Timofei Gritsaev, Evgeny Katasonov, Ildus Sadrtdinov, Esmeralda S. Whitammer, Dmitry Vetrov: On the Generation Order in GFlowNets https://arxiv.org/abs/2610.41648 https://arxiv.org/pdf/2610.41648 https://arxiv.org/html/2610.41648
November 2, 2026 at 6:43 AM
Timofei Gritsaev, Evgeny Katasonov, Ildus Sadrtdinov, Esmeralda S. Whitammer, Dmitry Vetrov: On the Generation Order in GFlowNets https://arxiv.org/abs/2610.41648 https://arxiv.org/pdf/2610.41648 https://arxiv.org/html/2610.41648
Omar G. Younis, Joseph D. Viviano, Chirayu Anant Haryan, MD Vasimuddin, Sanchit Misra, Yoshua Bengio: Mixture of GFlowNets https://arxiv.org/abs/2610.17007 https://arxiv.org/pdf/2610.17007 https://arxiv.org/html/2610.17007
October 14, 2026 at 6:48 AM
Omar G. Younis, Joseph D. Viviano, Chirayu Anant Haryan, MD Vasimuddin, Sanchit Misra, Yoshua Bengio: Mixture of GFlowNets https://arxiv.org/abs/2610.17007 https://arxiv.org/pdf/2610.17007 https://arxiv.org/html/2610.17007
Rui Hu, Yu Chen, Zhuoran Li, Longbo Huang: Towards Understanding the Training of GFlowNets: A Theoretical Perspective https://arxiv.org/abs/2610.13711 https://arxiv.org/pdf/2610.13711 https://arxiv.org/html/2610.13711
October 12, 2026 at 6:44 AM
Rui Hu, Yu Chen, Zhuoran Li, Longbo Huang: Towards Understanding the Training of GFlowNets: A Theoretical Perspective https://arxiv.org/abs/2610.13711 https://arxiv.org/pdf/2610.13711 https://arxiv.org/html/2610.13711
Tiago da Silva, Amauri H. Souza, Salem Lahlou: Learning a Mixture of GFlowNets https://arxiv.org/abs/2610.07562 https://arxiv.org/pdf/2610.07562 https://arxiv.org/html/2610.07562
October 7, 2026 at 6:44 AM
Tiago da Silva, Amauri H. Souza, Salem Lahlou: Learning a Mixture of GFlowNets https://arxiv.org/abs/2610.07562 https://arxiv.org/pdf/2610.07562 https://arxiv.org/html/2610.07562
New arXiv paper proposes using Generative Flow Networks to create diverse synthetic expert conversations for LLM post-training, addressing mode collapse in direct prompting by sampling latent interaction…
#GFlowNets #SyntheticData #LLMPostTraining #Prompting
https://arxiv.org/abs/2609.38359
#GFlowNets #SyntheticData #LLMPostTraining #Prompting
https://arxiv.org/abs/2609.38359
October 1, 2026 at 4:01 AM
New arXiv paper proposes using Generative Flow Networks to create diverse synthetic expert conversations for LLM post-training, addressing mode collapse in direct prompting by sampling latent interaction…
#GFlowNets #SyntheticData #LLMPostTraining #Prompting
https://arxiv.org/abs/2609.38359
#GFlowNets #SyntheticData #LLMPostTraining #Prompting
https://arxiv.org/abs/2609.38359
Tiago da Silva, Diego Mesquita, Salem Lahlou: Particle GFlowNets: Rethinking Generative Marginalization Models https://arxiv.org/abs/2609.11538 https://arxiv.org/pdf/2609.11538 https://arxiv.org/html/2609.11538
September 11, 2026 at 6:43 AM
Tiago da Silva, Diego Mesquita, Salem Lahlou: Particle GFlowNets: Rethinking Generative Marginalization Models https://arxiv.org/abs/2609.11538 https://arxiv.org/pdf/2609.11538 https://arxiv.org/html/2609.11538
New preprint 🚨!
