Downloads · 30 days
0
higgs-kernels
galqiwi/higgs-kernels
higgs-kernels is a machine learning model from galqiwi. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
CUDA kernels for HIGGS quantization, packaged for the Hugging Face Kernel Hub.
Access
Public
Updated Feb 14, 2026
Repo size
13.9 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.so27.8 MB · 100%
At a glance
- Access
- Public
- Created
- Feb 13, 2026
- Updated
- Feb 14, 2026
- SHA
- 1c90f831
Model card
From the Hugging Face model README
higgs-kernels
CUDA kernels for HIGGS quantization, packaged for the Hugging Face Kernel Hub.
Extracted from galqiwi/higgs-kernels.
Kernels
higgs_dequantize_2_256- codebook lookup: uint8 indices -> 2D fp16/bf16 vectorshiggs_quantize_2_256_f16- nearest codebook entry search (fp16)higgs_quantize_2_256_bf16- nearest codebook entry search (bf16)
Usage
from kernels import get_kernel
higgs = get_kernel("galqiwi/higgs-kernels")
out = higgs.higgs_dequantize_2_256_kernel(x_uint8, grid_256x2)
indices = higgs.higgs_quantize_2_256_kernel(x_fp16_Nx2, grid_256x2, grid_norms_256)
Grid data
Pre-trained codebook included in grids.safetensors (256x2, key "2_256").
from kernels import get_kernel
higgs = get_kernel("galqiwi/higgs-kernels")
grid = higgs.load_optimal_grid_2_256(device="cuda", dtype=torch.float16)
License
Apache-2.0
- Created
- Feb 13, 2026
- Updated
- Feb 14, 2026