New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Mhc Algorithm

by benchflow-ai in benchflow-ai/skillsbench

Implement mHC (Manifold-Constrained Hyper-Connections) for stabilizing deep network training.

Apache-2.0Auto-check passedResearch & Science

Install Mhc Algorithm

skills CLI
$ npx skills add benchflow-ai/skillsbench --skill mhc-algorithm -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install benchflow-ai/skillsbench mhc-algorithm --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/benchflow-ai/skillsbench.git skills-src && mkdir -p .claude/skills && cp -r skills-src/tasks-extra/mhc-layer-impl/environment/skills/mhc-algorithm .claude/skills/mhc-algorithm && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
mhc-algorithm
GitHub stars
1.8k
Token cost
~1k tokens
SKILL.md length
166 words
Files
6 (incl. references)
Skills in repo
178
Repo updated
First seen
Licence
Apache-2.0

At a glance

Implement mHC (Manifold-Constrained Hyper-Connections) for stabilizing deep network training.

  • Implementing residual connection improvements with doubly stochastic matrices via Sinkhorn-Knopp algorithm
  • SKILL.md covers Overview, Quick Reference, Installation and Minimal Example, plus 3 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md
  • Tasks that involve Academic paper search

What it does

Mhc Algorithm is an agent skill from benchflow-ai/skillsbench. Implement mHC (Manifold-Constrained Hyper-Connections) for stabilizing deep network training. Use when implementing residual connection improvements with doubly stochastic matrices via Sinkhorn-Knopp algorithm. Based on DeepSeek's 2025 paper (arXiv:2512.24880).

Its SKILL.md is about 1k tokens, which your agent loads only when the skill is triggered. The skill folder holds 6 other files, including reference files (for example `references/core-concepts.md`, `references/gpt-integration.md` and `references/module-implementation.md`).

It sits in Research & Science, covering Academic paper search. It works with DeepSeek and arXiv. The repository describes itself as: SkillsBench evaluates how well skills work and how effective agents are at using them. The licence is Apache-2.0.

When your agent uses it

  • Implementing residual connection improvements with doubly stochastic matrices via Sinkhorn-Knopp algorithm
  • Tasks that involve Academic paper search

Example prompts

  • “/mhc-algorithm”

Requirements

  • Python 3

What it can do on your machine

Read from SKILL.md and the folder at commit 9a1f4dd. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md (its code samples are python).

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    Links to these hosts (documentation or services it may open):

    • arxiv.org
    • en.wikipedia.org

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Mhc Algorithm loads about 1k tokens when it runs, and up to ~4.5k if it reads all its reference files. Until then it costs about 69 tokens; SKILL.md has 166 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~69
When it runs · the whole SKILL.md, loaded when a task matches
~1k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~4.5k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from benchflow-ai/skillsbench at commit 9a1f4dd, republished under its Apache-2.0 licence (© benchflow-ai). 166 words, ~1,029 tokens.

Download SKILL.mdSave it as .claude/skills/mhc-algorithm/SKILL.md (or your agent's skills folder). This skill also uses 5 other files; get the full folder from GitHub.
name
mhc-algorithm
description
Implement mHC (Manifold-Constrained Hyper-Connections) for stabilizing deep network training. Use when implementing residual connection improvements with doubly stochastic matrices via Sinkhorn-Knopp algorithm. Based on DeepSeek's 2025 paper (arXiv:2512.24880).

mHC: Manifold-Constrained Hyper-Connections

Overview

mHC (Manifold-Constrained Hyper-Connections) stabilizes deep network training by constraining residual mixing matrices to be doubly stochastic. It provides:

  • Stable Training: Lower gradient norm variance via doubly stochastic constraints
  • Multiple Streams: Hyper-Connections with learnable mixing across residual streams
  • Sinkhorn Projection: Log-space Sinkhorn-Knopp algorithm for doubly stochastic projection
  • GPT Integration: Pattern for wrapping attention and MLP layers

Two components:

  • HyperConnections Module: Core PyTorch module with H_res, H_pre, H_post matrices
  • Sinkhorn-Knopp: Log-space projection to doubly stochastic manifold

Quick Reference

TopicReference
Core Concepts & MathCore Concepts
Sinkhorn AlgorithmSinkhorn-Knopp
HyperConnections ModuleModule Implementation
GPT IntegrationGPT Integration
Common PitfallsPitfalls

Installation

python
# Required packages
pip install torch einops numpy

Minimal Example

python
import torch
import torch.nn as nn
from einops import rearrange, einsum

def sinkhorn_knopp(logits, num_iters=20, tau=0.05):
    log_alpha = logits / tau
    for _ in range(num_iters):
        log_alpha = log_alpha - torch.logsumexp(log_alpha, dim=-1, keepdim=True)
        log_alpha = log_alpha - torch.logsumexp(log_alpha, dim=-2, keepdim=True)
    return torch.exp(log_alpha)

class HyperConnections(nn.Module):
    def __init__(self, num_streams, dim, branch=None, layer_idx=0):
        super().__init__()
        self.num_streams = num_streams
        self.branch = branch