I’m really excited about this work led by @ishume.bsky.social :)
We introduce Grinch, a framework that uses GFlowNets to construct quantum optical circuit graphs from the bottom up, edge-by-edge.
📜: arxiv.org/abs/2609.08073
#AIForScience #ML4Phys
I’m really excited about this work led by @ishume.bsky.social :)
We introduce Grinch, a framework that uses GFlowNets to construct quantum optical circuit graphs from the bottom up, edge-by-edge.
📜: arxiv.org/abs/2609.08073
#AIForScience #ML4Phys
September 10, 2026 at 5:58 PM
New preprint 🚨!
I’m really excited about this work led by @ishume.bsky.social :)
We introduce Grinch, a framework that uses GFlowNets to construct quantum optical circuit graphs from the bottom up, edge-by-edge.
📜: arxiv.org/abs/2609.08073
#AIForScience #ML4Phys
I’m really excited about this work led by @ishume.bsky.social :)
We introduce Grinch, a framework that uses GFlowNets to construct quantum optical circuit graphs from the bottom up, edge-by-edge.
📜: arxiv.org/abs/2609.08073
#AIForScience #ML4Phys
Excited to share Grinch! We use GFlowNets to build graphs of quantum optical experiments, guiding design with logic-based masking functions and finding multiple high-fidelity solutions for target states and photonic gates while respecting connectivity constraints :)
arxiv.org/abs/2609.08073
arxiv.org/abs/2609.08073
September 10, 2026 at 7:51 PM
Excited to share Grinch! We use GFlowNets to build graphs of quantum optical experiments, guiding design with logic-based masking functions and finding multiple high-fidelity solutions for target states and photonic gates while respecting connectivity constraints :)
arxiv.org/abs/2609.08073
arxiv.org/abs/2609.08073
Grinch, a new reward-driven generative framework using GFlowNets, enables automated design of quantum optical circuits by learning directly from fidelity-based rewards without pre-existing datasets. Demonstrates diverse high-fidelity solutions for entangled states, graph state...
Bottom-Up Generative Design of Quantum Optical Circuits with GFlowNets
arxiv.org
September 9, 2026 at 11:35 AM
Grinch, a new reward-driven generative framework using GFlowNets, enables automated design of quantum optical circuits by learning directly from fidelity-based rewards without pre-existing datasets. Demonstrates diverse high-fidelity solutions for entangled states, graph state...
Grinch, ein neues belohnungsgesteuertes generatives Framework mit GFlowNets, ermöglicht das automatisierte Design von quantenoptischen Schaltkreisen durch direktes Lernen aus Fidelitätsbasierten Belohnungen ohne vorhandene Datensätze. Demonstriert vielfältige hochfidele Lösung...
Bottom-Up Generatives Design von Quantenoptischen Schaltkreisen mit GFlowNets
arxiv.org
September 9, 2026 at 11:37 AM
Grinch, ein neues belohnungsgesteuertes generatives Framework mit GFlowNets, ermöglicht das automatisierte Design von quantenoptischen Schaltkreisen durch direktes Lernen aus Fidelitätsbasierten Belohnungen ohne vorhandene Datensätze. Demonstriert vielfältige hochfidele Lösung...
Grinch — это новая управляемая вознаграждением обобщающая структура, использующая GFlowNets, которая обеспечивает автоматизированное проектирование квантово-оптических схем путем прямого обучения на основе вознаграждений, основанных на точности, без предварительно существующих...
Генеративное проектирование квантово-оптических схем снизу вверх с использованием GFlowNets
arxiv.org
September 9, 2026 at 11:38 AM
Grinch — это новая управляемая вознаграждением обобщающая структура, использующая GFlowNets, которая обеспечивает автоматизированное проектирование квантово-оптических схем путем прямого обучения на основе вознаграждений, основанных на точности, без предварительно существующих...
