reducing memory and latency during inference without compromising their long-context abilities. It introduces a concept of Retrieval Heads and Streaming Heads to efficiently manage attention across tokens. By applying a full Key and Value (KV) cache to retrieval heads and a lightweight, constant-length KV cache to streaming heads, DuoAttention achieves significant reductions in memory usage and decoding time for LLMs. The framework uses an optimization-based algorithm with synthetic data to accurately identify retrieval heads, enabling efficient inference with minimal accuracy loss compared to full attention. DuoAttention also supports quantization techniques for further memory optimization, allowing for decoding of up to 3.3 million tokens on a single GPU.

Edgen is a local GenAI API server that serves as a drop-in replacement for OpenAI's API. It provides multi-endpoint support for chat completions and speech-to-text, is model agnostic, offers optimized inference, and features model caching. Built in Rust, Edgen is natively compiled for Windows, MacOS, and Linux, eliminating the need for Docker. It allows users to utilize GenAI locally on their devices for free and with data privacy. With features like session caching, GPU support, and support for various endpoints, Edgen offers a scalable, reliable, and cost-effective solution for running GenAI applications locally.

EasyDist is an automated parallelization system and infrastructure designed for multiple ecosystems. It offers usability by making parallelizing training or inference code effortless with just a single line of change. It ensures ecological compatibility by serving as a centralized source of truth for SPMD rules at the operator-level for various machine learning frameworks. EasyDist decouples auto-parallel algorithms from specific frameworks and IRs, allowing for the development and benchmarking of different auto-parallel algorithms in a flexible manner. The architecture includes MetaOp, MetaIR, and the ShardCombine Algorithm for SPMD sharding rules without manual annotations.

Since the emergence of chatGPT in 2022, the acceleration of Large Language Model has become increasingly important. Here is a list of papers on LLMs inference and serving.

LLM-Viewer is a tool for visualizing Language and Learning Models (LLMs) and analyzing performance on different hardware platforms. It enables network-wise analysis, considering factors such as peak memory consumption and total inference time cost. With LLM-Viewer, users can gain valuable insights into LLM inference and performance optimization. The tool can be used in a web browser or as a command line interface (CLI) for easy configuration and visualization. The ongoing project aims to enhance features like showing tensor shapes, expanding hardware platform compatibility, and supporting more LLMs with manual model graph configuration.

ServerlessLLM is a fast, affordable, and easy-to-use library designed for multi-LLM serving, optimized for environments with limited GPU resources. It supports loading various leading LLM inference libraries, achieving fast load times, and reducing model switching overhead. The library facilitates easy deployment via Ray Cluster and Kubernetes, integrates with the OpenAI Query API, and is actively maintained by contributors.

Sarathi-Serve is the official OSDI'24 artifact submission for paper #444, focusing on 'Taming Throughput-Latency Tradeoff in LLM Inference'. It is a research prototype built on top of CUDA 12.1, designed to optimize throughput-latency tradeoff in Large Language Models (LLM) inference. The tool provides a Python environment for users to install and reproduce results from the associated experiments. Users can refer to specific folders for individual figures and are encouraged to cite the paper if they use the tool in their work.

Aphrodite is the official backend engine for PygmalionAI, serving as the inference endpoint for the website. It allows serving Hugging Face-compatible models with fast speeds. Features include continuous batching, efficient K/V management, optimized CUDA kernels, quantization support, distributed inference, and 8-bit KV Cache. The engine requires Linux OS and Python 3.8 to 3.12, with CUDA >= 11 for build requirements. It supports various GPUs, CPUs, TPUs, and Inferentia. Users can limit GPU memory utilization and access full commands via CLI.

