Toolkit for efficient experimentation with Speech Recognition, Text2Speech and NLP

Up to 200x Faster Dot Products & Similarity Metrics — for Python, Rust, C, JS, and Swift, supporting f64, f32, f16 real & complex, i8, and bit vectors using SIMD for both AVX2, AVX-512, NEON, SVE, & SVE2 📐

Half-precision floating point types f16 and bf16 for Rust.

Stage 3 IEEE 754 half-precision floating-point ponyfill

float16 provides IEEE 754 half-precision format (binary16) with correct conversions to/from float32

🎯 Accumulated Gradients for TensorFlow 2

PyTorch half precision gemm lib w/ fused optional bias + optional relu/gelu

Code for testing the native float16 matrix multiplication performance on Tesla P100 and V100 GPU based on cublasHgemm

half float library for C and for z80

CPP20 implementation of a 16-bit floating-point type mimicking most of the IEEE 754 behavior. Single file and header-only.

Arduino library to implement float16 data type

Bfloat16 conversion utilities for Go/Golang

A basic library for using OpenAI:s CLIP model in C#

TFLite applications: Optimized .tflite models (i.e. lightweight and low latency) and code to run directly on your Microcontroller!

Main purpose of this library is to provide functions for conversion to and from half precision (16bit) floating point numbers. It also provides functions for basic arithmetic and comparison of half floats.

FP16 Half precision floating point (IEEE754 2008) adder + multiplier

Half-precision floating-point positive infinity.

The bias of a half-precision floating-point number's exponent.

Square root of half-precision floating-point epsilon.

Difference between one and the smallest value greater than one that can be represented as a half-precision floating-point number.