2601.13160v1.pdf

{
  "file/base": {
    "record": {
      "extension": ".pdf",
      "hash": "359bd5e60aae936abad48014d286ad2adf1de0b6a8b2c4d9b5a8b0357c097c7a",
      "media_type": "application/pdf",
      "media_type_prefix": "application",
      "name": "2601.13160v1.pdf",
      "size": 1182428
    },
    "source": {
      "execution_id": "d403a9c3-96bb-4e12-9289-f5410edb0edd",
      "id": "file/base",
      "type": "Model",
      "version": "1.0.0"
    }
  },
  "file/pdf": {
    "private": false,
    "record": {
      "creation_date": "2026-01-21T03:04:31Z",
      "keywords": [],
      "modified_date": "2026-01-21T03:04:31Z",
      "page_count": 25,
      "producer": "pikepdf 8.15.1",
      "title": "Training instability in deep learning follows low-dimensional dynamical principles",
      "version": "1.7"
    },
    "source": {
      "execution_id": "c022afea-9581-4e83-84cb-42190b2f47e4",
      "id": "dorsal/pdf",
      "type": "Model",
      "version": "1.1.0"
    }
  }
}