Grinch, un nuevo marco generativo impulsado por recompensas que utiliza GFlowNets, permite el diseño automatizado de circuitos ópticos cuánticos aprendiendo directamente de recompensas basadas en fidelidad sin conjuntos de datos preexistentes. Demuestra soluciones diversas de ...
Diseño Generativo de Abajo hacia Arriba de Circuitos Ópticos Cuánticos con GFlowNets
arxiv.org
September 9, 2026 at 11:36 AM
Grinch, un nuevo marco generativo impulsado por recompensas que utiliza GFlowNets, permite el diseño automatizado de circuitos ópticos cuánticos aprendiendo directamente de recompensas basadas en fidelidad sin conjuntos de datos preexistentes. Demuestra soluciones diversas de ...
Grinch, un nouveau framework génératif basé sur les récompenses utilisant GFlowNets, permet la conception automatisée de circuits optiques quantiques en apprenant directement des récompenses basées sur la fidélité sans ensembles de données préexistants. Démontre des solutions ...
Conception générative ascendante de circuits optiques quantiques avec GFlowNets
arxiv.org
September 9, 2026 at 11:36 AM
Grinch, un nouveau framework génératif basé sur les récompenses utilisant GFlowNets, permet la conception automatisée de circuits optiques quantiques en apprenant directement des récompenses basées sur la fidélité sans ensembles de données préexistants. Démontre des solutions ...
Grinch, um novo framework generativo orientado por recompensa usando GFlowNets, permite o design automatizado de circuitos ópticos quânticos ao aprender diretamente de recompensas baseadas em fidelidade sem conjuntos de dados pré-existentes. Demonstra diversas soluções de alta...
Design Generativo Bottom-Up de Circuitos Ópticos Quânticos com GFlowNets
arxiv.org
September 9, 2026 at 11:39 AM
Grinch, um novo framework generativo orientado por recompensa usando GFlowNets, permite o design automatizado de circuitos ópticos quânticos ao aprender diretamente de recompensas baseadas em fidelidade sem conjuntos de dados pré-existentes. Demonstra diversas soluções de alta...
Grinch는 GFlowNets를 사용하는 새로운 보상 기반 생성 프레임워크로, 기존 데이터셋 없이 충실도 기반 보상으로부터 직접 학습하여 양자 광학 회로의 자동 설계를 가능하게 합니다. 얽힌 상태 및 그래프 상태에 대한 다양한 높은 충실도 해법을 제시합니다...
#포토닉양자 #양자회로설계 #연구
#포토닉양자 #양자회로설계 #연구
GFlowNets를 이용한 양자 광학 회로의 상향식 생성 설계
arxiv.org
September 9, 2026 at 11:36 AM
Grinch، إطار عمل توليدي جديد يعتمد على المكافآت ويستخدم GFlowNets، يتيح التصميم الآلي للدوائر البصرية الكمومية من خلال التعلم المباشر من المكافآت المستندة إلى الدقة دون الحاجة إلى مجموعات بيانات موجودة مسبقاً. يوضح حلولاً متنوعة عالية الدقة للحالات المتشابكة وحالات الرسم البيا...
التصميم التوليدي من الأسفل للأعلى للدوائر البصرية الكمومية باستخدام GFlowNets
arxiv.org
September 9, 2026 at 11:38 AM
Grinch، إطار عمل توليدي جديد يعتمد على المكافآت ويستخدم GFlowNets، يتيح التصميم الآلي للدوائر البصرية الكمومية من خلال التعلم المباشر من المكافآت المستندة إلى الدقة دون الحاجة إلى مجموعات بيانات موجودة مسبقاً. يوضح حلولاً متنوعة عالية الدقة للحالات المتشابكة وحالات الرسم البيا...
