OpenSkillIndex← back to search
claude-code

Awesome-omni-skill mhc

Implements Manifold-Constrained Hyper-Connections (mHC) to solve residual connection issues using Doubly Stochastic Matrices.

install
source · Clone the upstream repo
git clone https://github.com/diegosouzapw/awesome-omni-skill
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/diegosouzapw/awesome-omni-skill "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data-ai/mhc" ~/.claude/skills/diegosouzapw-awesome-omni-skill-mhc && rm -rf "$T"
manifest: skills/data-ai/mhc/SKILL.md
source content

mHC Skill

Manifold-Constrained Hyper-Connections (mHC) uses Doubly Stochastic Matrices to improve Deep Learning stability.

Contents

  • Examples
    • Full JAX implementation of 
      sinkhorn_knopp
      and 
      mhc_layer_forward
      .
  • Deep Theory
    • Motivation, stability proofs, and scalability arguments.

Usage

Use this skill when implementing Deep Transformers (1000+ layers) where standard residual connections fail (Gradient Vanishing, Representation Collapse).

# Quick Ref: Sinkhorn-Knopp (See examples.md for full context)
def sinkhorn_knopp(log_matrix, n_iters=20):
    M = jnp.exp(log_matrix)
    def body(i, m):
        m /= m.sum(axis=1, keepdims=True)
        m /= m.sum(axis=0, keepdims=True)
        return m
    return jax.lax.fori_loop(0, n_iters, body, M)