MarginFlow turns SIS proposal design into GFlowNet learning, matching 99.8% effective samples zero-shot
Tsinghua · hf · 2026-09-30
Tsinghua's MarginFlow proves the ideal sequential importance sampling proposal for counting/uniformly sampling binary matrices with fixed margins is exactly a GFlowNet policy with unit reward, turning proposal design into a learning problem amortized across margins via a set transformer. Trained on 1904 margins, it matches or beats the best of 31 analytic proposals on 1187/1190 zero-shot margins (median effective sample fraction 99.8%), and on the 56 hardest margins lifts the median from 10.3% to 94.1%.
More from Research
- Explaining just 5% of token positions retains nearly all audit success across 4.7M explanations — aisilab · 2026-09-30
- NTU's Persistence Forcing hits FID 1.63 on ImageNet 256 by heterogeneous refinement in pixel-space DiTs — NanyangTechnologicalUniversity · 2026-09-30
- IBM's Q&D trains proactive agents to ask better questions, beating a 15x larger model — ibm · 2026-09-30
- Researcher predicts AI labs will soon pivot from math conjectures to materials and drug discovery — tak3sh8 · 2026-09-30
- Video lecture series by Stephen Wright, Yousef Saad and Peter Bartlett on ML optimization now available — caglar_ee · 2026-09-30
- Arbor: open-source framework for AI agents doing autonomous long-horizon research — burkov · 2026-09-30