RLTasksby ulam.ai
Sign inAll tasks
Machine learning research5ML-4

5ML-4

Five frontier machine-learning environments for diagnosing complex systems under severe model misspecification. Across sparse expert routing, multimodal alignment, neural operators, preference optimization, and world-model planning, agents get only five experiments to identify one of 256 mechanism stacks, estimate latent parameters, and produce calibrated intervals for sealed transfer forecasts.

ML researchExperiment designRLVR
Version5.0.0
Environments5
RewardScalar · 0–1
DeliveryPrivate ZIP
01 Task contract

A real, versioned RL environment.

Five frontier machine-learning environments for diagnosing complex systems under severe model misspecification. Across sparse expert routing, multimodal alignment, neural operators, preference optimization, and world-model planning, agents get only five experiments to identify one of 256 mechanism stacks, estimate latent parameters, and produce calibrated intervals for sealed transfer forecasts.

Agent objective

  • Interact with the supplied stateful environment.
  • Produce verifier-checkable actions or artifacts.
  • Maximise scalar reward under the package contract.

Evaluation

  • 5 verifier-backed environments.
  • Reported reward range 0–1.
  • Package-specific public and private checks.

Delivery boundary

  • Private object stored in Cloudflare R2.
  • Authenticated entitlement required.
  • Short-lived signed URL per download.
02 What you will work on

Distinct environments, one demanding research contract.

Enough detail to understand the intellectual terrain; generated instances, hidden mechanisms, and solution paths remain inside the private package.

EnvironmentMathematical or technical frontierAdaptive research problem
SparseExpertRoutingLabMixture-of-experts routing and systemsSeparate routing mechanisms that match average utilization but diverge under overflow, tail-token, shift, and communication stress.
MultimodalAlignmentStressLabCross-modal alignment and groundingUse corruption, conflict, missing-modality, and localization tests to distinguish stacks near modality-dominance transitions.
NeuralOperatorInductiveBiasLabNeural operators and physical constraintsReverse-engineer architecture and constraints across resolution, boundaries, stiffness, geometry, and long rollouts.
PreferenceOptimizationForensicsPreference learning and safetyProbe noisy, shifted, adversarial, and safety-conflicting comparisons to expose pipelines that look alike on ordinary accuracy.
WorldModelPlanningAutopsyModel-based RL and temporal abstractionTest long-horizon planning under stochasticity, aliasing, delay, bias, and shift where one-step prediction hides compounding errors.
03 Why it is interesting

What the supplied evaluation reveals.

In a direct sealed evaluation, a GPT-5.6 Pro chat instance made legal experiment calls on one fresh expert case per environment. All submissions were valid, but it selected the wrong 256-way mode every time, averaging 0.3128 with no passes. The five-call sample is small, yet it cleanly exposes forecasting, interval, and latent-estimation difficulty beyond competent experiment selection. The package passed 121 source and wheel tests plus ten replays.

We publish aggregate behavior and task structure, while withholding generated instances, hidden labels, exact successful probes, private checks, and solution trajectories.

04 Supplied evaluation

Observed evaluation result.

Shown with its provenance and limitations; it is not a performance guarantee.

i
Methodology matters

Reported result from the evaluation artifact supplied with this package.

Evaluated system / policyGPT-5.6 Pro

As identified by the supplied artifact.

Mean reward0.3128

1 reported runs.

Result artifactIncluded

gpt56_blind_report.md

Public result record

Machine-readable provenance and the exact displayed metric are available in results.json.

Open result JSON
05 Private delivery

The package stays off the public website.

The paid ZIP will live in a private R2 bucket. Vercel authorizes the buyer and issues a 2–5 minute object URL; R2 serves the bytes directly.

Included with purchase

  • Exact package version 5.0.0
  • Environment and task contracts
  • Verifier or scoring interface
  • Supplied reference/evaluation artifacts
  • Purchase record and licence v1.1
delivery flow
Authenticated buyer + entitlement check
+ private R2 object + 2–5 minute signed URL
= direct, auditable download

Package SHA-256
14448e1c7a42dd2692ef3ea9bd9e39158e5ba306ca23472431da19ba787d5a0c
06 Licence v1.1

Commercial use, without exclusivity.

One purchase licenses this identified item to one legal organisation for worldwide, perpetual commercial model training, evaluation, research and development. Redistribution and resale of the package are not permitted.

Read the full licenceYotta Content LTD · business customers only
5ML-4

Ready to add this environment?

Back to marketplace
Stripe checkout

Business purchase confirmation

Sign in or create an account, then complete secure Stripe Checkout. Access is granted only by the verified payment webhook.