        # Initialize H_res near identity (use small negative for gradient flow)
        init_h_res = torch.full((num_streams, num_streams), -0.1)
        init_h_res.fill_diagonal_(0.0)
        self.H_res_logits = nn.Parameter(init_h_res)

        # H_pre/H_post for depth connections
        init_h_pre = torch.full((1, num_streams), -0.1)
        init_h_pre[0, layer_idx % num_streams] = 0.0
        self.H_pre_logits = nn.Parameter(init_h_pre)
        self.H_post_logits = nn.Parameter(torch.zeros(1, num_streams))

    def forward(self, x):
        s = self.num_streams
        x = rearrange(x, "(b s) t d -> b t s d", s=s)

        h_res = sinkhorn_knopp(self.H_res_logits)
        x_mixed = einsum(h_res, x, "s t, b n s d -> b n t d")

        h_pre = self.H_pre_logits.softmax(dim=-1)
        branch_in = einsum(h_pre, x, "v s, b n s d -> b n v d").squeeze(-2)

        branch_out = self.branch(branch_in) if self.branch else branch_in

        h_post = self.H_post_logits.softmax(dim=-1)
        depth_out = einsum(branch_out, h_post, "b t d, v s -> b t s d")

        output = x_mixed + depth_out
        return rearrange(output, "b t s d -> (b s) t d")

Common Imports

python
import torch
import torch.nn as nn
import torch.nn.functional as F
from einops import rearrange, einsum, repeat, reduce

When to Use What

ScenarioApproach
Standard residual connectionNo mHC needed
Deep networks (>12 layers) with stability issuesUse mHC with num_streams=4
GPT/Transformer trainingWrap both attention and MLP with HyperConnections
Custom Sinkhorn iterationsAdjust num_iters (20 default) and tau (0.05 default)
Memory-constrained trainingReduce num_streams or batch size

External Resources

© benchflow-ai, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 5 other files (references) in tasks-extra/mhc-layer-impl/environment/skills/mhc-algorithm of benchflow-ai/skillsbench.

  • SKILL.md
  • references/core-concepts.md
  • references/gpt-integration.md
  • references/module-implementation.md
  • references/pitfalls.md
  • references/sinkhorn-knopp.md

Open the folder on GitHubat commit 9a1f4dd

Compare with similar skills

Mhc Algorithm next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Mhc Algorithm compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Mhc Algorithm this skillbenchflow-ai/skillsbench1.8k—~1kAutomated safety check: PassApache-2.0
Nature Academic Searchwp-a/nature-academic-search299—~1.4kAutomated safety check: PassMIT
Summaryalaliqing/claude-paper3441 repos~2kAutomated safety check: NotesMIT
Read arXiv Paperkarpathy/nanochat58k2 repos~494Automated safety check: PassMIT
Literature Reviewneflibata-feng/MyArxiv-Agent12621 repos~5.9kAutomated safety check: NotesMIT
Citation ManagementK-Dense-AI/claude-scientific-writer2.4k3 repos~3.9kAutomated safety check: NotesMIT

Similar skills

  • Nature Academic Search

    wp-a/nature-academic-search

    A skill your agent uses when users ask to 找文献、做文献检索、查论文、查临床试验、核验引用、去重文献、设计 PubMed/MeSH 检索式、追踪上下游引文、解析 DOI/PMID/PMCID/arXiv/OpenAlex/Semantic Scholar/NCT ID， 或导出 RIS、BibTeX、NBIB、ENW；also use for…

    299 GitHub stars~1.4k tokensUpdated 6 days ago
    Research & ScienceAuto-check passed

  • Summary

    alaliqing/claude-paper

    Use this for a quick summary of a research paper's core ideas and key points.

    344 GitHub starsUsed in 1 repo~2k tokens
    Research & ScienceAuto-check: notes

  • Read arXiv Paper

    karpathy/nanochat

    Fetches the TeX source of an arXiv paper from its URL, reads it and writes a markdown summary tied to the nanochat project.

    58k GitHub starsUsed in 2 repos~494 tokens
    Research & ScienceAuto-check passed

  • Literature Review

    neflibata-feng/MyArxiv-Agent

    Conduct comprehensive, systematic literature reviews using multiple academic databases (PubMed, arXiv, bioRxiv, Semantic Scholar, etc.).

    126 GitHub starsUsed in 21 repos~5.9k tokens
    Research & ScienceAuto-check: notes

  • Citation Management

    K-Dense-AI/claude-scientific-writer

    Finds papers in OpenAlex, PubMed and Google Scholar, turns DOIs, PMIDs and arXiv IDs into clean BibTeX, and validates citations for a manuscript or thesis.

