Research map / Probabilistic and scientific modeling / Variational and generative sampling
Tandem Inference: An Out-of-Core Streaming Algorithm for Very Large-Scale Relational Inference
Sriram Srinivasan, Eriq Augustine, Lise Getoor
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 TI streaming inference instead of SGD/ADMM | PSL inference · JESTER-FULL | total runtime: about 8 times | Improved |
| use TI streaming inference instead of SGD/ADMM | PSL inference · JESTER | total runtime: over 5 times | Improved |
| use TI instead of SGD/ADMM | PSL inference · EPINIONS, CITESEER-ER, LASTFM | total runtime | Improved |
| use TI instead of SGD/ADMM | PSL inference · CITESEER and CORA | convergence speed | Hurt |
| use TI instead of in-memory SGD/ADMM | PSL inference · FRIENDSHIP-500M | feasibility under 400GB RAM: under four hours using only 10GB | Improved |
| use TI instead of in-memory methods | PSL inference · FRIENDSHIP-1B | feasibility under 400GB RAM: under nine hours using only 10GB | Improved |
| decrease page size to 10 and 100 (ablation) | TI · JESTER-FULL | peak JVM memory | Hurt |
| decrease page size to 10 and 100 (ablation) | TI · JESTER-FULL | time per iteration | Hurt |
| increase page size above 1000 (ablation) | TI · JESTER-FULL | time per iteration | No clear effect |
| increase page size (ablation) | TI · JESTER-FULL | I/O operations per iteration | Improved |
| use constant learning rate instead of decaying (ablation) | SGD · LASTFM, JESTER | iterations to converge | Hurt |
| use Adam adaptive learning rate instead of time-decayed (ablation) | SGD · LASTFM, JESTER | iterations to converge | No clear effect |
Closest papers by meaning
- OptScale: Probabilistic Optimality for Inference-time Scaling (AAAI 2026 · similarity 0.86)
- The Silent Hyperparameter: Quantifying the Impact of Inference Backends on LLM Reproducibility (NeurIPS 2026 · similarity 0.86)
- Tandem Transformers for Inference Efficient LLMs (ICML 2024 · similarity 0.86)
- Entropy-Gated Latent Recursion (NeurIPS 2026 · similarity 0.85)
- B'MOJO: Hybrid State Space Realizations of Foundation Models with Eidetic and Fading Memory (NeurIPS 2024 · similarity 0.85)
- Scaling Laws Meet Model Architecture: Toward Inference-Efficient LLMs (ICLR 2026 · similarity 0.85)
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.
