[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-qwen-image-pe-visual-design-prompt-enhancer":3,"topics-all":38,"news-related-089086a5-bcc1-41a9-b5d2-e8ae17529f25":66},{"id":4,"title":5,"summary":6,"content":7,"original_url":8,"source_id":9,"tags":10,"translations":24,"news_slug":31,"published_at":32,"created_at":33,"modified_at":34,"is_published":35,"publish_type":36,"image_url":14,"view_count":37},"089086a5-bcc1-41a9-b5d2-e8ae17529f25","阿里开源文生图提示词改写器：密度不够反而更糟","阿里把文生图的提示词改写单独立成一个可下载模型，Qwen-Image-PE 的 9B 权重已落在 Hugging Face。论文自测里 9B 综合分 60.28 压过 GPT-5.5 的 60.10，397B 到 62.17。更反直觉的是标度实验：信息密度不够时，改写不是无效而是负收益，低于原始请求。","一句\"一只在雨中弹吉他的柯基\"，会被改写成一整段英文长描述，再附一个 16:9 宽高比建议，然后才轮到文生图模型接手。Qwen-Image-PE 把这一步做成了能单独下载的模型——9B 权重已挂在 Hugging Face 的 Qwen-Image-2.1-PE-T2I 上，推理代码同步进了 Qwen 官方仓库（[arXiv:2610.33472](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.33472)）。\n\n## 靶子不是\"更好看\"，是\"会设计\"\n\n传统提示词增强只把短需求扩写成更长的描述。Qwen-Image-PE 换了个靶子——光影、构图、布局、细节四个设计维度。\n\n训练分两段。第一段是 Caption-Based Reverse Distillation：不收集人写的好提示词，而是反过来把\"设计得好的图\"密集描述出来，让改写器从用户需求预测出这种描述。论文用 t-SNE 比了 1000 条请求，改写提示词与真实训练 caption 分布更近：unigram\u002Fbigram 的 JS 散度从 0.229\u002F0.539 降到 0.105\u002F0.329，MAUVE 从 0.032 升到 0.414（对照是 Qwen-3.8-Max）。\n\n第二段是 Groupwise Design Reward。多模态裁判把同一请求的一组候选图拼成网格，四个设计维度一次性联合打分，所有候选共享同一组参照。它只需 D×K 次裁判调用，低于逐点打分的 D×G，远低于穷举两两比较。\n\n## 密度不够，改写反而帮倒忙\n\n三条标度轴：训练数据从 1.25 万加到 10 万，Overall 从 55.87 爬到 57.61，只涨 1.74 分且增量一路递减；模型从 Qwen3.5-2B 换到 397B-A17B，从 49.56 涨到 59.86，涨 10.30 分；信息密度从 L1 到 L4，从 44.65 涨到 57.31，涨 12.66 分——密度最猛。\n\n关键在于 L1 和 L2 低于完全不增强的原始请求（54.36），只有 L3、L4 才反超 1.03 和 2.95 分。密度不够时改写不是无效，而是负收益：把用户本就含糊的需求，改写成另一种含糊。\n\n## 三盆冷水\n\n第一，数字全是论文自测，Qwen-Image-Bench 是 Qwen 自己的基准。397B 综合分 62.17，比原始请求高 7.81 分，同一底座上压过 GPT-5.5（60.10）、Gemini-3.1-Pro（60.04）、FLUX 3 Image Klein PE（59.90），均属论文自报成绩。\n\n第二，拿第一的 397B 是内部模型（表格标了 †），放出来的只有 9B。9B 综合分 60.28 也压过那批对手，但 Layout 一项 60.72 落后 GPT-5.5 的 61.60；397B 的 Layout 是 61.11，对 61.15 基本打平。\n\n第三，增强器补输入端，补不过底座。同一张表里直接出图的 GPT Image 2 综合分 64.69，仍高于 Qwen-Image-2.1 加 397B 增强器后的 62.17（底座不同，仅供参考）。迁移实验也画出边界：FLUX 3 Image Klein 上原生增强器 56.36，9B 与 397B 是 57.81、59.10；混元 HunyuanImage-2.1 上原生 PE-32B 只有 48.59，9B 和 397B 是 51.04、51.66。\n\n## 所以呢\n\n提示词改写正在变成可单独训练、下载、部署的模型品类。9B 权重落地意味着它不再需要闭源 API，也意味着你要自己占一张卡。\n\n值得抄走的是密度门槛：改写器的价值不来自说得更多，而来自说的那部分是不是真信息。密度不足时反而拉低成品质量的改写器，比没改写更糟。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.33472","7437aeb9-930c-4866-a2e9-48003c1a792b",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"499f4b56-819d-49a3-9609-33e775143b86","multimodal",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source",{"id":19,"name":20,"slug":20,"description":14,"color":14},"c187600e-804c-4697-b828-1e4330e0eb10","qwen",{"id":22,"name":23,"slug":23,"description":14,"color":14},"c883fd20-1d66-4fb7-9fc7-320fa7f87023","text-to-image",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"5a21a4fd-3ec9-42b6-a5da-32622d708b1f","en","Alibaba Ships a 9B Prompt Rewriter for Text-to-Image","Alibaba's 9B Qwen-Image-PE rewrites text-to-image prompts, beating GPT-5.5 on the team's own benchmark; sparse prompts make it hurt.","A prompt like \"a corgi playing guitar in the rain\" now gets rewritten into a long English description with a suggested 16:9 aspect ratio before the image model ever sees it. Qwen-Image-PE turns that step into a model you can download on its own: the 9B checkpoint sits on Hugging Face, and the inference code lives in the QwenLM\u002FQwen-Image-2.1 repository ([arXiv:2610.33472](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.33472)).\n\n## The target is not prettier, it is design\n\nClassic prompt enhancement just expands a short request into a longer description. Qwen-Image-PE moves the target to four design dimensions, namely lighting, composition, layout and detail, and trains in two stages.