Optimal Sparsity Code Models
Optimal Sparsity Code is an experimental checkpoint family released with the authors' sparsity study. These are distinct trained configurations grouped for research comparison.
Research release
Each identifier records an architecture configuration
The family belongs to the Optimal Sparsity paper and its author-maintained training code. It is a controlled research release rather than a single general assistant.Optimal Sparsity Code Models official model card
Read the variant name
In each identifier, d is hidden width, E is the number of experts, and k is the number selected. The last two fields give total and active parameters. For example, d512-E8-k2-320M-A170M and d2048-E128-k2-52.2B-A1.5B represent different model sizes and routing costs.Optimal Sparsity author repository and identifier format
Match the code experiment exactly
Practical guidance. Record the full architecture identifier and evaluation settings when using these coding checkpoints. A shared total parameter count does not identify the same number of active experts.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Research release |
| Access | Downloadable |
| Model identifiers | optimal-sparsity-code-d512-E8-k2-320M-A170Moptimal-sparsity-code-d512-E16-k2-520M-A170Moptimal-sparsity-code-d512-E32-k2-920M-A170Moptimal-sparsity-code-d512-E64-k2-1.7B-A170Moptimal-sparsity-code-d512-E128-k2-3.3B-A170Moptimal-sparsity-code-d512-E256-k2-6.6B-A170Moptimal-sparsity-code-d512-E8-k4-320M-A220Moptimal-sparsity-code-d512-E16-k4-520M-A220Moptimal-sparsity-code-d512-E32-k4-920M-A220Moptimal-sparsity-code-d512-E64-k4-1.7B-A220Moptimal-sparsity-code-d512-E128-k4-3.3B-A220Moptimal-sparsity-code-d512-E256-k4-6.6B-A220Moptimal-sparsity-code-d512-E8-k8-320M-A320Moptimal-sparsity-code-d512-E16-k8-520M-A320Moptimal-sparsity-code-d512-E32-k8-920M-A320Moptimal-sparsity-code-d512-E64-k8-1.7B-A320Moptimal-sparsity-code-d512-E128-k8-3.3B-A320Moptimal-sparsity-code-d512-E256-k8-6.6B-A320Moptimal-sparsity-code-d512-E16-k16-520M-A520Moptimal-sparsity-code-d512-E32-k16-920M-A520Moptimal-sparsity-code-d512-E64-k16-1.7B-A520Moptimal-sparsity-code-d512-E128-k16-3.3B-A520Moptimal-sparsity-code-d512-E256-k16-6.6B-A520Moptimal-sparsity-code-d1024-E8-k2-1.1B-A470Moptimal-sparsity-code-d1024-E16-k2-1.9B-A470Moptimal-sparsity-code-d1024-E32-k2-3.5B-A470Moptimal-sparsity-code-d1024-E64-k2-6.7B-A470Moptimal-sparsity-code-d1024-E128-k2-13.2B-A470Moptimal-sparsity-code-d1024-E256-k2-26.0B-A470Moptimal-sparsity-code-d1024-E8-k4-1.1B-A670Moptimal-sparsity-code-d1024-E16-k4-1.9B-A670Moptimal-sparsity-code-d1024-E32-k4-3.5B-A670Moptimal-sparsity-code-d1024-E64-k4-6.7B-A670Moptimal-sparsity-code-d1024-E128-k4-13.2B-A670Moptimal-sparsity-code-d1024-E256-k4-26.0B-A670Moptimal-sparsity-code-d1024-E8-k8-1.1B-A1.1Boptimal-sparsity-code-d1024-E16-k8-1.9B-A1.1Boptimal-sparsity-code-d1024-E32-k8-3.5B-A1.1Boptimal-sparsity-code-d1024-E64-k8-6.7B-A1.1Boptimal-sparsity-code-d1024-E128-k8-13.2B-A1.1Boptimal-sparsity-code-d1024-E256-k8-26.0B-A1.1Boptimal-sparsity-code-d1024-E16-k16-1.9B-A1.9Boptimal-sparsity-code-d1024-E32-k16-3.5B-A1.9Boptimal-sparsity-code-d1024-E64-k16-6.7B-A1.9Boptimal-sparsity-code-d1024-E128-k16-13.2B-A1.9Boptimal-sparsity-code-d1024-E256-k16-26.0B-A1.9Boptimal-sparsity-code-d2048-E8-k2-3.9B-A1.5Boptimal-sparsity-code-d2048-E16-k2-7.1B-A1.5Boptimal-sparsity-code-d2048-E32-k2-13.6B-A1.5Boptimal-sparsity-code-d2048-E64-k2-26.4B-A1.5Boptimal-sparsity-code-d2048-E128-k2-52.2B-A1.5Boptimal-sparsity-code-d2048-E8-k4-3.9B-A2.3Boptimal-sparsity-code-d2048-E16-k4-7.1B-A2.3Boptimal-sparsity-code-d2048-E32-k4-13.6B-A2.3Boptimal-sparsity-code-d2048-E64-k4-26.4B-A2.3Boptimal-sparsity-code-d2048-E128-k4-52.2B-A2.3Boptimal-sparsity-code-d2048-E8-k8-3.9B-A3.9Boptimal-sparsity-code-d2048-E16-k8-7.1B-A3.9Boptimal-sparsity-code-d2048-E32-k8-13.6B-A3.9Boptimal-sparsity-code-d2048-E64-k8-26.4B-A3.9Boptimal-sparsity-code-d2048-E128-k8-52.2B-A3.9Boptimal-sparsity-code-d2048-E16-k16-7.1B-A7.1Boptimal-sparsity-code-d2048-E32-k16-13.6B-A7.1Boptimal-sparsity-code-d2048-E64-k16-26.4B-A7.1Boptimal-sparsity-code-d2048-E128-k16-52.2B-A7.1B |
| Access | Downloadable model checkpoints; Apache 2.0 model cards.Optimal Sparsity Code Models official model card |
| Coverage | 65 published identifiers are grouped in this article.Optimal Sparsity Code Models official model card |
Related models
- Drop-Upcycling Research Models
LLM-jp / NII LLMC · Drop-Upcycling
Published dense and expert-model experiments for studying partial weight reinitialization during upcycling.
Launch date not confirmedResearch releaseDownloadable - Optimal Sparsity Math Models
LLM-jp / NII LLMC · Optimal Sparsity
The math branch of a controlled expert-model research sweep, with exact variant identifiers kept together.
Launch date not confirmedResearch releaseDownloadable