{"id":"283371b110a39c6161f0adbe7d77a32d714d500979981fbc38ffd6dc28ea1b92","title":"NextGAN: a proposed approach to simpler, scalable GAN training | arXiv","argument":"In the abstract for NextGAN, Xinhao Lou and coauthors argue that conventional adversarial losses guide generator updates indirectly: they supervise critic values but do not directly specify the gradients used to update the generator. Their method trains a scalar critic to match its input gradient to displacements between paired reference and generated features, measured in frozen representation spaces. For a fixed generator, the authors say this regression is equivalent to an adversarial objective with a gradient-energy penalty; the generator still uses the standard potential-based GAN update.\n\nThe authors report adapting pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores or trajectories. On ImageNet, they report one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; an adversarial form reaches 1.36. For text-to-video generation, one trained Wan2.1-1.3B model supports one-, two- and four-step sampling. The authors report the highest VBench total scores among compared 1.3B methods at one and two steps, and a score of 83.13 at two steps. These are results reported in the abstract; the supplied evidence does not include the paper’s full evaluation details.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the abstract and publisher page, not the paper’s full evaluation details; it does not give enough information here to assess the reported comparisons or methods in depth.","speaker":{"name":"Lou, Xinhao, Xie, Dongye, Sun, Peng, Zhuang, Chenyi, Yu, Kaicheng, Lin, Tao","roleAtSource":"Article author"},"source":{"title":"NextGAN: Towards Simple and Scalable GAN Training for Image and Video Generation","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2611.06270","publishedAt":"2026-11-04","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors say conventional adversarial losses supervise critic values without directly specifying the generator-update gradients.","excerpt":"conventional adversarial losses supervise critic values without directly specifying the gradients used to update the generator","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"},{"text":"NEXTGAN trains a scalar critic to regress its input gradient onto displacements between independently paired reference and generated features in frozen representation spaces.","excerpt":"trains a scalar critic by regressing its input gradient onto displacements between independently paired reference and generated features at interpolated locations in frozen representation spaces","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"},{"text":"For a fixed generator, the authors say the regression is equivalent to an adversarial objective with a gradient-energy penalty, while the generator retains the standard potential-based GAN update.","excerpt":"For a fixed generator, this regression is equivalent to an adversarial objective with a gradient-energy penalty, while the generator retains the standard potential-based GAN update.","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"},{"text":"The authors report adapting pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores or trajectories.","excerpt":"adapts pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores, or trajectories","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"},{"text":"On ImageNet 256 × 256, the authors report one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; the adversarial form reaches 1.36.","excerpt":"On ImageNet $256 \\times 256$, NEXTGAN achieves one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; its adversarial form further reaches 1.36.","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"},{"text":"For text-to-video generation, the authors report that one trained Wan2.1-1.3B model supports one-, two- and four-step sampling, with the highest VBench total scores among compared 1.3B methods at one and two steps.","excerpt":"a single trained Wan2.1-1.3B model supports one-, two-, and four-step sampling, achieving the highest VBench total scores among the compared 1.3B methods at one and two steps.","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"},{"text":"At two steps, the authors report a VBench score of 83.13, exceeding rCM and AnyFlow by 1.25 and 0.47 points, respectively.","excerpt":"At two steps, NEXTGAN scores 83.13, exceeding rCM and AnyFlow by 1.25 and 0.47 points, respectively.","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.06270","checkedAt":"2026-11-09T12:50:08.142Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How does NEXTGAN perform on image and video settings beyond the ImageNet and Wan2.1-1.3B results reported in the abstract?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"NextGAN: a proposed approach to simpler, scalable GAN training | arXiv","summary":"In the abstract for NextGAN, Xinhao Lou and coauthors argue that conventional adversarial losses guide generator updates indirectly: they supervise critic values but do not directly specify the gradients used to update the generator. Their method trains a scalar critic to match its input gradient to displacements between paired reference and generated features, measured in frozen representation spaces. For a fixed generator, the authors say this regression is equivalent to an adversarial objective with a gradient-energy penalty; the generator still uses the standard potential-based GAN update.