Use cases · Computer science
Training runs, ablations, benchmarks and algorithm analysis, set up properly, run to completion and reported with the curves and tables behind them.
Real runs
Each one is a complete run in production: the question as it was asked, every step Propab took, and everything it made.
On a small transformer trained on a character-level language modelling task, ablate positional encodings, layer norm placement and dropout one at a time, and measure the effect on validation loss. Deliver the ablation table, curves and a short report.
Ready today
Start from one of these, or ask in your own words. Propab combines them when your question needs more than one.
Algorithms and theory
Where a problem family becomes intractable on this machine, with fitted scaling across solvers and the theory that explains it.
Algorithms and theory
Runtime and solution quality of graph algorithms across generated graph families and sizes.
Algorithms and theory
Solver comparison on SMT-LIB with every model and unsat core checked.
Algorithms and theory
Closed forms, recurrences and inequalities checked symbolically to support an argument.
Algorithms and theory
Quality-vs-time curves of heuristics on TSPLIB-style instances against known optima.
Compilers and languages
An SMT proof or counterexample that an optimisation preserves semantics for all inputs.
Compilers and languages
Grammar written and tested, or ASTs of real code analysed.
Compilers and languages
IR built and run through optimisation passes, with before/after IR and JIT results.
Computer architecture
Arithmetic intensity vs attainable performance for a kernel on a named CPU/GPU from its datasheet numbers.
Computer vision
COCO-style mAP/AR for a set of detections the researcher has.
Computer vision
Feature detection, matching, homography and geometric correction on images.
Computer vision
Transfer-learning accuracy with calibration on a small public dataset (CIFAR-scale), CPU-sized.
Computer vision
Train and compare architectures on ImageNet-1k under a matched recipe.
Needs a GPU you connectComputer vision
Accuracy of pretrained models on CIFAR-10-C by corruption type and severity.
Computer vision
Detection/segmentation model trained and evaluated on COCO-style data.
Needs a GPU you connectComputer vision
Find mislabeled test items and compare accuracy on a re-collected test set (CIFAR-10.1).
Computer vision
NeRF/3D Gaussian splatting quality (PSNR/SSIM/LPIPS) on a standard scene set.
Needs a GPU you connectFormal methods
TLC run over a bounded instance: invariants hold or a counterexample trace.
Formal methods
Alloy analysis of a relational model within a stated scope.
Formal methods
Counterexamples to a Python function's pre/postconditions.
Formal methods
Minimal failing input for a stated property of real code.
Formal methods
A theorem stated and proved in Coq using the standard library.
Formal methods
SMT encoding of a transformation with a proof or counterexample.
Human-computer interaction
Mixed-effects analysis of a within-subjects experiment with effect sizes and CIs.
Human-computer interaction
Required sample size and a preregistration document with the analysis script dry-run on simulated data.
Human-computer interaction
Model fit, throughput and per-condition comparison from pointing trials.
Human-computer interaction
Codebook, coded excerpts and inter-rater agreement for qualitative data the researcher provides.
Human-computer interaction
Usage patterns and retention from in-the-wild interaction logs.
Human-computer interaction
Reliability (alpha/omega) and factor structure of a questionnaire.
Human-computer interaction
Re-run a published analysis from its OSF materials and compare numbers.
Information retrieval
nDCG/MAP/recall of BM25 on a public collection with qrels.
Information retrieval
Zero-shot dense-retriever results on all BEIR datasets.
Needs a GPU you connectInformation retrieval
Paired tests with multiple-comparison control over per-query metrics from run files.
Information retrieval
Retrieval and answer quality of a RAG pipeline on a QA benchmark.
Needs a GPU you connectInformation retrieval
SQL analytics over a large document or interaction corpus.
Machine learning
Seed-level significance test and effect size for a claimed gain.
Machine learning
Scaling exponents with bootstrap intervals from a run table.
Machine learning
Causal ablation of components on a CPU-sized task with seeds and compute matched.
Machine learning
Tuned baselines on an OpenML suite with the search budget disclosed.
Machine learning
Matched-recipe training runs of models too large for CPU.
Needs a GPU you connectMachine learning
lm-evaluation-harness scores for open models with the harness version pinned.
Needs a GPU you connectMachine learning
N-gram/near-duplicate overlap of a benchmark with a pretraining-corpus sample.
Machine learning
Re-run a paper's code from GitHub and compare its numbers.
Natural language and LLMs
Tokenizer behaviour and perplexity of small models on a corpus.
Natural language and LLMs
BLEU, chrF and ROUGE with signatures for system outputs.
Natural language and LLMs
Fine-tuned small model (≤ ~125M params) with evaluation against the base.
Natural language and LLMs
Parameter-efficient fine-tuning of a 1B+ model with a disclosed search budget.
Needs a GPU you connectNatural language and LLMs
BBQ/CrowS-Pairs-style results disaggregated by group.
Needs a GPU you connectNatural language and LLMs
Embedding index and retrieval quality over a document set.
Natural language and LLMs
TF-IDF/linear and small-transformer baselines on a public dataset with error analysis.
Natural language and LLMs
Throughput/latency of a served model under load (vLLM).
Needs a GPU you connectRobotics and control
MuJoCo/PyBullet simulation with a feedback controller and tracking metrics.
Robotics and control
LQR, pole placement, Bode and root-locus analysis of a plant.
Robotics and control
EKF/UKF/particle-filter estimates and consistency checks on a dataset.
Robotics and control
OMPL planners compared on success rate and path quality.
Robotics and control
Habitat navigation agents on standard scene datasets.
Needs a GPU you connectRobotics and control
Least-squares pose-graph optimisation with loop closures on a dataset.
Security and cryptography
Symbolic execution of a binary to a target block (e.g. a crackme's success path).
Security and cryptography
Sections, symbols, disassembly and emulation of an ELF binary.
Security and cryptography
Working attack (padding oracle, CBC bit-flip, small-exponent RSA) with explanation.
Security and cryptography
CVE counts, CWE mix and CVSS trends for a product or class.
Security and cryptography
Rule matches and false-positive checks over files the researcher supplies.
Security and cryptography
Attack AUC against a trained model and the effect of a defence.
Systems and databases
Query results, plans and engine comparison over multi-GB columnar data.
Systems and databases
Plan quality and regret across queries and scale factors.
Systems and databases
A distributed protocol run under a seeded scheduler until a correctness bug appears, with the replay.
Systems and databases
Percentile analysis with CIs and change-point detection on latency measurements.
Systems and databases
A paper's artifact re-executed from Zenodo/GitHub with results compared.