Projects sharing features with Fastllm

mini-sglang is a collection of tools for large language model inference, serving as an OpenAI-compatible inference server, a memory-efficient prefill engine, and a tensor parallelism runtime. It also functions as a local batch processing engine for offline benchmarking and ablation studies. The project focuses on acceleration and memory management through a KV cache manager that reuses precomputed caches for shared request prefixes. It handles large model workloads by distributing tasks across multiple GPUs and manages peak memory consumption by splitting long input sequences into smaller chu

ipex-llm is an acceleration library and inference engine designed to optimize the execution and finetuning of large language models on Intel GPUs and NPUs. It provides a HuggingFace compatible model backend and a dedicated quantization toolkit for converting model weights into low-bit precision formats. The project facilitates distributed inference by splitting large model workloads across multiple accelerators using pipeline and tensor parallelism. It enables the deployment of models on Intel Arc, Flex, and Max GPUs to increase throughput and reduce latency. The library covers a broad range

gpt-fast is a PyTorch transformer inference engine designed for text generation using a native tensor library implementation. It provides a runtime for executing large language models without the need for external C++ extensions. The project implements speculative decoding to accelerate generation by using a small draft model for token prediction and a larger model for verification. It further optimizes performance through a compiled prefill stage and a multi-GPU tensor parallelism library that shards linear layers across multiple graphics processing units. Memory efficiency is managed throu

Tengine is a suite of tooling and a lightweight execution engine designed for running deep learning models on constrained embedded hardware. It provides an infrastructure for converting neural network models, quantizing weights, optimizing operator kernels, and benchmarking inference performance across CPU, GPU, and NPU units. The project features an automated operator kernel optimizer to generate high-efficiency kernels and a model quantization tool that reduces precision to integer formats to lower memory usage. It includes a dedicated hardware benchmarking tool to evaluate the execution sp

Metaseq is a transformer sequence modeling toolkit designed for training, fine-tuning, and deploying sequence-to-sequence models using open pre-trained weights. It provides a comprehensive framework for large language model training, including dedicated tools for sequence dataset processing and a standalone inference server for generating text via API requests. The project features specialized utilities for model quantization to reduce parameter precision to eight bits, which lowers memory usage and increases inference speed. It also includes a checkpoint conversion pipeline to transform mode

Paddle-Lite is a deep learning inference engine and edge computing runtime designed to execute trained models on mobile and edge devices. It provides a hardware-accelerated inference framework and a decoupled runtime with a minimal binary footprint to operate in resource-constrained environments without third-party dependencies. The project includes a model quantization tool for reducing precision and size via static and dynamic quantization, as well as a computation graph optimizer. These tools reduce latency and memory usage by fusing operators and pruning the model intermediate representat

Lorax is a GPU-accelerated inference server and multi-adapter engine designed for serving large language models. It functions as a high-throughput system capable of deploying models via Kubernetes and managing the dynamic swapping of Low-Rank Adaptation adapters per request. The server distinguishes itself through multi-adapter dynamic batching, which allows requests using different adapter weights to be processed in a single GPU forward pass. It employs just-in-time adapter loading and weighted adapter merging to maximize throughput and enable multi-tasking without sacrificing performance.

jetson-inference is a set of libraries and tools for executing optimized deep learning models on embedded GPU hardware. Its primary purpose is to enable real-time computer vision and AI inference at the edge with low latency and high throughput. The project distinguishes itself through high-performance streaming analytics and the ability to execute concurrent AI pipelines on auto-grade silicon. It provides specialized support for multi-sensor stream processing, utilizing zero-copy data transport to load camera frames directly into GPU memory. The codebase covers a broad surface of capabiliti

TNN is a deep learning inference framework designed to execute pre-trained neural networks across mobile, desktop, and server hardware. It functions as a hardware-accelerated runtime and model compression toolkit, providing a unified interface for deploying models in diverse environments. The framework includes an ONNX model converter to transform models from various training frameworks into a standardized internal format. It distinguishes itself through a combination of model compression tools—including weight quantization and static-code pruning—and a memory management system that reuses bu

