Search Results for "compression"
Sort By:
Showing 42 open source projects for "compression"
View related business solutions-
Our Free Plans just got better! | Auth0You asked, we delivered! Auth0 is excited to expand our Free and Paid plans to include more options so you can focus on building, deploying, and scaling applications without having to worry about your security. Auth0 now, thank yourself later.
-
Compliant and Reliable File Transfers Backed by Top Security CertificationsStop relying on non-certified, legacy file transfer tools that creak under the weight of modern security demands. Get full audit trails, advanced access controls and more supported by an award-winning team of experts. Start your free 25-day trial today.
-
1
DeepSpeed
Deep learning optimization library: makes distributed training easy
...Achieve excellent system throughput and efficiently scale to thousands of GPUs 3. Train/Inference on resource constrained GPU systems 4. Achieve unprecedented low latency and high throughput for inference 5. Achieve extreme compression for an unparalleled inference latency and model size reduction with low costs DeepSpeed offers a confluence of system innovations, that has made large scale DL training effective, and efficient, greatly improved ease of use, and redefined the DL training landscape in terms of scale that is possible. These innovations such as ZeRO, 3D-Parallelism, DeepSpeed-MoE, ZeRO-Infinity, etc. fall under the training pillar. -
2
Wan2.2
Wan2.2: Open and Advanced Large-Scale Video Generative Model
...The model is trained on significantly larger datasets than its predecessor, greatly enhancing motion complexity, semantic understanding, and aesthetic diversity. Wan2.2 also open-sources a 5-billion parameter high-compression VAE-based hybrid text-image-to-video (TI2V) model that supports 720P video generation at 24fps on consumer-grade GPUs like the RTX 4090. It supports multiple video generation tasks including text-to-video. -
3
LLM Action
Technical principles related to large models
LLM-Action is a knowledge/tutorial/repository that shares principles, techniques, and real-world experience related to large language models (LLMs), focusing on LLM engineering, deployment, optimization, inference, compression, and tooling. It organizes content in domains like training, inference, compression, alignment, evaluation, pipelines, and applications. Sections covering infrastructure, engineering, and deployment. Repository templates, sample code, and resource links. Articles/code on LLM compression (quantization, pruning). -
4
NNCF
Neural Network Compression Framework for enhanced OpenVINO
NNCF (Neural Network Compression Framework) is an optimization toolkit for deep learning models, designed to apply quantization, pruning, and other techniques to improve inference efficiency. -
Ship Agents FasterGemini Enterprise Agent Platform lets you rapidly build, scale, govern and optimize production-ready agents grounded in your organization's data. The platform enables developers to build custom or pre-built agents for virtually any use case. New customers get $300 in free credits.
-
5
FramePack
Lets make video diffusion practical
...It’s useful for diffusion and generative models that learn from sequential image datasets, as well as classical pipelines that batch many related frames. With a simple API and examples, it invites experimentation on tradeoffs between compression, fidelity, and speed. -
6
KVCache-Factory
Unified KV Cache Compression Methods for Auto-Regressive Models
KVCache-Factory is an open-source research framework designed to explore and implement unified key-value cache compression techniques for autoregressive transformer models. In large language models, the key-value cache stores intermediate attention states that enable efficient token generation during inference, but these caches can consume large amounts of GPU memory when handling long contexts. KVCache-Factory provides a platform for implementing and evaluating multiple compression strategies that reduce memory usage while preserving model performance. ... -
7
Audiogen Codec
48khz stereo neural audio codec for general audio
AGC (Audiogen Codec) is a convolutional autoencoder based on the DAC architecture, which holds SOTA. We found that training with EMA and adding a perceptual loss term with CLAP features improved performance. These codecs, being low compression, outperform Meta's EnCodec and DAC on general audio as validated from internal blind ELO games. We trained (relatively) very low compression codecs in the pursuit of solving a core issue regarding general music and audio generation, low acoustic quality, and audible artifacts, which hinder industry use for these models. Our hope is to encourage researchers to build hierarchical generative audio models that can efficiently use high sequence length representations without sacrificing semantic abilities. -
8
TurboQuant+
Implementation of TurboQuant (ICLR 2026)
TurboQuant Plus is an extended and enhanced version of quantization tooling aimed at improving neural network efficiency through advanced compression and optimization strategies. It builds upon the concept of reducing model precision to accelerate inference while attempting to maintain or recover accuracy through refined techniques. The project explores additional enhancements such as improved calibration, adaptive quantization, and potentially hybrid precision approaches that combine multiple levels of compression. -
9
R-KV
Redundancy-aware KV Cache Compression for Reasoning Models
R-KV is an open-source research project that focuses on improving the efficiency of large language model inference through key-value cache compression techniques. Modern transformer models rely heavily on KV caches during autoregressive decoding, which store intermediate attention states to accelerate generation. However, these caches can consume large amounts of memory, especially in reasoning-oriented models with long context windows. R-KV introduces a method for compressing the KV cache during decoding, allowing models to maintain reasoning performance while reducing memory consumption and computational overhead. ... -
MongoDB Atlas runs apps anywhereMongoDB Atlas gives you the freedom to build and run modern applications anywhere—across AWS, Azure, and Google Cloud. With global availability in over 115 regions, Atlas lets you deploy close to your users, meet compliance needs, and scale with confidence across any geography.
