{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreihjdergrltfnh4hf7ipudhgtosonjfyai5kcq4d4gso5y5lc7ejde",
    "uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3moqi7aiafi72"
  },
  "path": "/t/c-inf-soft-sparsity-engine-maths-and-code-from-1972/176964#post_5",
  "publishedAt": "2026-06-20T17:24:04.000Z",
  "site": "https://discuss.huggingface.co",
  "textContent": "Thanks. The Fortran code (subroutines) for inducing sparsity with infinitely differentiable loss functions (not kludges for L0 or L1) and using a metric and its inverse to endogenously prevent basis collapse (no hyperparameters) would probably qualify as a model. Several AIs indicated that it may be, among other things, representational learning engine. The maths definitely has applications to ML and AI.",
  "title": "C-inf soft sparsity engine maths and code from 1972"
}