This is a compiled CUDA kernel package, not a model. It provides the Mamba selective-scan operation (selective_scan_fn, Mamba, Mamba2, etc.) for use with the Hugging Face kernels library. WEO-SAS built it to support the mamba-* variants of sen2sr, which need a compiled selective-scan kernel for their Mamba/Swin2SR backbone. This is an internal/experimental build — it has not been validated for production use.

This is the repository card of kernels-community/mamba-ssm that has been pushed on the Hub. It was built to be used with the kernels library. This card was automatically generated.

How to use

# make sure `kernels` is installed: `pip install -U kernels`
from kernels import get_kernel

kernel_module = get_kernel("WEO-SAS/mamba-ssm")
selective_scan_fn = kernel_module.selective_scan_fn

selective_scan_fn(...)

Available functions

selective_scan_fn
mamba_inner_fn
falcon_mamba_inner_fn
selective_state_update
mamba_chunk_scan_combined
mamba_split_conv1d_scan_combined
Mamba
Mamba2
MambaLMHeadModel

Benchmarks

No benchmark available yet.

Downloads last month: -

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support