Research map / Architectures and efficiency / RNNs and normalization
Device Placement Optimization with Reinforcement Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use RL-based device placement instead of the expert-designed placement | NMT · NMT batch 64, 1 CPU + 2 GPUs | running time: 4.99 -> 4.04 | Helped |
| Use RL-based device placement instead of the expert-designed placement | NMT · NMT batch 64, 1 CPU + 4 GPUs | running time: 4.73 -> 3.92 | Helped |
| Use RL-based device placement instead of the expert-designed placement | NMT · WMT14 English-German training | total training time: 229.57 -> 165.73 hours | Helped |
| Use RL-based placement instead of a single-GPU placement | RNNLM · RNNLM batch 64, 1 CPU + 2 GPUs | running time: 1.57 -> 1.57 | No clear effect |
| Use RL-based placement instead of a single-GPU placement | RNNLM · RNNLM batch 64, 1 CPU + 4 GPUs | running time: 1.57 -> 1.57 | No clear effect |
| Use RL-based placement instead of a single-GPU placement | Inception-V3 · batch 32, 1 CPU + 2 GPUs | running time: 4.60 -> 4.60 | No clear effect |
| Use RL-based placement instead of expert-designed placement | Inception-V3 · batch 32, 1 CPU + 4 GPUs | running time: 10.65 -> 3.85 | Helped |
Closest papers by meaning
- A Hierarchical Model for Device Placement (ICLR 2018 · similarity 0.94)
- Spotlight: Optimizing Device Placement for Training Deep Neural Networks (ICML 2018 · similarity 0.90)
- A Structure-Aware Framework for Learning Device Placements on Computation Graphs (NeurIPS 2024 · similarity 0.89)
- Integrated Hardware Architecture and Device Placement Search (ICML 2024 · similarity 0.88)
- Chip Placement with Diffusion Models (ICML 2025 · similarity 0.83)
- FlashRNN: I/O-Aware Optimization of Traditional RNNs on modern hardware (ICLR 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.