Research map / Architectures and efficiency / RNNs and normalization
Simple, Distributed, and Accelerated Probabilistic Programming
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 Edward2 with TPU data parallelism | VAE · 64x64 ImageNet | speedup: optimal linear speedup from 1 to 256 | Helped |
| Use Edward2 with TPU data parallelism | Image Transformer · 256x256 CelebA-HQ | speedup: optimal linear speedup from 1 to 256 | Helped |
| Run NUTS on GPUs using Edward2 | Bayesian logistic regression · Covertype | speedup over Stan: 100x | Helped |
| Run NUTS on GPUs using Edward2 | Bayesian logistic regression · Covertype | speedup over PyMC3: 37x | Helped |
| Use Edward2 tracing instead of handwritten TensorFlow code | NUTS · Covertype | time per leapfrog step: smaller than between-run variation | No clear effect |
Closest papers by meaning
- Deep Probabilistic Programming (ICLR 2017 · similarity 0.91)
- A Fast Convoluted Story: Scaling Probabilistic Inference for Integer Arithmetics (NeurIPS 2024 · similarity 0.86)
- Synergistic Tensor and Pipeline Parallelism (NeurIPS 2025 · similarity 0.85)
- Scaling Tractable Probabilistic Circuits: A Systems Perspective (ICML 2024 · similarity 0.85)
- Pfeife: Automatic Pipeline Parallelism for PyTorch (ICML 2025 · similarity 0.84)
- DOGE-Train: Discrete Optimization on GPU with End-to-End Training (AAAI 2024 · similarity 0.84)
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.