-
10
MemPalace
The highest-scoring AI memory system ever benchmarked
...It operates fully locally using tools like ChromaDB, meaning it requires no API keys, cloud services, or external dependencies once installed. MemPalace emphasizes fidelity over compression, preserving full conversational history to maintain reasoning, nuance, and decision-making context that is typically lost in other systems. -
11
WavTokenizer
SOTA discrete acoustic codec models with 40/75 tokens per second
...It is built to represent speech, music, and general audio with extremely low bitrate, making it ideal as a front-end for large audio language models like GPT-4o and similar architectures. The model uses a single-quantizer design together with temporal compression to achieve extreme compression without sacrificing reconstruction fidelity. Its architecture incorporates a broader vector-quantization space, extended contextual windows, and improved attention networks, combined with multi-scale discriminators and inverse Fourier transform blocks to enhance waveform reconstruction. Extensive experiments show that WavTokenizer matches or surpasses previous neural codecs across speech, music, and general audio on both objective metrics and subjective listening tests. -
12
PaddleOCR
Awesome multilingual OCR toolkits based on PaddlePaddle
...It features a PPOCR series of high-quality pre-trained models, which includes: ultra lightweight ppocr_mobile series models, general ppocr_server series models, and ultra lightweight compression ppocr_mobile_slim series models. PaddleOCR is easy to install and easy to use on Windows, Linux, MacOS and other systems. -
13
LMDeploy
LMDeploy is a toolkit for compressing, deploying, and serving LLMs
LMDeploy is a toolkit designed for compressing, deploying, and serving large language models (LLMs). It offers tools and workflows to optimize LLMs for production environments, ensuring efficient performance and scalability. LMDeploy supports various model architectures and provides deployment solutions across different platforms. -
14
Claw Compactor
14-stage Fusion Pipeline for LLM token compression
Claw Compactor is a utility designed to optimize and manage the context limitations inherent in AI agent systems, particularly those built on OpenClaw-like architectures. It addresses the challenge of finite context windows in language models by compressing or summarizing historical interactions while preserving essential information. The system works by transforming older conversation data into condensed representations that maintain continuity without exceeding token limits. This approach... -
15
Deep Lake
Data Lake for Deep Learning. Build, manage, and query datasets
...Use one API to upload, download, and stream datasets to/from AWS S3/S3-compatible storage, GCP, Activeloop cloud, or local storage. Store images, audios and videos in their native compression. Deeplake automatically decompresses them to raw data only when needed, e.g., when training a model. Treat your cloud datasets as if they are a collection of NumPy arrays in your system's memory. Slice them, index them, or iterate through them. -
16
AIMET
AIMET is a library that provides advanced quantization and compression
...QuIC has a mission to help migrate the ecosystem toward fixed-point inference. With this goal, QuIC presents the AI Model Efficiency Toolkit (AIMET) - a library that provides advanced quantization and compression techniques for trained neural network models. AIMET enables neural networks to run more efficiently on fixed-point AI hardware accelerators. Quantized inference is significantly faster than floating point inference. For example, models that we’ve run on the Qualcomm® Hexagon™ DSP rather than on the Qualcomm® Kryo™ CPU have resulted in a 5x to 15x speedup. ... -
17
DOLMA
Data and tools for generating and inspecting OLMo pre-training data
DOLMA (Data Optimization and Learning for Model Alignment) is a framework designed to manage large-scale datasets for training and fine-tuning language models efficiently. -
18
FlexLLMGen
Running large language models on a single GPU
