{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "sauer-et-al-stylegant-text-image-adversarial-2023",
    "title": "StyleGAN-T: making text-conditioned GANs larger and easier to steer",
    "summary": "A GAN can turn a random seed into an image quickly, but open-ended text adds a much harder conditioning problem. StyleGAN-T revises the synthesis network and discriminator, incorporates pretrained visual and text features, and alternates generator training with adaptation of the text encoder. Per-prompt truncation moves a latent sample toward a typical sample for that prompt, trading variety for alignment. Sauer and colleagues demonstrate the combined design at large scale. The contribution is the specific adversarial image system, rather than invention of its inherited GAN or pretrained representation components.",
    "boundary": "The reported low-resolution result does not establish high-resolution superiority: at 256 pixels its distributional image score trails several diffusion and autoregressive comparators. Training uses 250 million pairs and four weeks on 64 A100 GPUs. Sampling times refer to historical hardware and settings; comparator hardware and measurement sources differ. Strong text-alignment guidance can introduce artifacts, while truncation reduces diversity. Attribute binding and readable text remain failure cases. Smooth selected latent edits are not a guarantee of arbitrary semantic control.",
    "area": "Adversarial image generation",
    "topic": "Artificial intelligence",
    "year": 2023,
    "lane": "HUMAN",
    "result": "ALGORITHM",
    "package_id": "sauer-et-al-stylegant-text-image-adversarial-2023",
    "target": {
      "record_id": "urn:uuid:832ec2d4-4e2d-4efe-a5ce-fc025b6f22d8",
      "version": 1,
      "content_hash": "sha256:bf0bf35cf36aab858da9e99665ee60f907d711c522f2dca6cc04e019b14ecd9e"
    },
    "humans": [
      "Axel Sauer",
      "Tero Karras",
      "Samuli Laine",
      "Andreas Geiger",
      "Timo Aila"
    ],
    "authors_complete": true,
    "human_role_note": "Source-listed human researchers and engineers for the exact method or resource. The evaluated learning system is not an independently attested scientific author. Earlier methods, funding and data-collection roles remain separate. Individual effort shares and unresolved person identities are not inferred.",
    "agents": [],
    "related": [],
    "subresults": [],
    "scientific_assessment": {
      "source_review": "SOURCE_CURATED",
      "scientific_review": "NOT_REVIEWED_BY_WRS",
      "formal_replay": "NOT_RUN",
      "provider_attestation": "NONE"
    },
    "sources": [
      {
        "id": "cur20261017-arxiv-2301-09515v1",
        "title": "StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis",
        "url": "https://arxiv.org/abs/2301.09515v1",
        "date": "2023-01-23"
      }
    ],
    "points": "70000",
    "published_grade": {
      "id": "sauer-et-al-stylegant-text-image-adversarial-2023",
      "entry_id": "sauer-et-al-stylegant-text-image-adversarial-2023",
      "band": "A",
      "modifier": "-",
      "target": {
        "record_id": "urn:uuid:832ec2d4-4e2d-4efe-a5ce-fc025b6f22d8",
        "version": 1,
        "content_hash": "sha256:bf0bf35cf36aab858da9e99665ee60f907d711c522f2dca6cc04e019b14ecd9e"
      },
      "what_changed": "A GAN can turn a random seed into an image quickly, but open-ended text adds a much harder conditioning problem. StyleGAN-T revises the synthesis network and discriminator, incorporates pretrained visual and text features, and alternates generator training with adaptation of the text encoder. Per-prompt truncation moves a latent sample toward a typical sample for that prompt, trading variety for alignment. Sauer and colleagues demonstrate the combined design at large scale. The contribution is the specific adversarial image system, rather than invention of its inherited GAN or pretrained representation components.",
      "scope": "A GAN can turn a random seed into an image quickly, but open-ended text adds a much harder conditioning problem. StyleGAN-T revises the synthesis network and discriminator, incorporates pretrained visual and text features, and alternates generator training with adaptation of the text encoder. Per-prompt truncation moves a latent sample toward a typical sample for that prompt, trading variety for alignment. Sauer and colleagues demonstrate the combined design at large scale. The contribution is the specific adversarial image system, rather than invention of its inherited GAN or pretrained representation components. The reported low-resolution result does not establish high-resolution superiority: at 256 pixels its distributional image score trails several diffusion and autoregressive comparators. Training uses 250 million pairs and four weeks on 64 A100 GPUs. Sampling times refer to historical hardware and settings; comparator hardware and measurement sources differ. Strong text-alignment guidance can introduce artifacts, while truncation reduces diversity. Attribute binding and readable text remain failure cases. Smooth selected latent edits are not a guarantee of arbitrary semantic control.",
      "status": "PROVISIONAL_EDITORIAL",
      "scientific_correctness_certified": false