\n\nThe authors report adapting pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores or trajectories. On ImageNet, they report one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; an adversarial form reaches 1.36. For text-to-video generation, one trained Wan2.1-1.3B model supports one-, two- and four-step sampling. The authors report the highest VBench total scores among compared 1.3B methods at one and two steps, and a score of 83.13 at two steps. These are results reported in the abstract; the supplied evidence does not include the paper’s full evaluation details.","question":"How does NEXTGAN perform on image and video settings beyond the ImageNet and Wan2.1-1.3B results reported in the abstract?","claims":[{"text":"The authors say conventional adversarial losses supervise critic values without directly specifying the generator-update gradients.","excerpt":"conventional adversarial losses supervise critic values without directly specifying the gradients used to update the generator","support":"source-stated"},{"text":"NEXTGAN trains a scalar critic to regress its input gradient onto displacements between independently paired reference and generated features in frozen representation spaces.","excerpt":"trains a scalar critic by regressing its input gradient onto displacements between independently paired reference and generated features at interpolated locations in frozen representation spaces","support":"source-stated"},{"text":"For a fixed generator, the authors say the regression is equivalent to an adversarial objective with a gradient-energy penalty, while the generator retains the standard potential-based GAN update.","excerpt":"For a fixed generator, this regression is equivalent to an adversarial objective with a gradient-energy penalty, while the generator retains the standard potential-based GAN update.","support":"source-stated"},{"text":"The authors report adapting pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores or trajectories.","excerpt":"adapts pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores, or trajectories","support":"source-stated"},{"text":"On ImageNet 256 × 256, the authors report one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; the adversarial form reaches 1.36.","excerpt":"On ImageNet $256 \\times 256$, NEXTGAN achieves one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; its adversarial form further reaches 1.36.","support":"source-stated"},{"text":"For text-to-video generation, the authors report that one trained Wan2.1-1.3B model supports one-, two- and four-step sampling, with the highest VBench total scores among compared 1.3B methods at one and two steps.","excerpt":"a single trained Wan2.1-1.3B model supports one-, two-, and four-step sampling, achieving the highest VBench total scores among the compared 1.3B methods at one and two steps.","support":"source-stated"},{"text":"At two steps, the authors report a VBench score of 83.13, exceeding rCM and AnyFlow by 1.25 and 0.47 points, respectively.","excerpt":"At two steps, NEXTGAN scores 83.13, exceeding rCM and AnyFlow by 1.25 and 0.47 points, respectively.","support":"source-stated"}],"limitations":["The supplied evidence is the abstract and publisher page, not the paper’s full evaluation details; it does not give enough information here to assess the reported comparisons or methods in depth."],"mediaId":"0d9cab90bcd84fca343d99d1f780b23aee62c18f55052ec0d6836e11cd8a9d48","correctionReason":null,"id":"283371b110a39c6161f0adbe7d77a32d714d500979981fbc38ffd6dc28ea1b92","publicationId":"fcfbfc8d-8e85-4b7a-8a65-0d81a0756263","version":1,"sourceUrl":"https://arxiv.org/abs/2611.06270","sourceTitle":"NextGAN: Towards Simple and Scalable GAN Training for Image and Video Generation","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-11-04","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-05T05:00:28.788Z","createdAt":"2026-11-09T12:35:18.610Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"0d9cab90bcd84fca343d99d1f780b23aee62c18f55052ec0d6836e11cd8a9d48","kind":"article","url":"https://arxiv.org/abs/2611.06270","sourceUrl":"https://arxiv.org/abs/2611.06270","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-09T12:34:13.533Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.06270","checkedAt":"2026-11-09T12:50:08.142Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"0d9cab90bcd84fca343d99d1f780b23aee62c18f55052ec0d6836e11cd8a9d48","confidence":0.96,"provider":"typesafe","model":"jev-1.13.0","inputHash":"ca07ab68c1885399d846ac46e1b3066ece1ad788332adc6eedfa5f0ee226c1e4","decidedAt":"2026-11-09T12:34:16.444Z"},"provenance":{"textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observationId":"1aab9ef4f31f182c383ddf7640b244ec2b05096ab0822beaaf861e01a5c468f2","retrievedAt":"2026-11-09T12:34:13.034Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"9bc1138b-141e-4493-8ec6-6c618d7c43f3"},"reviewer":{"model":"gpt-6-luna","runId":"beed3a2d-cea5-441b-832e-f7888da42bcd"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-09T12:35:18.610Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