Here is our latest work on using adversarial agents to solve optimization problems with evolutionary search. Accepted to ICML 2026 Workshop Scalable Learning and Optimization for Efficient Multimodal AI Agents (SCALE)</p>\n","updatedAt":"2026-06-09T14:16:58.302Z","author":{"_id":"67e8761654cc82de7b632159","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/9anqjWVQEBFRG7lyX-rcA.png","fullname":"john donaghy","name":"johnnyd-gensyn","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":6,"isUserFollowing":false}},"numEdits":3,"identifiedLanguage":{"language":"en","probability":0.8815875053405762},"editors":["johnnyd-gensyn"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/9anqjWVQEBFRG7lyX-rcA.png"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2605.27130","authors":[{"_id":"6a281f07e7d78ea7587e5045","name":"John Donaghy","hidden":false},{"_id":"6a281f07e7d78ea7587e5046","name":"Shikhar Rastogi","hidden":false}],"publishedAt":"2026-05-26T00:00:00.000Z","submittedOnDailyAt":"2026-06-09T00:00:00.000Z","title":"DEI: Diversity in Evolutionary Inference for Quality-Diversity Search","submittedOnDailyBy":{"_id":"67e8761654cc82de7b632159","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/9anqjWVQEBFRG7lyX-rcA.png","isPro":false,"fullname":"john donaghy","user":"johnnyd-gensyn","type":"user","name":"johnnyd-gensyn"},"summary":"We present DEI: Diversity in Evolutionary Inference, a distributed Quality-Diversity (QD) search framework that assigns heterogeneous large language models (LLMs) as mutation operators across peer nodes communicating with non-blocking collective operations. Unlike homogeneous parallel search, which replicates a single model's inductive biases across all workers, DEI treats each LLM's distinct creative prior as a complementary source of behavioral novelty. Extending the Digital Red Queen framework with DEI, nodes share local optimal solutions at the end of each round to seed the next round's population. This creates cross-model adversarial pressure that drives robustness beyond intra-model self-play. Evaluated on the Core War domain, a competitive programming benchmark in which Redcode warrior programs battle inside a simulated machine, a four-node heterogeneous ensemble (GPT-5.4-mini, Claude Sonnet 4.6, GPT-5.2, and Claude Haiku 4.5) achieves 124 percent higher merged-archive QD-Score (45.90 vs. 20.46) and 28 percent higher coverage (80.6 percent vs. 63.0 percent of cells) than a single-node baseline at equal total LLM-call budget. The heterogeneous ensemble also outperforms an equally-budgeted homogeneous ensemble on QD-Score, coverage, and held-out solution generality across all four model families. These results provide the first empirical evidence that model diversity, not merely parallelism, is the key driver of gain in distributed LLM-based QD search.","upvotes":9,"discussionId":"6a281f07e7d78ea7587e5047","githubRepo":"https://github.com/gensyn-ai/dei","githubRepoAddedBy":"user","ai_summary":"A distributed Quality-Diversity search framework uses heterogeneous large language models as mutation operators to enhance evolutionary inference, demonstrating that model diversity improves performance over homogeneous parallel approaches.","ai_keywords":["Quality-Diversity","evolutionary inference","distributed search","heterogeneous large language models","mutation operators","peer nodes","non-blocking collective operations","Digital Red Queen framework","local optimal solutions","cross-model adversarial pressure","Core War","Redcode warrior programs","QD-Score","coverage","held-out solution generality"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","githubStars":2,"organization":{"_id":"66d2530bb26010e571f0ea9b","name":"Gensyn","fullname":"Gensyn","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/66d252ec8a438492b0d6e4ce/KD6rJavI2N74-2aT7NiZc.jpeg"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"6387a1d43811d7ebe6e2f187","avatarUrl":"/avatars/5989f81f319b909300d3480fb2cda5e2.svg","isPro":false,"fullname":"Shikhar