    },
    "assessment_disposition": {
      "status": "ASSESSED",
      "label": "Provisionally assessed",
      "reason": "A current grade and separate reference score identify this exact contribution.",
      "target": {
        "record_id": "urn:uuid:832ec2d4-4e2d-4efe-a5ce-fc025b6f22d8",
        "version": 1,
        "content_hash": "sha256:bf0bf35cf36aab858da9e99665ee60f907d711c522f2dca6cc04e019b14ecd9e"
      },
      "edition": "ratings-e039106f30dec177"
    }
  },
  "version": {
    "kind": "PUBLIC_RECORD_VIEW",
    "previous": null,
    "published_at": "2026-10-23T00:00:00Z",
    "publisher": {
      "principal_id": "urn:uuid:80116931-785b-40a2-a24b-9d59db0c7e33",
      "role": "WRS_EDITORIAL_PUBLICATION"
    },
    "record": {
      "contributors": [
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Axel Sauer",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis",
            "url": "https://arxiv.org/abs/2301.09515v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Tero Karras",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis",
            "url": "https://arxiv.org/abs/2301.09515v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Samuli Laine",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis",
            "url": "https://arxiv.org/abs/2301.09515v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Andreas Geiger",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis",
            "url": "https://arxiv.org/abs/2301.09515v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Timo Aila",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis",
            "url": "https://arxiv.org/abs/2301.09515v1"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis",
            "url": "https://arxiv.org/abs/2301.09515v1"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "The reported low-resolution result does not establish high-resolution superiority: at 256 pixels its distributional image score trails several diffusion and autoregressive comparators. Training uses 250 million pairs and four weeks on 64 A100 GPUs. Sampling times refer to historical hardware and settings; comparator hardware and measurement sources differ. Strong text-alignment guidance can introduce artifacts, while truncation reduces diversity. Attribute binding and readable text remain failure cases. Smooth selected latent edits are not a guarantee of arbitrary semantic control.",
        "domain": "Adversarial image generation"
      },
      "sources": [
        {
          "citation": "StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis",
          "url": "https://arxiv.org/abs/2301.09515v1"
        }
      ],
      "statement": "# StyleGAN-T: making text-conditioned GANs larger and easier to steer\n\nA GAN can turn a random seed into an image quickly, but open-ended text adds a much harder conditioning problem. StyleGAN-T revises the synthesis network and discriminator, incorporates pretrained visual and text features, and alternates generator training with adaptation of the text encoder. Per-prompt truncation moves a latent sample toward a typical sample for that prompt, trading variety for alignment. Sauer and colleagues demonstrate the combined design at large scale. The contribution is the specific adversarial image system, rather than invention of its inherited GAN or pretrained representation components.\n\n## Scientific scope\n\nThe reported low-resolution result does not establish high-resolution superiority: at 256 pixels its distributional image score trails several diffusion and autoregressive comparators. Training uses 250 million pairs and four weeks on 64 A100 GPUs. Sampling times refer to historical hardware and settings; comparator hardware and measurement sources differ. Strong text-alignment guidance can introduce artifacts, while truncation reduces diversity. Attribute binding and readable text remain failure cases. Smooth selected latent edits are not a guarantee of arbitrary semantic control.\n\n## Reading status\n\nThis is a source-linked WRS catalogue note. A full explanatory guide and independent scientific reconstruction have not been completed for this entry. Read the identified original work for its complete narrative and arguments.\n\n## People and systems\n\nAxel Sauer; Tero Karras; Samuli Laine; Andreas Geiger; Timo Aila\n\nSource-listed human researchers and engineers for the exact method or resource. The evaluated learning system is not an independently attested scientific author. Earlier methods, funding and data-collection roles remain separate. Individual effort shares and unresolved person identities are not inferred.\n\n\n\n## Original sources\n\n- [StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis](https://arxiv.org/abs/2301.09515v1)\n",
      "summary": "A GAN can turn a random seed into an image quickly, but open-ended text adds a much harder conditioning problem. StyleGAN-T revises the synthesis network and discriminator, incorporates pretrained visual and text features, and alternates generator training with adaptation of the text encoder. Per-prompt truncation moves a latent sample toward a typical sample for that prompt, trading variety for alignment. Sauer and colleagues demonstrate the combined design at large scale. The contribution is the specific adversarial image system, rather than invention of its inherited GAN or pretrained representation components.",