Lou, Xinhao","url":null,"kind":"unknown"},{"name":"Xie, Dongye","url":null,"kind":"unknown"},{"name":"Sun, Peng","url":null,"kind":"unknown"},{"name":"Zhuang, Chenyi","url":null,"kind":"unknown"},{"name":"Yu, Kaicheng","url":null,"kind":"unknown"},{"name":"Lin, Tao","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Lou, Xinhao","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lou, Xinhao","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.06270","locator":"head/meta[name=author|citation_author]"},"assertionId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T12:50:08.142Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","sourceUrl":"https://arxiv.org/abs/2611.06270","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z"}},{"actor":{"name":"Xie, Dongye","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Xie, Dongye","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.06270","locator":"head/meta[name=author|citation_author]"},"assertionId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T12:50:08.142Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","sourceUrl":"https://arxiv.org/abs/2611.06270","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z"}},{"actor":{"name":"Sun, Peng","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sun, Peng","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.06270","locator":"head/meta[name=author|citation_author]"},"assertionId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T12:50:08.142Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","sourceUrl":"https://arxiv.org/abs/2611.06270","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z"}},{"actor":{"name":"Zhuang, Chenyi","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zhuang, Chenyi","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.06270","locator":"head/meta[name=author|citation_author]"},"assertionId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T12:50:08.142Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","sourceUrl":"https://arxiv.org/abs/2611.06270","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z"}},{"actor":{"name":"Yu, Kaicheng","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yu, Kaicheng","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.06270","locator":"head/meta[name=author|citation_author]"},"assertionId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T12:50:08.142Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","sourceUrl":"https://arxiv.org/abs/2611.06270","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z"}},{"actor":{"name":"Lin, Tao","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lin, Tao","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.06270","locator":"head/meta[name=author|citation_author]"},"assertionId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T12:50:08.142Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","sourceUrl":"https://arxiv.org/abs/2611.06270","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z"}}],"participationCoverage":{"total":6,"returned":6,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.06270","snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z","basis":"publisher-page-metadata","sourceLinks":46,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/283371b110a39c6161f0adbe7d77a32d714d500979981fbc38ffd6dc28ea1b92/source"}},"canonicalUrl":"https://escapemeridian.com/readings/nextgan-a-proposed-approach-to-simpler-scalable-gan-training-283371b1","formats":{"html":"/readings/nextgan-a-proposed-approach-to-simpler-scalable-gan-training-283371b1","markdown":"/readings/nextgan-a-proposed-approach-to-simpler-scalable-gan-training-283371b1.md","json":"/readings/nextgan-a-proposed-approach-to-simpler-scalable-gan-training-283371b1.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[{"id":"term-pre-training","name":"Pre-training","explanation":"A model's initial training on a large dataset, before further adaptation for particular uses.","referenceUrl":"https://developers.google.com/machine-learning/glossary#pre-training"}],"entities":[{"id":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"283371b110a39c6161f0adbe7d77a32d714d500979981fbc38ffd6dc28ea1b92","revision":"fcfbfc8d-8e85-4b7a-8a65-0d81a0756263","revisionKind":"publication-id","title":"NextGAN: a proposed approach to simpler, scalable GAN training | arXiv","url":"https://escapemeridian.com/readings/nextgan-a-proposed-approach-to-simpler-scalable-gan-training-283371b1"},"source":{"type":"source","id":"https://arxiv.org/abs/2611.06270","revision":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","revisionKind":"sha256-source-text","title":"NextGAN: Towards Simple and Scalable GAN Training for Image and Video Generation","url":"https://arxiv.org/abs/2611.06270"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"283371b110a39c6161f0adbe7d77a32d714d500979981fbc38ffd6dc28ea1b92","title":"NextGAN: a proposed approach to simpler, scalable GAN training | arXiv","argument":"In the abstract for NextGAN, Xinhao Lou and coauthors argue that conventional adversarial losses guide generator updates indirectly: they supervise critic values but do not directly specify the gradients used to update the generator. Their method trains a scalar critic to match its input gradient to displacements between paired reference and generated features, measured in frozen representation spaces. For a fixed generator, the authors say this regression is equivalent to an adversarial objective with a gradient-energy penalty; the generator still uses the standard potential-based GAN update.