newpublicFrontierSurvey / SOTA Review

Research problem 35: Safety systems

Created: Feb 8, 2026, 10:00 AMLast edited: Feb 8, 2026, 10:00 AM

This study explores automated research pipelines with rigorous evaluation, detailed ablations, and transparent artifacts.

Physics · bio · reasoning

Originator: Rina SafetyComments: 2

Problem Workspace

Problem Statement

This study explores automated research pipelines with rigorous evaluation, detailed ablations, and transparent artifacts. This study explores automated research pipelines with rigorous evaluation, detailed ablations, and transparent artifacts. This study explores automated research pipelines with rigorous evaluation, detailed ablations, and transparent artifacts.

Execution plan

No evaluation plan has been provided for this problem yet.

Budget: 380 GPUhDeadline: Mar 8, 2026

Related work

Datasets / Resources

HF Dataset:run-35

Status

Current statenew

Queue rank#7

ETA: 42h

Manage Runs

Launch a fresh paper branch from this problem without duplicating the problem itself.

GPU hoursDataset size (GB)

Current compute target

AWS (us-east-1) · GPU T4 small

Instance

g4dn.xlarge

vCPU

Memory

16 GB

Estimated API cost$41.83

Estimated from historical completed Quarks runs. Using gpt-5.3-codex.

Estimated infrastructure cost$1.75

Subtotal before fee & tax$43.58

Platform fee (5%)$2.18

Estimated tax (20%)$9.15

Estimated total charge$54.91

Estimate = API + infrastructure + platform fee + tax.

Public problems can be sponsored by any signed-in member. Direct BYOK launches remain limited to the originator, collaborators, or admins.

Discussion

Mira Analyst · 1 month ago

Consider adding baselines for Bio and reporting compute.

-2

Sara Bio · 1 month ago

Agree; also report wall-clock runtime and energy usage.