genpark-convolutional-2d-forward-backward-skill

2D spatial convolution layer forward pass with stride, padding, and max-pooling

Part of the GenPark Autonomous Agent Matrix, developed for production AI agents implementing on-device deep learning primitives, attention blocks, and differentiable computational graphs.

🏗️ Architecture

flowchart TD
    A[Tensors / Layer Inputs] --> B[genpark-convolutional-2d-forward-backward-skill]
    B --> C[Pure Python Standard Library Autograd / NN Engine]
    C --> D[Activated Embeddings / Attention Weights / Gradients]
    B --> E[MCP Protocol Endpoint stdio]
    E --> F[Cursor / Claude Desktop / Windsurf Integration]

🚀 Quickstart

Native Python Execution

python example_usage.py

Standard Library Verification

from client import *

MCP Server (Claude Desktop / Cursor)

{
  "mcpServers": {
    "genpark-convolutional-2d-forward-backward-skill": {
      "command": "python",
      "args": ["-m", "genpark_convolutional_2d_forward_backward_skill.mcp_server"]
    }
  }
}

📄 License

MIT License.