      "title": "StyleGAN-T: making text-conditioned GANs larger and easier to steer"
    },
    "record_id": "urn:uuid:832ec2d4-4e2d-4efe-a5ce-fc025b6f22d8",
    "version": 1,
    "source_ref": {
      "record_id": "urn:uuid:832ec2d4-4e2d-4efe-a5ce-fc025b6f22d8",
      "version": 1,
      "content_hash": "sha256:bf0bf35cf36aab858da9e99665ee60f907d711c522f2dca6cc04e019b14ecd9e"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:1bad418051a5520ae41e356ffaffa1a082db36f2674509579d95e18ff2c71077"
  },
  "target": {
    "record_id": "urn:uuid:832ec2d4-4e2d-4efe-a5ce-fc025b6f22d8",
    "version": 1,
    "content_hash": "sha256:bf0bf35cf36aab858da9e99665ee60f907d711c522f2dca6cc04e019b14ecd9e"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "sauer-et-al-stylegant-text-image-adversarial-2023",
    "entry_id": "sauer-et-al-stylegant-text-image-adversarial-2023",
    "revision": 1,
    "title": "StyleGAN-T: making text-conditioned GANs larger and easier to steer",
    "context": "Adversarial image generation",
    "track": "Reusable scientific method, experimental result or historical framework",
    "band": "A",
    "modifier": "-",
    "status": "PROVISIONAL_EDITORIAL",
    "assessed_on": "2026-10-23",
    "evidence_cutoff": "2026-10-23",
    "achievement_state": "SUPPORTED_SCOPED_RESULT",
    "what_changed": "A GAN can turn a random seed into an image quickly, but open-ended text adds a much harder conditioning problem. StyleGAN-T revises the synthesis network and discriminator, incorporates pretrained visual and text features, and alternates generator training with adaptation of the text encoder. Per-prompt truncation moves a latent sample toward a typical sample for that prompt, trading variety for alignment. Sauer and colleagues demonstrate the combined design at large scale. The contribution is the specific adversarial image system, rather than invention of its inherited GAN or pretrained representation components.",
    "scope": "A GAN can turn a random seed into an image quickly, but open-ended text adds a much harder conditioning problem. StyleGAN-T revises the synthesis network and discriminator, incorporates pretrained visual and text features, and alternates generator training with adaptation of the text encoder. Per-prompt truncation moves a latent sample toward a typical sample for that prompt, trading variety for alignment. Sauer and colleagues demonstrate the combined design at large scale. The contribution is the specific adversarial image system, rather than invention of its inherited GAN or pretrained representation components. The reported low-resolution result does not establish high-resolution superiority: at 256 pixels its distributional image score trails several diffusion and autoregressive comparators. Training uses 250 million pairs and four weeks on 64 A100 GPUs. Sampling times refer to historical hardware and settings; comparator hardware and measurement sources differ. Strong text-alignment guidance can introduce artifacts, while truncation reduces diversity. Attribute binding and readable text remain failure cases. Smooth selected latent edits are not a guarantee of arbitrary semantic control.",
    "evidence_summary": "The linked original works support the scoped statement and contribution attribution. WRS has not independently replayed experiments or reconstructed a full mathematical proof. The reported low-resolution result does not establish high-resolution superiority: at 256 pixels its distributional image score trails several diffusion and autoregressive comparators. Training uses 250 million pairs and four weeks on 64 A100 GPUs. Sampling times refer to historical hardware and settings; comparator hardware and measurement sources differ. Strong text-alignment guidance can introduce artifacts, while truncation reduces diversity. Attribute binding and readable text remain failure cases. Smooth selected latent edits are not a guarantee of arbitrary semantic control.",
    "references": [
      {
        "title": "StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis",
        "url": "https://arxiv.org/abs/2301.09515v1",
        "locator": "Original scoped contribution, statement/method and evidence limits; source-specific attribution retained."
      }
    ],
    "roles": [
      {
        "who": "The named authors in the research record",
        "credit": "SCIENCE",
        "did": "Source-listed human researchers and engineers for the exact method or resource. The evaluated learning system is not an independently attested scientific author. Earlier methods, funding and data-collection roles remain separate. Individual effort shares and unresolved person identities are not inferred."
      }
    ],
    "target": {
      "record_id": "urn:uuid:832ec2d4-4e2d-4efe-a5ce-fc025b6f22d8",
      "version": 1,
      "content_hash": "sha256:bf0bf35cf36aab858da9e99665ee60f907d711c522f2dca6cc04e019b14ecd9e"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "70000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": null,
  "reading_status": "SOURCE_LINKED_NOTE"
}