FasterTransformer is a high-performance inference optimization library and distributed runtime designed to accelerate the execution of transformer models. It provides a toolkit for reducing model precision and parallelizing execution across multiple GPUs to increase throughput and reduce latency for large language models. The framework utilizes a C++ backend with custom CUDA kernels to replace generic operations with optimized GPU instructions. It implements tensor and pipeline parallelism to shard model weights and distribute compute operations across multiple devices. The system includes c

DeepSpeed is a distributed deep learning optimization library and framework designed for the training and inference of massive AI models. It serves as a model parallelism orchestrator and a toolkit for scaling large language models across multiple GPUs and compute nodes. The project distinguishes itself through 3D parallelism orchestration, which combines data, pipeline, and tensor parallelism. It utilizes ZeRO-based memory partitioning to eliminate redundant storage and employs CPU-offload memory management to move weights and optimizer states to system RAM. Additionally, it provides special

Qwen-7B is a pretrained causal language model designed for natural language generation, text processing, and complex reasoning tasks. It is available as an instruction-tuned model optimized for conversational interactions and a tool-use model capable of executing function calls and interacting with external APIs. The project provides a quantized version of the model to reduce GPU memory usage and supports the development of autonomous agents that can execute code and perform functions to complete complex goals. The system covers a wide range of capabilities including model fine-tuning throug

Sglang is a high-performance inference engine and serving system designed for large language and multimodal models. It provides a programmable interface for orchestrating complex generation workflows, enabling developers to coordinate multi-turn dialogues, tool invocations, and reasoning chains through a domain-specific language. The platform is built to support production-scale deployments, offering an OpenAI-compatible API that allows for integration with existing application ecosystems. The system distinguishes itself through a disaggregated architecture that separates compute-intensive pr

vllm-omni is a high-throughput serving engine and distributed inference framework designed for omni-modal models. It serves as a multi-modal model API server capable of generating text, image, video, and audio data, providing a standardized interface for remote client access. The system features a non-autoregressive generation engine for parallel media production and a robot policy inference server that acts as a real-time communication bridge to robotic hardware using specialized protocols. It supports hybrid execution models that combine sequential token generation with parallelized media g

LightLLM is a high-performance serving framework for deploying and executing large language models. It functions as a multi-GPU inference engine and server capable of handling dense architectures, mixture-of-experts designs, and multimodal models that process both text and images. The system is distinguished by its specialized support for Mixture-of-Experts models using expert parallelism and fused kernels. It implements structured text generation through deterministic state machines and pushdown automata to enforce precise output formats. To optimize throughput, the framework employs specula

ggml is a low-level C++ tensor library and machine learning inference engine designed for performing mathematical operations on multi-dimensional arrays across diverse hardware platforms. It provides a foundational toolset for executing machine learning models and calculating mathematical gradients through an automatic differentiation library. The project features a quantized tensor framework that converts floating-point weights into integer representations to reduce memory usage and increase inference speed. It utilizes a custom binary format for model serialization to ensure rapid loading a

gpt-fast is a PyTorch transformer inference engine designed for low-latency text generation. It functions as a distributed GPU inference library, a quantized model runner, and a speculative decoding framework. The system utilizes a speculative decoding workflow where a small draft model predicts token sequences for verification by a larger model to accelerate generation. It supports quantized model execution to reduce memory footprint and implements tensor parallelism to split computations across multiple GPUs. The project includes a standardized evaluation harness to measure the accuracy an

LitGPT is a training and deployment framework for large language models, providing a suite of tools for pretraining, finetuning, quantizing, evaluating, and serving models within a production environment. It includes a dedicated training pipeline for adapting pretrained models to specific tasks, a quantization tool for reducing weight precision, and an inference server for hosting models via web interfaces. The framework supports high-performance model development through custom architecture implementation and the use of predefined recipes to standardize pretraining and finetuning. It enables