Rastogi","user":"shikharras","type":"user"},{"_id":"67a2bf2fe6242abd8d559af2","avatarUrl":"/avatars/67046c518d3303ff8417c419c75ab1cf.svg","isPro":false,"fullname":"Gabriel P Andrade","user":"gpandrad","type":"user"},{"_id":"67e9cb2df10424cf470b7b63","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/1bbHOzXiFbqedevkxdru9.png","isPro":false,"fullname":"Steve Glasper","user":"steevg","type":"user"},{"_id":"68dabf5d2f7e86ce6584b3d1","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/Dl_QainUfD2fm_sPoSZSC.png","isPro":false,"fullname":"Judson Bonneville","user":"jbonneville","type":"user"},{"_id":"68551c67d124b99adf1cf112","avatarUrl":"/avatars/75dd4f6391ed5d03faadbf4a8ede3574.svg","isPro":false,"fullname":"Xavier Finlayson","user":"X-underscore","type":"user"},{"_id":"67ec363f09073e6c29c154e3","avatarUrl":"/avatars/267787df1715f2ceb22ac0be4c5e48ef.svg","isPro":false,"fullname":"gasoline","user":"gasoline2255","type":"user"},{"_id":"67e8761654cc82de7b632159","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/9anqjWVQEBFRG7lyX-rcA.png","isPro":false,"fullname":"john donaghy","user":"johnnyd-gensyn","type":"user"},{"_id":"68937744c2759b97d101ccd1","avatarUrl":"/avatars/a34bf2f0b00a6bf93541131f9ccaafcd.svg","isPro":false,"fullname":"sanjay das","user":"sanjay00090","type":"user"},{"_id":"684dc8d46ab9daa0eb3586f4","avatarUrl":"/avatars/5f29094f5dd4e55e8e56ceb43916e5a8.svg","isPro":false,"fullname":"goku","user":"goku0007","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"66d2530bb26010e571f0ea9b","name":"Gensyn","fullname":"Gensyn","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/66d252ec8a438492b0d6e4ce/KD6rJavI2N74-2aT7NiZc.jpeg"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2605/2605.27130.md"}">
DEI: Diversity in Evolutionary Inference for Quality-Diversity Search
Abstract
A distributed Quality-Diversity search framework uses heterogeneous large language models as mutation operators to enhance evolutionary inference, demonstrating that model diversity improves performance over homogeneous parallel approaches.
We present DEI: Diversity in Evolutionary Inference, a distributed Quality-Diversity (QD) search framework that assigns heterogeneous large language models (LLMs) as mutation operators across peer nodes communicating with non-blocking collective operations. Unlike homogeneous parallel search, which replicates a single model's inductive biases across all workers, DEI treats each LLM's distinct creative prior as a complementary source of behavioral novelty. Extending the Digital Red Queen framework with DEI, nodes share local optimal solutions at the end of each round to seed the next round's population. This creates cross-model adversarial pressure that drives robustness beyond intra-model self-play. Evaluated on the Core War domain, a competitive programming benchmark in which Redcode warrior programs battle inside a simulated machine, a four-node heterogeneous ensemble (GPT-5.4-mini, Claude Sonnet 4.6, GPT-5.2, and Claude Haiku 4.5) achieves 124 percent higher merged-archive QD-Score (45.90 vs. 20.46) and 28 percent higher coverage (80.6 percent vs. 63.0 percent of cells) than a single-node baseline at equal total LLM-call budget. The heterogeneous ensemble also outperforms an equally-budgeted homogeneous ensemble on QD-Score, coverage, and held-out solution generality across all four model families. These results provide the first empirical evidence that model diversity, not merely parallelism, is the key driver of gain in distributed LLM-based QD search.
Community
Here is our latest work on using adversarial agents to solve optimization problems with evolutionary search. Accepted to ICML 2026 Workshop Scalable Learning and Optimization for Efficient Multimodal AI Agents (SCALE)
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2605.27130 in a model README.md to link it from this page.
Cite arxiv.org/abs/2605.27130 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2605.27130 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.