\n\nThe authors report adapting pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores or trajectories. On ImageNet, they report one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; an adversarial form reaches 1.36. For text-to-video generation, one trained Wan2.1-1.3B model supports one-, two- and four-step sampling. The authors report the highest VBench total scores among compared 1.3B methods at one and two steps, and a score of 83.13 at two steps. These are results reported in the abstract; the supplied evidence does not include the paper’s full evaluation details.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the abstract and publisher page, not the paper’s full evaluation details; it does not give enough information here to assess the reported comparisons or methods in depth.","speaker":{"name":"Lou, Xinhao, Xie, Dongye, Sun, Peng, Zhuang, Chenyi, Yu, Kaicheng, Lin, Tao","roleAtSource":"Article author"},"source":{"title":"NextGAN: Towards Simple and Scalable GAN Training for Image and Video Generation","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2611.06270","publishedAt":"2026-11-04","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors say conventional adversarial losses supervise critic values without directly specifying the generator-update gradients.","excerpt":"conventional adversarial losses supervise critic values without directly specifying the gradients used to update the generator","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"},{"text":"NEXTGAN trains a scalar critic to regress its input gradient onto displacements between independently paired reference and generated features in frozen representation spaces.","excerpt":"trains a scalar critic by regressing its input gradient onto displacements between independently paired reference and generated features at interpolated locations in frozen representation spaces","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"},{"text":"For a fixed generator, the authors say the regression is equivalent to an adversarial objective with a gradient-energy penalty, while the generator retains the standard potential-based GAN update.","excerpt":"For a fixed generator, this regression is equivalent to an adversarial objective with a gradient-energy penalty, while the generator retains the standard potential-based GAN update.","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"},{"text":"The authors report adapting pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores or trajectories.","excerpt":"adapts pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores, or trajectories","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"},{"text":"On ImageNet 256 × 256, the authors report one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; the adversarial form reaches 1.36.","excerpt":"On ImageNet $256 \\times 256$, NEXTGAN achieves one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; its adversarial form further reaches 1.36.","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"},{"text":"For text-to-video generation, the authors report that one trained Wan2.1-1.3B model supports one-, two- and four-step sampling, with the highest VBench total scores among compared 1.3B methods at one and two steps.","excerpt":"a single trained Wan2.1-1.3B model supports one-, two-, and four-step sampling, achieving the highest VBench total scores among the compared 1.3B methods at one and two steps.","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"},{"text":"At two steps, the authors report a VBench score of 83.13, exceeding rCM and AnyFlow by 1.25 and 0.47 points, respectively.","excerpt":"At two steps, NEXTGAN scores 83.13, exceeding rCM and AnyFlow by 1.25 and 0.47 points, respectively.","url":"https://arxiv.org/abs/2611.06270","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.06270","checkedAt":"2026-11-09T12:50:08.142Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How does NEXTGAN perform on image and video settings beyond the ImageNet and Wan2.1-1.3B results reported in the abstract?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"NextGAN: a proposed approach to simpler, scalable GAN training | arXiv","summary":"In the abstract for NextGAN, Xinhao Lou and coauthors argue that conventional adversarial losses guide generator updates indirectly: they supervise critic values but do not directly specify the gradients used to update the generator. Their method trains a scalar critic to match its input gradient to displacements between paired reference and generated features, measured in frozen representation spaces. For a fixed generator, the authors say this regression is equivalent to an adversarial objective with a gradient-energy penalty; the generator still uses the standard potential-based GAN update.