Math RL Tasks B
AI benchmark · Math
- Tier
- B why
- Snapshot
What is Math RL Tasks?
Math RL Tasks scores models on 20 constructive mathematics problems where an executable verifier checks a submitted mathematical object or certificate and returns a binary reward.
Math mostly
“Math RL Tasks is a public collection of Ulam's mathematical construction problems for reinforcement learning with verifiable rewards (RLVR) and model evaluation.”
huggingface.co/datasets/ulamai/Math-RL-Tasks ↗
Tier B
Established: 15 or more models scored, or curated RL directories rate it.
How much stands behind a tier: usage, who backs it, a confirmed line from its own site or paper, and a clear read on X. High with three of the four, medium with two, low otherwise.
- listed with Ulam
- in curated RL directories (3 pts)
How we got to tier B
- Adoption: listed with Ulam. Tier C.
- On X: unclear, none buzz No move.
- In curated RL directories: 3 points, a floor of B. Raised to B.
- Confirmed: a line from its own site or paper places it.
A clear view on X moves a tier one step up or down; most entries don't move. How tiers work.
Training methods
Methods Math RL Tasks uses, offers or builds for, with public evidence: its own website says so, or two independent sites do.
- RLVR Reinforcement Learning with Verifiable Rewards On its own website
“2.564 / 4 A-SFT + V-RLVR · rank 2”
ulam.ai“Strict RLVR rows are an explicitly promoted subset—not a label for the whole corpus. The public release provides 10 inspection examples, schema, and quality-gate definitions.”
ulam.ai - DPO Direct Preference Optimization On its own website
“Training views: SFT, DPO, critic and repair data, RLVR/GRPO tasks, process-reward steps, private evaluation, and regression tests.”
ulam.ai“Preferences Verifier-ranked pairs Failed and corrected attempts become DPO-compatible signal.”
ulam.ai - GRPO Group Relative Policy Optimization On its own website
“Training views: SFT, DPO, critic and repair data, RLVR/GRPO tasks, process-reward steps, private evaluation, and regression tests.”
ulam.ai“Identity Exact checkpoint lineage Base, SFT, DPO, and GRPO students remain distinguishable.”
ulam.ai - SFT Supervised Fine-Tuning On its own website
“Tests shaped verification rewards alone and after specialized SFT, separating proof-critique skill from broad completion reliability.”
ulam.ai“Useful for tactic SFT, critics, proof repair, preference data, RLVR, and regression. Formal status applies to Lean-accepted proof steps.”
ulam.ai
Where Math RL Tasks sits
Its domain on the map, live. Hover a logo for its name; click to open it. Open it on the full map
Next: browse all benchmarks. Entry from the RL Research daily scrape of public sources, 2026-10-31 snapshot.