actypedef/AURA
AURA: Augmented Representation for Unified Accuracy-aware Quantization
AURA helps machine learning engineers and researchers optimize the performance of deep learning models by quantizing both weights and activations to low-bit augmented matrices. It takes a trained model and identifies parts that can be compressed without significant accuracy loss, producing a smaller, more efficient model that runs faster on specialized hardware. This is ideal for those working on deploying large language models or other complex neural networks.
No commits in the last 6 months.
Use this if you need to reduce the memory footprint and increase the inference speed of your deep learning models while preserving accuracy, especially for deployment on hardware that supports low-bit data types.
Not ideal if you are a data scientist primarily focused on model training and experimentation without immediate concerns about deployment efficiency or specialized low-bit hardware.
Stars
8
Forks
—
Language
Cuda
License
—
Category
Last pushed
Sep 24, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/actypedef/AURA"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
bitsandbytes-foundation/bitsandbytes
Accessible large language models via k-bit quantization for PyTorch.
intel/neural-compressor
SOTA low-bit LLM quantization (INT8/FP8/MXFP8/INT4/MXFP4/NVFP4) & sparsity; leading model...
dropbox/hqq
Official implementation of Half-Quadratic Quantization (HQQ)
OpenGVLab/OmniQuant
[ICLR2024 spotlight] OmniQuant is a simple and powerful quantization technique for LLMs.
Hsu1023/DuQuant
[NeurIPS 2024 Oral🔥] DuQuant: Distributing Outliers via Dual Transformation Makes Stronger...