\n\nThe authors report adapting pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores or trajectories. On ImageNet, they report one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; an adversarial form reaches 1.36. For text-to-video generation, one trained Wan2.1-1.3B model supports one-, two- and four-step sampling. The authors report the highest VBench total scores among compared 1.3B methods at one and two steps, and a score of 83.13 at two steps. These are results reported in the abstract; the supplied evidence does not include the paper’s full evaluation details.","question":"How does NEXTGAN perform on image and video settings beyond the ImageNet and Wan2.1-1.3B results reported in the abstract?","claims":[{"text":"The authors say conventional adversarial losses supervise critic values without directly specifying the generator-update gradients.","excerpt":"conventional adversarial losses supervise critic values without directly specifying the gradients used to update the generator","support":"source-stated"},{"text":"NEXTGAN trains a scalar critic to regress its input gradient onto displacements between independently paired reference and generated features in frozen representation spaces.","excerpt":"trains a scalar critic by regressing its input gradient onto displacements between independently paired reference and generated features at interpolated locations in frozen representation spaces","support":"source-stated"},{"text":"For a fixed generator, the authors say the regression is equivalent to an adversarial objective with a gradient-energy penalty, while the generator retains the standard potential-based GAN update.","excerpt":"For a fixed generator, this regression is equivalent to an adversarial objective with a gradient-energy penalty, while the generator retains the standard potential-based GAN update.","support":"source-stated"},{"text":"The authors report adapting pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores or trajectories.","excerpt":"adapts pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores, or trajectories","support":"source-stated"},{"text":"On ImageNet 256 × 256, the authors report one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; the adversarial form reaches 1.36.","excerpt":"On ImageNet $256 \\times 256$, NEXTGAN achieves one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; its adversarial form further reaches 1.36.","support":"source-stated"},{"text":"For text-to-video generation, the authors report that one trained Wan2.1-1.3B model supports one-, two- and four-step sampling, with the highest VBench total scores among compared 1.3B methods at one and two steps.","excerpt":"a single trained Wan2.1-1.3B model supports one-, two-, and four-step sampling, achieving the highest VBench total scores among the compared 1.3B methods at one and two steps.","support":"source-stated"},{"text":"At two steps, the authors report a VBench score of 83.13, exceeding rCM and AnyFlow by 1.25 and 0.47 points, respectively.","excerpt":"At two steps, NEXTGAN scores 83.13, exceeding rCM and AnyFlow by 1.25 and 0.47 points, respectively.","support":"source-stated"}],"limitations":["The supplied evidence is the abstract and publisher page, not the paper’s full evaluation details; it does not give enough information here to assess the reported comparisons or methods in depth."],"mediaId":"0d9cab90bcd84fca343d99d1f780b23aee62c18f55052ec0d6836e11cd8a9d48","correctionReason":null,"id":"283371b110a39c6161f0adbe7d77a32d714d500979981fbc38ffd6dc28ea1b92","publicationId":"fcfbfc8d-8e85-4b7a-8a65-0d81a0756263","version":1,"sourceUrl":"https://arxiv.org/abs/2611.06270","sourceTitle":"NextGAN: Towards Simple and Scalable GAN Training for Image and Video Generation","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-11-04","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-05T05:00:28.788Z","createdAt":"2026-11-09T12:35:18.610Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"0d9cab90bcd84fca343d99d1f780b23aee62c18f55052ec0d6836e11cd8a9d48","kind":"article","url":"https://arxiv.org/abs/2611.06270","sourceUrl":"https://arxiv.org/abs/2611.06270","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-09T12:34:13.533Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.06270","checkedAt":"2026-11-09T12:50:08.142Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"0d9cab90bcd84fca343d99d1f780b23aee62c18f55052ec0d6836e11cd8a9d48","confidence":0.96,"provider":"typesafe","model":"jev-1.13.0","inputHash":"ca07ab68c1885399d846ac46e1b3066ece1ad788332adc6eedfa5f0ee226c1e4","decidedAt":"2026-11-09T12:34:16.444Z"},"provenance":{"textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observationId":"1aab9ef4f31f182c383ddf7640b244ec2b05096ab0822beaaf861e01a5c468f2","retrievedAt":"2026-11-09T12:34:13.034Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"9bc1138b-141e-4493-8ec6-6c618d7c43f3"},"reviewer":{"model":"gpt-6-luna","runId":"beed3a2d-cea5-441b-832e-f7888da42bcd"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-09T12:35:18.610Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Lou, Xinhao","url":null,"kind":"unknown"},{"name":"Xie, Dongye","url":null,"kind":"unknown"},{"name":"Sun, Peng","url":null,"kind":"unknown"},{"name":"Zhuang, Chenyi","url":null,"kind":"unknown"},{"name":"Yu, Kaicheng","url":null,"kind":"unknown"},{"name":"Lin, Tao","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Lou, Xinhao","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lou, Xinhao","