Xilinx/brevitas

Health v2 · maintenance only

Flags: no_license

Adoption not part of the score

1567 stars · 248 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Brevitas is a PyTorch library for neural network quantization supporting both post-training quantization (PTQ) and quantization-aware training (QAT). It provides quantized implementations of common PyTorch layers with fine-grained control over quantization settings, and is closely tied to Xilinx FPGA deployment flows.

Use cases

• quantize a PyTorch model to low bit-widths
• run quantization-aware training for a CNN
• apply post-training quantization to an ImageNet classifier
• prepare a neural network for FPGA deployment
• quantize weights, activations, and biases of conv and attention layers
• compress a model for hardware acceleration

When to choose

• you need PTQ or QAT in PyTorch with per-tensor quantization control
• you are targeting Xilinx/FPGA or other low-precision hardware inference
• you want quantized layers like QuantConv2d, QuantLSTM, or QuantMultiheadAttention

When to avoid

• you need a one-click quantization tool with no configuration
• your project is outside PyTorch
• you need production support from an official vendor product (Brevitas is a research project)

Facets

library · maturity active

machine-learning deep-learning llm-training gpu-computing deep-learning machine-learning gpu-computing hardware python windows quantization quantization-aware-training post-training-quantization pytorch fpga xilinx neural-networks model-compression low-precision linux macos gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Xilinx/brevitas")