Downloads · 30 days
0
gdhnes/SNN-4-ALL-AE-checkpoints
SNN-4-ALL-AE-checkpoints is a machine learning model from gdhnes. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for custom. The card lists the license as bsd-2-clause.
This repository contains checksum-verified checkpoints prepared for the MICRO 2026 Artifact Evaluation package of:
Downloads · 30 days
0
Access
Public
Updated Jul 21, 2026
Repo size
2.4 GB
Likes
0
Public
Click a slice to open those files.
.pth2.4 GB · 100%
From the Hugging Face model README
This repository contains checksum-verified checkpoints prepared for the MICRO 2026 Artifact Evaluation package of:
SNN-4-All: Breaking the Edge Fine-Tuning Barrier with Heterogeneous Sparsity Exploitation
GDHNES/SNN-4-ALL-AE-releasemicro2026-ae-v1.1The Server SpikingVGG-11 baseline and TwiST checkpoints were produced by full training under the recovered coupled historical protocol. Their independently evaluated accuracies were 58.1% and 59.5%, versus 59.4% and 60.2% reported in the paper; both satisfy the pre-declared +/-3 percentage-point tolerance.
The Orin Spikformer checkpoints support checksum-verified, complete-test and resource replay. This is checkpoint-based replay evidence, not a new full-training reproduction.
Verify every checkpoint against CHECKSUMS.sha256 before use. Exact provenance
and the redistribution license are documented in
CHECKPOINT_LICENSES.md and checkpoint_manifest.json.
The public SpikeGPT-216M checkpoint is not duplicated here. The source artifact downloads the upstream Apache-2.0 checkpoint directly from the immutable revision recorded in its downloader and verifies its SHA-256.
An anonymous, token-free download from an empty cache was completed against a
full immutable Hugging Face revision. All four SHA-256 checks passed. The four
downloaded checkpoints then reproduced the expected complete-test accuracies:
58.1% and 59.5% for Server SpikingVGG-11 baseline/TwiST, and 75.9% and 75.7%
for Orin Spikformer baseline/TwiST. See ANONYMOUS_VALIDATION.md.