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.06270","locator":"head/meta[name=author|citation_author]"},"assertionId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T12:50:08.142Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","sourceUrl":"https://arxiv.org/abs/2611.06270","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z"}},{"actor":{"name":"Xie, Dongye","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Xie, Dongye","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.06270","locator":"head/meta[name=author|citation_author]"},"assertionId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T12:50:08.142Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","sourceUrl":"https://arxiv.org/abs/2611.06270","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z"}},{"actor":{"name":"Sun, Peng","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sun, Peng","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.06270","locator":"head/meta[name=author|citation_author]"},"assertionId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T12:50:08.142Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","sourceUrl":"https://arxiv.org/abs/2611.06270","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z"}},{"actor":{"name":"Zhuang, Chenyi","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zhuang, Chenyi","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.06270","locator":"head/meta[name=author|citation_author]"},"assertionId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T12:50:08.142Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","sourceUrl":"https://arxiv.org/abs/2611.06270","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z"}},{"actor":{"name":"Yu, Kaicheng","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yu, Kaicheng","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.06270","locator":"head/meta[name=author|citation_author]"},"assertionId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T12:50:08.142Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","sourceUrl":"https://arxiv.org/abs/2611.06270","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z"}},{"actor":{"name":"Lin, Tao","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lin, Tao","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.06270","locator":"head/meta[name=author|citation_author]"},"assertionId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T12:50:08.142Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","sourceUrl":"https://arxiv.org/abs/2611.06270","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z"}}],"participationCoverage":{"total":6,"returned":6,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.06270","snapshotId":"4954f9ebdd785732aa767becce83b447162b08da659e0efbf39e7862e161be49","contentSha256":"f339d53f0bf8de3a0e8c60e9217d01c4b27cc4e332e61616371c3e3aeabcc55e","textHash":"f09924b4fc7611d94f320d80ac654f6dea4fad87c27797a193361e8407960d57","observedAt":"2026-11-09T12:50:08.142Z","basis":"publisher-page-metadata","sourceLinks":46,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/283371b110a39c6161f0adbe7d77a32d714d500979981fbc38ffd6dc28ea1b92/source"}}}],"audits":[{"publicationId":"fcfbfc8d-8e85-4b7a-8a65-0d81a0756263","verdict":"uncertain","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading consistently attributes the problem framing, method and benchmark results to Xinhao Lou and coauthors, presenting them as the authors' claims and reported results rather than established facts.","excerpt":"Authors:Xinhao Lou, Dongye Xie, Peng Sun, Chenyi Zhuang, Kaicheng Yu, Tao Lin"},{"dimension":"summary","status":"uncertain","explanation":"The reading is accurate but not understandable on its own to a non-specialist. Central terms such as GAN, adversarial training, critic, FID, VBench, one-step sampling and frozen representation spaces are never explained. The lead also omits the plain-language point the abstract supports: the method targets one- and few-step image and video generation. Smallest repair: open with that point, gloss GAN, critic, FID and VBench through tooltips, and state that the significance is fewer generation steps, per the authors' results.","excerpt":"Adversarial training enables one- and few-step image and video generation"},{"dimension":"date","status":"pass","explanation":"The originalPublished value of 2026-11-04 matches the arXiv v1 submission date. It is not the later feed date of 2026-11-05.","excerpt":"[v1]\nWed, 4 Nov 2026 10:16:06 UTC (9,244 KB)"},{"dimension":"support","status":"pass","explanation":"Every claim matches the abstract verbatim or closely, including the 14B scale, the FIDs (2.87, 1.54, 1.36), the VBench score of 83.13 and the margins over rCM and AnyFlow. The limitation note and the question do not overreach. The summary drops the phrases 'independently paired' and 'at interpolated locations' from the method description, but this does not distort the method.","excerpt":"At two steps, NEXTGAN scores 83.13, exceeding rCM and AnyFlow by 1.25 and 0.47 points, respectively."},{"dimension":"media","status":"pass","explanation":"The selected media is the verified arXiv abstract page itself. It is the direct original source.","excerpt":null}],"createdAt":"2026-11-09T12:35:18.610Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}