Awesome Mamba is a curated collection of groundbreaking research papers and articles on Mamba Architecture, a pioneering framework in deep learning known for its selective state spaces and efficiency in processing complex data structures. The repository offers a comprehensive exploration of Mamba architecture through categorized research papers covering various domains like visual recognition, speech processing, remote sensing, video processing, activity recognition, image enhancement, medical imaging, reinforcement learning, natural language processing, 3D recognition, multi-modal understanding, time series analysis, graph neural networks, point cloud analysis, and tabular data handling.

The 'unilm' repository is a collection of tools, models, and architectures for Foundation Models and General AI, focusing on tasks such as NLP, MT, Speech, Document AI, and Multimodal AI. It includes various pre-trained models, such as UniLM, InfoXLM, DeltaLM, MiniLM, AdaLM, BEiT, LayoutLM, WavLM, VALL-E, and more, designed for tasks like language understanding, generation, translation, vision, speech, and multimodal processing. The repository also features toolkits like s2s-ft for sequence-to-sequence fine-tuning and Aggressive Decoding for efficient sequence-to-sequence decoding. Additionally, it offers applications like TrOCR for OCR, LayoutReader for reading order detection, and XLM-T for multilingual NMT.

LLM App Stack, also known as Emerging Architectures for LLM Applications, is a comprehensive list of available tools, projects, and vendors at each layer of the LLM app stack. It covers various categories such as Data Pipelines, Embedding Models, Vector Databases, Playgrounds, Orchestrators, APIs/Plugins, LLM Caches, Logging/Monitoring/Eval, Validators, LLM APIs (proprietary and open source), App Hosting Platforms, Cloud Providers, and Opinionated Clouds. The repository aims to provide a detailed overview of tools and projects for building, deploying, and maintaining enterprise data solutions, AI models, and applications.

Awesome deep logic is a curated list of papers and resources focusing on integrating symbolic logic into deep neural networks. It includes surveys, tutorials, and research papers that explore the intersection of logic and deep learning. The repository aims to provide valuable insights and knowledge on how logic can be used to enhance reasoning, knowledge regularization, weak supervision, and explainability in neural networks.

Welcome to the ultimate hub for on-device Large Language Models (LLMs)! This repository is your go-to resource for all things related to LLMs designed for on-device deployment. Whether you're a seasoned researcher, an innovative developer, or an enthusiastic learner, this comprehensive collection of cutting-edge knowledge is your gateway to understanding, leveraging, and contributing to the exciting world of on-device LLMs.

DuoAttention is a framework designed to optimize long-context large language models (LLMs) by reducing memory and latency during inference without compromising their long-context abilities. It introduces a concept of Retrieval Heads and Streaming Heads to efficiently manage attention across tokens. By applying a full Key and Value (KV) cache to retrieval heads and a lightweight, constant-length KV cache to streaming heads, DuoAttention achieves significant reductions in memory usage and decoding time for LLMs. The framework uses an optimization-based algorithm with synthetic data to accurately identify retrieval heads, enabling efficient inference with minimal accuracy loss compared to full attention. DuoAttention also supports quantization techniques for further memory optimization, allowing for decoding of up to 3.3 million tokens on a single GPU.

LLM notes repository contains detailed analysis on transformer models, language model compression, inference and deployment, high-performance computing, and system optimization methods. It includes discussions on various algorithms, frameworks, and performance analysis related to large language models and high-performance computing. The repository serves as a comprehensive resource for understanding and optimizing language models and computing systems.

A curated list of high-quality papers on resource-efficient Large Language Models (LLMs) with a focus on various aspects such as architecture design, pre-training, fine-tuning, inference, system design, and evaluation metrics. The repository covers topics like efficient transformer architectures, non-transformer architectures, memory efficiency, data efficiency, model compression, dynamic acceleration, deployment optimization, support infrastructure, and other related systems. It also provides detailed information on computation metrics, memory metrics, energy metrics, financial cost metrics, network communication metrics, and other metrics relevant to resource-efficient LLMs. The repository includes benchmarks for evaluating the efficiency of NLP models and references for further reading.