Grinch, מסגרת גנרטיבית חדשה מונעת תגמול המשתמשת ב-GFlowNets, מאפשרת עיצוב אוטומטי של מעגלי אופטיקה קוונטית על ידי למידה ישירה מתגמולים מבוססי נאמנות ללא מערכות נתונים קיימות. מדגימה פתרונות רבים בעלי נאמנות גבוהה למצבים שזורים, מצב גרף...
#פוטוני_קוונטי #עיצוב_מעגלים_קוונטיים #מחקר
#פוטוני_קוונטי #עיצוב_מעגלים_קוונטיים #מחקר
עיצוב גנרטיבי מלמטה למעלה של מעגלי אופטיקה קוונטית עם GFlowNets
iq.fp2.dev
September 9, 2026 at 11:39 AM
Grinch, מסגרת גנרטיבית חדשה מונעת תגמול המשתמשת ב-GFlowNets, מאפשרת עיצוב אוטומטי של מעגלי אופטיקה קוונטית על ידי למידה ישירה מתגמולים מבוססי נאמנות ללא מערכות נתונים קיימות. מדגימה פתרונות רבים בעלי נאמנות גבוהה למצבים שזורים, מצב גרף...
#פוטוני_קוונטי #עיצוב_מעגלים_קוונטיים #מחקר
#פוטוני_קוונטי #עיצוב_מעגלים_קוונטיים #מחקר
Grinch, GFlowNets का उपयोग करने वाली एक नई पुरस्कार-संचालित जनरेटिव फ्रेमवर्क, पूर्व-मौजूदा डेटासेट के बिना निष्ठा-आधारित पुरस्कार से सीधे सीखकर क्वांटम ऑप्टिकल सर्किट के स्वचालित डिजाइन को सक्षम बनाता है। उलझी हुई अवस्थाओं, ग्राफ अवस्था के लिए विविध उच्च-निष्ठा समाधान प्रदर्श...
GFlowNets के साथ क्वांटम ऑप्टिकल सर्किट का तल-ऊपर की ओर जनरेटिव डिजाइन
arxiv.org
September 9, 2026 at 11:37 AM
Grinch, GFlowNets का उपयोग करने वाली एक नई पुरस्कार-संचालित जनरेटिव फ्रेमवर्क, पूर्व-मौजूदा डेटासेट के बिना निष्ठा-आधारित पुरस्कार से सीधे सीखकर क्वांटम ऑप्टिकल सर्किट के स्वचालित डिजाइन को सक्षम बनाता है। उलझी हुई अवस्थाओं, ग्राफ अवस्था के लिए विविध उच्च-निष्ठा समाधान प्रदर्श...
Grinch是一個使用GFlowNets的新型獎勵驅動生成框架，通過直接從保真度獎勵學習，無需預先存在的數據集，實現量子光學電路的自動化設計。展示了用於糾纏態、圖態等的多樣化高保真度解決方案...
#光子量子 #量子電路設計 #研究
#光子量子 #量子電路設計 #研究
使用GFlowNets的量子光學電路自下而上生成式設計
arxiv.org
September 9, 2026 at 11:35 AM
Grinch, một khung công việc tạo sinh dựa trên phần thưởng mới sử dụng GFlowNets, cho phép thiết kế tự động các mạch quang tử bằng cách học trực tiếp từ phần thưởng dựa trên độ chính xác mà không cần bộ dữ liệu sẵn có. Thể hiện các giải pháp độ chính xác cao đa dạng cho các trạ...
Thiết kế Mạch Quang Tử Tạo sinh từ Dưới lên bằng GFlowNets
arxiv.org
September 9, 2026 at 11:38 AM
Grinch, một khung công việc tạo sinh dựa trên phần thưởng mới sử dụng GFlowNets, cho phép thiết kế tự động các mạch quang tử bằng cách học trực tiếp từ phần thưởng dựa trên độ chính xác mà không cần bộ dữ liệu sẵn có. Thể hiện các giải pháp độ chính xác cao đa dạng cho các trạ...
