以下部分引用 AI 总结,现阶段 AI 仍然有幻觉。请以内容的原文为准。
数据新鲜度群聊 07-14 ✓卖方 07-14 ✓Wrap 断档15天 ✕News 07-15 ✓X 断档12天 ✕页面生成 07-15
FlowPlace
5 atoms · 跨 1 天 · 首见 2026-04-28 · 最近 2026-04-28
三色: 🟦 fact 5 · 🟥 take 0 · stance ▲0/▼0/◆0
来源: X 5
时态: fresh:5
标签: 好信源:5 · 好数字:1
事实 (fact) (5)
- 2026-04-28 ⭐ · achieves better PPA metrics, 10-50x faster sampling efficiency, and zero overlaps on OpenROAD and ICCAD 2015 benchmarks
The experimental results on OpenROAD and ICCAD 2015 benchmarks show FlowPlace achieves better PPA metrics, 10-50x faster sampling efficiency, and zero overlaps.
tags: 好数字·好信源 · → daily
value: qty=10-50x faster sampling efficiency
- 2026-04-28 · is introduced as a flow-based generative framework
In this paper is introduced FlowPlace, a flow-based generative framework that incorporates mask-guided generation, deterministic flow matching, and hard-constraint sampling to efficiently produce overlap-free chip placements.
tags: 好信源 · → daily
- 2026-04-28 · integrates boundary-aware sampling prior to generate macro placements with structural regularities
The proposed approach integrates a modular, boundary-aware sampling prior to generate macro placements with the structural regularities characteristic of high-quality designs, providing high-quality and scalable data for model training.
tags: 好信源 · → daily
- 2026-04-28 · learns a deterministic trajectory rather than reversing a stochastic process like diffusion models
FlowPlace learns a deterministic trajectory, rather than learning to reverse a stochastic process via a sequence of denoising steps like diffusion models.
tags: 好信源 · → daily
- 2026-04-28 · offers advantages including fewer sampling steps, flexible prior choice, and simpler training paradigm
This offers several advantages, including fewer sampling steps due to deterministic dynamics, flexible choice of prior distribution, and simpler training paradigm without noise schedule tuning.
tags: 好信源 · → daily
时间轴 (近 20)
- 2026-04-28 ·
fact · is introduced as a flow-based generative framework · → daily
- 2026-04-28 ·
fact · integrates boundary-aware sampling prior to generate macro p · → daily
- 2026-04-28 ·
fact · learns a deterministic trajectory rather than reversing a st · → daily
- 2026-04-28 ·
fact · offers advantages including fewer sampling steps, flexible p · → daily
- 2026-04-28 ·
fact · achieves better PPA metrics, 10-50x faster sampling efficien · → daily
← 实体目录 · 系统日志