    2.4k GitHub starsUsed in 3 repos~3.9k tokens
    Research & ScienceAuto-check: notes

  • Openalex Database

    neflibata-feng/MyArxiv-Agent

    Query and analyze scholarly literature using the OpenAlex database.

    126 GitHub starsUsed in 13 repos~3k tokens
    Research & ScienceAuto-check passed

More from benchflow-ai/skillsbench

All 178 skills in this repo

  • Lean4 Memories

    benchflow-ai/skillsbench

    This skill should be used when working on Lean 4 formalization projects to maintain persistent memory of successful proof patterns, failed approaches, project conventions, and user preferences…

    1.8k GitHub stars~3.2k tokensUpdated 2 mo ago
    Auto-check passed

  • Senior Data Engineer

    benchflow-ai/skillsbench

    World-class data engineering skill for building scalable data pipelines, ETL/ELT systems, real-time streaming, and data infrastructure.

    1.8k GitHub stars~5.9k tokensUpdated 2 mo ago
    Auto-check passed

  • Ac Branch Pi Model

    benchflow-ai/skillsbench

    AC branch pi-model power flow equations (P/Q and |S|) with transformer tap ratio and phase shift, matching acopf-math-model.md and MATPOWER branch fields.

    1.8k GitHub stars~1.1k tokensUpdated 2 mo ago
    Auto-check passed

  • Civ6lib

    benchflow-ai/skillsbench

    Civilization 6 district mechanics library. An agent skill from benchflow-ai/skillsbench.

    1.8k GitHub stars~1.7k tokensUpdated 2 mo ago
    Auto-check passed

  • D3 Visualization

    benchflow-ai/skillsbench

    Build deterministic, verifiable data visualizations with D3.js (v6).

    1.8k GitHub stars~1.5k tokensUpdated 2 mo ago
    Auto-check passed

  • Dc Power Flow

    benchflow-ai/skillsbench

    DC power flow analysis for power systems. An agent skill from benchflow-ai/skillsbench.

    1.8k GitHub stars~717 tokensUpdated 2 mo ago
    Auto-check passed

Works with

Categories

Questions about Mhc Algorithm

What does Mhc Algorithm do?

Implement mHC (Manifold-Constrained Hyper-Connections) for stabilizing deep network training. Mhc Algorithm is an agent skill from benchflow-ai/skillsbench. Implement mHC (Manifold-Constrained Hyper-Connections) for stabilizing deep network training.

When should I use Mhc Algorithm?

Mhc Algorithm fits situations like: implementing residual connection improvements with doubly stochastic matrices via Sinkhorn-Knopp algorithm; tasks that involve Academic paper search.

How do I install Mhc Algorithm in Claude Code?

Run `npx skills add benchflow-ai/skillsbench --skill mhc-algorithm -a claude-code`. Or copy the skill folder (tasks-extra/mhc-layer-impl/environment/skills/mhc-algorithm in benchflow-ai/skillsbench) into .claude/skills/mhc-algorithm in your project. Claude Code loads it when a task matches its description.

How do I install Mhc Algorithm in Codex?

Run `npx skills add benchflow-ai/skillsbench --skill mhc-algorithm -a codex`. Or copy the skill folder (tasks-extra/mhc-layer-impl/environment/skills/mhc-algorithm in benchflow-ai/skillsbench) into .agents/skills/mhc-algorithm in your project. Codex loads it when a task matches its description.

Can I use Mhc Algorithm in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add benchflow-ai/skillsbench --skill mhc-algorithm -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/mhc-algorithm, .gemini/skills/mhc-algorithm, .github/skills/mhc-algorithm and .opencode/skills/mhc-algorithm in your project.

What does Mhc Algorithm need to run?

SKILL.md names no scripts, command-line tools or credentials: Mhc Algorithm is instructions for the agent only. Our summary lists: Python 3.

Does Mhc Algorithm access the network?

SKILL.md names 2 domains. As links in the text: arxiv.org and en.wikipedia.org. This is read from the text; nothing was executed.

Is Mhc Algorithm safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Mhc Algorithm use?

Mhc Algorithm is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Mhc Algorithm use?

About 1k tokens (SKILL.md is roughly 4.1k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 3.5k tokens, read only when the agent opens those files.

What are the alternatives to Mhc Algorithm?

Skills that share tags, products or a category with Mhc Algorithm: Nature Academic Search (wp-a/nature-academic-search, 299 stars), Summary (alaliqing/claude-paper, 344 stars), Read arXiv Paper (karpathy/nanochat, 58k stars) and Literature Review (neflibata-feng/MyArxiv-Agent, 126 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Mhc Algorithm?

benchflow-ai (a GitHub organization) maintains it in benchflow-ai/skillsbench, which has 1,832 GitHub stars. The repository holds 178 skills in this directory. The repository was last updated on July 23, 2026.

Source: benchflow-ai/skillsbench on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.