Grinch是一个基于GFlowNets的新型奖励驱动生成框架,能够通过直接从基于保真度的奖励中学习,在无需预先数据集的情况下实现量子光学电路的自动化设计。展示了纠缠态、图态等多样化高保真解决方案...
#光子量子 #量子电路设计 #研究
#光子量子 #量子电路设计 #研究
利用GFlowNets的量子光学电路自底向上生成设计
iq.fp2.dev
September 9, 2026 at 11:35 AM
Grinchは、GFlowNetsを用いた新しい報酬駆動型生成フレームワークであり、既存のデータセットなしに忠実度ベースの報酬から直接学習することで、量子光学回路の自動設計を実現します。エンタングルド状態やグラフ状態などの多様な高忠実度ソリューションを示しています...
#光子量子 #量子回路設計 #研究
#光子量子 #量子回路設計 #研究
GFlowNetsを用いた量子光学回路のボトムアップ生成設計
arxiv.org
September 9, 2026 at 11:35 AM
Grinch เป็นกรอบการทำงานเชิงสร้างสรรค์ใหม่ที่ขับเคลื่อนโดยรางวัล ใช้ GFlowNets ซึ่งช่วยให้สามารถออกแบบวงจรแสงควอนตัมโดยอัตโนมัติ ผ่านการเรียนรู้โดยตรงจากรางวัลที่ขึ้นอยู่กับความเที่ยงตรง โดยไม่ต้องใช้ชุดข้อมูลที่มีอยู่ก่อนหน้า แสดงให้เห็นวิธีแก้ปัญหาที่มีความเที่ยงตรงสูงและหลาก...
การออกแบบสร้างสรรค์จากล่างขึ้นบนของวงจรแสงควอนตัมโดยใช้ GFlowNets
iq.fp2.dev
September 9, 2026 at 11:38 AM
Grinch เป็นกรอบการทำงานเชิงสร้างสรรค์ใหม่ที่ขับเคลื่อนโดยรางวัล ใช้ GFlowNets ซึ่งช่วยให้สามารถออกแบบวงจรแสงควอนตัมโดยอัตโนมัติ ผ่านการเรียนรู้โดยตรงจากรางวัลที่ขึ้นอยู่กับความเที่ยงตรง โดยไม่ต้องใช้ชุดข้อมูลที่มีอยู่ก่อนหน้า แสดงให้เห็นวิธีแก้ปัญหาที่มีความเที่ยงตรงสูงและหลาก...
![Emoji Legend
💨x# | "Increased Trending Rate" for a Word (meaningful value; "💨x8 - burger" will mean "burger" is trending at 8 times its own normal rate for this time of day)
🔓 | "Unlocked" Ongoing Emergency Word (such words are not weighted down by their own previous usage; their *ongoing* increased usage--weeks, months, years--would otherwise move them down the list.)
[PREVIOUSLY USED]
⚠️💨# | "Trend Strength Rating"
⚠️🌀💨 | "Trend Strength Rating" (see above)
⚠️🌀 CAT. # | [same; "CAT." = Category, like hurricane]
💬# | 10min Post Count
🆎# | 10min Word Count
🙂# | 10min Emoji Count](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:mcb6n67plnrlx4lg35natk2b/bafkreigp6bgag3jzkool6uxrujpcrwtn6zkd7cbfrmevhaxaqhhewgolmu)
Carine Ribeiro dos Santos, Ina P\"ohner: Risk-Sensitive Reward Composition for Conditional GFlowNets https://arxiv.org/abs/2609.01929 https://arxiv.org/pdf/2609.01929 https://arxiv.org/html/2609.01929
September 3, 2026 at 6:39 AM
Carine Ribeiro dos Santos, Ina P\"ohner: Risk-Sensitive Reward Composition for Conditional GFlowNets https://arxiv.org/abs/2609.01929 https://arxiv.org/pdf/2609.01929 https://arxiv.org/html/2609.01929