...The system focuses on high-throughput generation workloads where large batches of text must be processed quickly, such as large-scale data extraction or document analysis tasks. Instead of requiring expensive multi-GPU systems, the framework uses techniques such as memory offloading, compression, and optimized batching to run large models on commodity hardware. The architecture distributes computation and memory usage across the GPU, CPU, and disk in order to maximize the number of tokens processed during inference. This design allows organizations to deploy powerful language models for high-volume tasks without the infrastructure costs typically associated with large-scale AI systems. ... -
19
WFGY 3.0
A tension reasoning engine over 131 S-class problems
WFGY is an experimental open-source reasoning framework designed to improve the reliability and interpretability of large language model outputs through structured reasoning layers. The project introduces a conceptual reasoning engine that analyzes complex problems by identifying semantic compression errors and residual assumptions within a system’s reasoning process. Its architecture treats reasoning failures as measurable signals that can be detected and analyzed rather than simply observed as incorrect answers. Different versions of the framework, including WFGY 1.0, 2.0, and 3.0, represent stages of development where early conceptual ideas evolved into more structured reasoning engines and diagnostic tools. ... -
20
TurboQuant PyTorch
From-scratch PyTorch implementation of Google's TurboQuant
TurboQuant PyTorch is a specialized deep learning optimization framework designed to accelerate neural network inference and training through advanced quantization techniques within the PyTorch ecosystem. The project focuses on reducing the computational and memory footprint of models by converting floating-point representations into lower-precision formats while preserving performance. It provides tools for experimenting with different quantization strategies, enabling developers to balance... -
21
DeepSeek-OCR
Contexts Optical Compression
DeepSeek-OCR is an open-source optical character recognition solution built as part of the broader DeepSeek AI vision-language ecosystem. It is designed to extract text from images, PDFs, and scanned documents, and integrates with multimodal capabilities that understand layout, context, and visual elements beyond raw character recognition. The system treats OCR not simply as “read the text” but as “understand what the text is doing in the image”—for example distinguishing captions from body... -
22
Headroom
Compress tool outputs, logs, files, and RAG chunks
Headroom is a context optimization layer for LLM applications that compresses information before it reaches the model. It sits between an application and an LLM provider, intercepting requests and forwarding a shorter optimized prompt. The project is designed to reduce token usage while preserving the answer quality needed for agent workflows. It can compress tool outputs, logs, RAG chunks, files, and conversation history. Headroom can be used as a transparent proxy, a Python function, a... -
23
SparseML
Libraries for applying sparsification recipes to neural networks
SparseML is an optimization toolkit for training and deploying deep learning models using sparsification techniques like pruning and quantization to improve efficiency. -
24
PaddleX
PaddlePaddle End-to-End Development Toolkit
PaddleX is a deep learning full-process development tool based on the core framework, development kit, and tool components of Paddle. It has three characteristics opening up the whole process, integrating industrial practice, and being easy to use and integrate. Image classification and labeling is the most basic and simplest labeling task. Users only need to put pictures belonging to the same category in the same folder. When the model is trained, we need to divide the training set, the... -
25
NVIDIA Model Optimizer
A unified library of SOTA model optimization techniques
Model Optimizer is a unified library that provides state-of-the-art techniques for compressing and optimizing deep learning models to improve inference efficiency and deployment performance. It brings together multiple optimization strategies such as quantization, pruning, distillation, and speculative decoding into a single cohesive framework. The library is designed to reduce model size and computational requirements while maintaining accuracy, making it particularly valuable for deploying...