SynapseML is an Apache Spark machine learning library designed for building and scaling machine learning workflows and data pipelines across distributed clusters. It serves as a distributed machine learning pipeline framework and a distributed inference engine for executing hardware-accelerated predictions and deep learning tasks on large-scale datasets. The project functions as a cloud AI integration layer, allowing users to apply pretrained artificial intelligence services for text, vision, and speech within distributed pipelines. It also includes a dedicated suite of tools for distributed

Neural Compressor is a deep learning model compression toolkit and AI inference acceleration engine. It functions as an automated model quantization tool and hardware-aware model compiler designed to reduce the memory footprint of neural networks and decrease execution latency. The project provides specialized frameworks for optimizing large language models, utilizing weight-only quantization and hardware-specific kernels to improve the operational efficiency of generative AI workloads. It maps neural network operators to specialized CPU and GPU vector instructions to accelerate model executi

This project is a comprehensive technical course study guide and reference for learning the architectures and training methods of Transformers and large language models. It serves as a technical overview for understanding how neural networks process data and how to align model behavior with specific performance goals. The repository provides specialized guides on several key areas of model development. This includes detailed references for transformer architectures, implementation frameworks for retrieval-augmented generation and agentic workflows, and technical guides for model optimization

BigDL is a PyTorch acceleration framework and distributed inference engine designed for large language models. It provides a toolkit for running models on Intel hardware, integrating quantization tools and libraries for parameter-efficient fine-tuning. The project distinguishes itself through the use of pipeline parallelism to distribute model workloads across multiple hardware accelerators. It utilizes low-bit integer quantization and speculative decoding to reduce memory footprints and decrease text generation latency. The system covers broad capabilities in model optimization, including w

Distributed-llama is a distributed inference engine and command line tool for running large language models across multiple networked machines. It functions as a compute cluster manager that coordinates worker nodes to share the computational load of a single model. The system utilizes tensor parallelism to shard model weights across different hosts, allowing the execution of models that exceed the memory capacity of a single piece of hardware. It includes a dedicated format converter to transform standard model files into a compatible binary layout optimized for distributed loading. The eng

llama-rs is a local large language model inference engine implemented in Rust. It enables the execution of model computations on local hardware to generate text responses from user prompts. The project utilizes Rust-based tensor operations and direct-memory model mapping to handle high-performance linear algebra and efficient weight loading. It incorporates weight quantization to reduce the memory footprint of models by converting high-precision weights into smaller formats. The system includes a command-line interface for interactive chat sessions and one-off prompts, along with file-backed

gpt-oss is an open-weight large language model and reasoning engine designed for complex reasoning and agentic workflows. It functions as an AI agent framework and model serving API, allowing for local deployment and the hosting of standardized interfaces to expose model completions and internal reasoning processes. The project distinguishes itself as a quantized inference engine, utilizing tensor parallelism and weight quantization to run high-parameter models on limited hardware. It features a reasoning model that employs chain-of-thought processing to solve multi-step logical tasks. The s

llm-compressor is a quantization toolkit and post-training library designed to reduce the memory footprint and size of large language models. It provides a framework for compressing models using weight and activation quantization to enable more efficient deployment. The project distinguishes itself through a distributed quantization framework that utilizes data-parallel processing and disk-based weight offloading to handle massive model checkpoints that exceed available system memory. It includes specialized compressors for diverse architectures, including Mixture-of-Experts, Vision-Language,

Grok-1 is an open-weights large language model implementation featuring a sparse mixture-of-experts architecture. It is designed for high-performance text generation and natural language processing by activating only a subset of specialized expert layers per token. The model utilizes 8-bit weight quantization to reduce memory overhead and accelerate loading. To manage its high parameter count, the implementation supports activation sharding, which distributes the memory load across multiple hardware devices during execution. The project covers large-scale model inference, including text comp