\n\nStage one is Caption-Based Reverse Distillation (CRD). Instead of collecting prompts that humans wrote well, the team runs the process in reverse: densely caption images that were designed well, then train the enhancer to predict such a caption from a user request. On 1,000 requests they compare embedding distributions with t-SNE, and the rewritten prompts sit closer to the real training captions. Unigram and bigram Jensen-Shannon divergence drops from 0.229\u002F0.539 to 0.105\u002F0.329, and MAUVE rises from 0.032 to 0.414, with Qwen-3.8-Max as the comparison point.\n\nStage two is Groupwise Design Reward (GDR). A multimodal judge receives the candidate images for one request arranged in a grid, scores all of them jointly under the four design rubrics, and every candidate is scored against the same group of references. The accounting matters: GDR needs D times K judge calls, fewer than pointwise scoring at D times G and far below exhaustive pairwise judging at D times G choose 2.\n\n## The counterintuitive finding: too little information makes rewriting harmful\n\nThree scaling axes stand out. Growing the training set from 12.5K to 100K examples moves Overall from 55.87 to 57.61, a total of 1.74 points, with increments shrinking from 0.81 to 0.63 to 0.30. Swapping the model from Qwen3.5-2B to 397B-A17B moves it from 49.56 to 59.86, a gain of 10.30 points. Moving information density from L1 to L4 moves it from 44.65 to 57.31, a gain of 12.66 points, the densest of the three axes.\n\nThe detail that matters: L1 and L2 land below the unenhanced original request at 54.36, and only L3 and L4 overtake it, by 1.03 and 2.95 points. Below a density threshold the enhancer is not merely useless, it is negative. It takes a request that was already vague and rewrites it into a different kind of vague.\n\n## Three caveats\n\nFirst, every number comes from the team's own evaluation, and Qwen-Image-Bench is Qwen's own benchmark. The 397B variant reaches 62.17 Overall, 7.81 points above the original request, and on the same Qwen-Image-2.1 backbone it leads GPT-5.5 at 60.10, Gemini-3.1-Pro at 60.04 and FLUX 3 Image Klein PE at 59.90. All of these are self-reported paper numbers.\n\nSecond, the 397B model is an internal one, marked with a dagger in the paper's tables. Only the 9B is public. The 9B also clears the baselines above at 60.28 Overall, but its Layout score of 60.72 falls behind GPT-5.5 at 61.60; the 397B lands at 61.11 against 61.15, essentially a tie.\n\nThird, an enhancer improves the input side and still does not outrun the base model. In the same table, GPT Image 2 scores 64.69 Overall as a direct generator, above the 62.17 that Qwen-Image-2.1 reaches with the 397B enhancer attached. The backbones differ, so treat that as a reference rather than a head-to-head. The transfer experiments draw the same boundary. On FLUX 3 Image Klein the native enhancer gets 56.36 while the 9B and 397B reach 57.81 and 59.10. On HunyuanImage-2.1 the native PE-32B manages only 48.59, and Qwen's two variants score 51.04 and 51.66.\n\n## So what\n\nPrompt rewriting is turning into a model category of its own, something you can train, download and deploy separately. Shipping the 9B weights means it no longer requires a closed API, and it also means you own the GPU slot and the inference bill.\n\nThe part worth stealing is the density threshold, not the scores. An enhancer earns its keep only if what it adds is real information. A rewriter that drags the output down when density is low is worse than no rewriter at all.","qwen-image-pe-visual-design-prompt-enhancer","2026-10-28T01:00:00Z","2026-10-28T01:14:43.025888Z","2026-10-28T01:14:43.025900Z",true,"agent",439,[39,48,57],{"slug":40,"tag_slug":23,"title_zh":41,"title_en":42,"intro_zh":43,"intro_en":44,"id":45,"is_active":35,"created_at":46,"modified_at":47},"image-generation","图像生成：从文生图到逐框编辑的模型竞赛","Image Generation: from text-to-image to frame-level editing","文生图已是红海，战场正转向编辑与可控性：FLUX.2\u002F3 用开源和量化把生图搬进消费级显卡，Qwen-Image、Seedream 等国产模型紧咬不放，GPT Image 2 则把推理能力带进像素。