[NeurIPS 2025 (spotlight)] HBLLM: Wavelet-Enhanced High-Fidelity 1-Bit Quantization for LLMs
-
Updated
Dec 17, 2025 - Python
[NeurIPS 2025 (spotlight)] HBLLM: Wavelet-Enhanced High-Fidelity 1-Bit Quantization for LLMs
Code for paper "Accelerating Federated Learning for IoT in Big Data Analytics with Pruning, Quantization and Selective Updating"
RxLM-Med: A multimodal clinical AI agent featuring System 2 reasoning, cross-lingual hierarchical RAG (BM25 + FAISS + RRF), deterministic medical calculation engine, and Traffic Light Protocol (TLP) safety alignment — built on Qwen-VL with LoRA fine-tuning, SFT/DPO alignment, and INT4 quantization for real-world lab report interpretation.
Geometry-aware LLM quantization, per-direction bit allocation priced by measured KL divergence given injected perturbation.
NeurIPS 2019 MicroNet Challenge
Finding Storage- and Compute-Efficient Convolutional Neural Networks
compares different pretrained object classification with per-layer and per-channel quantization using pytorch
Quantization (post-training quantization) your (custom mobilenet_v2) models .h5 or .pb models using TensorFlow Lite 2.4
Weight quantization
custom K-means Quantization for Deep Learning NNs
High-performance LLM compression engine using SVD matrix decomposition, INT8 quantization, and output caching training acceleration.
LLM Quantization Gallery source data
Training neural nets with quantized weights on arbitrarily specified bit-depth
linear quantization with W8A16 for neural networks with PyTorch
A data visualization project for 3D mesh data by using normalization, quantization, and reconstruction with comprehensive error analysis.
Agentic framework for hardware-aware, layer-wise LLM quantization using sequential optimization and on-device benchmarking.
Development and evaluation of a convolutional neural network for fault classification in photovoltaic modules using thermal imaging, targeting execution on embedded edge devices.
To associate your repository with the quantization-algorithms topic, visit your repo's landing page and select "manage topics."