memory-efficient 开源项目推荐（174 个）

FareedKhan-dev /kimi-k3-in-c

2.78万亿参数Kimi K3在8.24 GB RAM中的单个中央处理器上运行推理。便携式C99：无BLAS、无框架、无图形处理器。

box /spout

Read and write spreadsheet files (CSV, XLSX and ODS), in a fast and scalable way

plasticityai /magnitude

A fast, efficient universal vector embedding utility package.

uhop /stream-json

一个流组件微库，用于以极低内存占用构建自定义 JSON 和 JSONC 处理流水线——通过受 SAX 启发的令牌 API，在 Node.js 或 Web Streams 上解析、过滤和转换远大于可用内存的 JSON。

alphadose /haxmap

Fastest and most memory efficient golang concurrent hashmap

microsoft /tolerant-php-parser

An early-stage PHP parser designed for IDE usage scenarios.

haoliuhl /ringattention

Large Context Attention

alphadose /ZenQ

A thread-safe queue faster and more resource efficient than golang's native channels

nanoexpress /nanoexpress

专业的 Node.js 后端框架

lucidrains /memory-efficient-attention-pytorch

Implementation of a memory efficient multi-head attention as proposed in the paper, "Self-attention Does Not Need O(n²) Memory"

zhuhanqing /APOLLO

APOLLO: SGD-like Memory, AdamW-level Performance; MLSys'25 Oustanding Paper Honorable Mention

Linketic /VGGT-X

VGGT-X: When VGGT Meets Dense Novel View Synthesis

daniel-rusu /pods4k

Performance-oriented Data Structures for Kotlin

DAMO-NLP-SG /Inf-CLIP

[CVPR 2025 Highlight] The official CLIP training codebase of Inf-CL: "Breaking the Memory Barrier: Near Infinite Batch Size Scaling for Contrastive Loss". A super memory-efficiency CLIP training scheme.

alibaba /EasyParallelLibrary

Easy Parallel Library（EPL）是一个通用且高效的深度学习框架，用于分布式模型训练。

dadhi /ImTools

Fast and memory-efficient immutable collections and helper data structures

SubhamTyagi /Last-Launcher

Lightweight: Faster than light, Low on memory

neopenx /Dragon

Dragon: A Computation Graph Virtual Machine Based Deep Learning Framework.

luyug /GC-DPR

Train Dense Passage Retriever (DPR) with a single GPU

abduallahmohamed /Social-Implicit

Code for: "Social-Implicit: Rethinking Trajectory Prediction Evaluation and The Effectiveness of Implicit Maximum Likelihood Estimation" Accepted @ ECCV2022