本专题追踪图像生成与编辑模型的发布、架构演进、加速量化与商业化数据，看「好图」如何变成「可用的图」。","Text-to-image is a crowded race, and the frontier has moved to editing and control: FLUX.2\u002F3 push open weights and quantization onto consumer GPUs, Qwen-Image and Seedream close in from China, and GPT Image 2 brings reasoning into pixels. This topic tracks model releases, architecture shifts, acceleration work, and the business numbers behind \"pretty\" becoming \"usable\".","52de7c59-7864-4569-8ae7-3d656d26a1a1","2026-11-20T07:20:23.534506Z","2026-11-20T07:20:23.534527Z",{"slug":49,"tag_slug":49,"title_zh":50,"title_en":51,"intro_zh":52,"intro_en":53,"id":54,"is_active":35,"created_at":55,"modified_at":56},"ai-for-science","AI for Science 2026：从 UniPert 到 GPT-Rosalind 的硬核进化","AI for Science 2026: from UniPert to GPT-Rosalind","生命科学、化学材料、物理世界模型——AI 正在从\"语言工具\"变成\"实验伙伴\"。本专题收录 AI 在三大科学方向的关键节点：UniPert 统一基因与化学扰动空间、GPT-Rosalind 端到端生命科学推理、达摩院 AI 智能体 28 小时找到 4 种超导新材料、Anthropic Claude Science 把工作台做成标准品。","From language tool to lab partner — AI is reshaping life sciences, chemistry\u002Fmaterials, and physical world models. This topic covers the key milestones: UniPert unifying genetic-chemical perturbation spaces, GPT-Rosalind's end-to-end life-sciences reasoning, DAMO's AI agent discovering 4 superconducting materials in 28 hours, and Anthropic's Claude Science workbench going mainstream.","988a4300-5fab-41c4-b5d8-63711a2dc757","2026-09-10T01:34:15.296649Z","2026-09-10T01:34:15.296663Z",{"slug":58,"tag_slug":58,"title_zh":59,"title_en":60,"intro_zh":61,"intro_en":62,"id":63,"is_active":35,"created_at":64,"modified_at":65},"h3-series","MiniMax H3 系列：从开源权重到 35 倍吞吐","MiniMax H3 Series: from open weights to 35x throughput","MiniMax H3 自 2026 年 8 月开源以来节奏密集：官方把生成、参考与编辑收回一个模型；ComfyUI 当天压进 RTX 3060；摩尔线程 3 小时完成国产 GPU 适配；fal 后训练版把吞吐拉到 35 倍；FastH3 蒸馏再砍推理成本。本专题持续追踪 H3 的发布—开源—蒸馏—部署全链路。","Since MiniMax open-sourced H3 in August 2026 the pace has been relentless: one unified omni-modal model, same-day ComfyUI support down to an RTX 3060, a 3-hour Day-0 port to Moore Threads GPUs, fal's post-trained H3 Max at 35x throughput, and FastH3 distillation cutting inference cost further. This topic tracks the full H3 chain — release, open weights, distillation, deployment.","83ef0daa-3c31-4cb3-86ed-e5ee58654d5f","2026-09-08T07:33:19.942193Z","2026-09-08T07:33:19.942209Z",{"items":67},[68,73,78,83,88,93],{"id":69,"title":70,"news_slug":71,"published_at":72},"2f9ad534-b737-490e-b97e-bf01f2b6d26b","交错生成新基准:闭源旗舰文本赢,图像一致性输给开源","intreal-interleaved-generation-benchmark","2026-11-01T05:27:08+00:00",{"id":74,"title":75,"news_slug":76,"published_at":77},"d83c4910-6dce-4171-8a44-9f68c9f841af","Qwen-Audio-3.1-ASR开源:转写自带说话人","qwen-audio-3-1-asr-speaker-diarization","2026-10-22T17:13:29+00:00",{"id":79,"title":80,"news_slug":81,"published_at":82},"bbbda7ff-b1c7-4142-b37f-2b3fd0547e02","Fibo 2.0 开源:生成编辑合一,登顶开源编辑榜","bria-fibo-2-0-unified-editing-generation","2026-10-21T21:08:11+00:00",{"id":84,"title":85,"news_slug":86,"published_at":87},"3bf416bb-a989-4b02-8341-dbafc7c96c48","MOSS-Image 开源:5B 双语生图,文字排版可编程","moss-image-bilingual-text-rendering-base","2026-10-18T19:08:44+00:00",{"id":89,"title":90,"news_slug":91,"published_at":92},"405b8f47-8862-4684-ab0e-d552ce093072","MJev-Spatial:空间理解不生成文字,单次前向快5.6倍","mjev-spatial-classification-spatial-intelligence","2026-10-18T17:14:25+00:00",{"id":94,"title":95,"news_slug":96,"published_at":97},"162c36d7-37de-4906-b0dd-cd5ec6f9f455","微软擦除基础模型LensErase:一图一掩码,万物皆可擦","lenserase-erase-anything-foundation-model","2026-10-15T21:08:38+00:00"]