Directory label
Alignment: RL environments and AI benchmarks
A label other catalogs put on entries, not one of the domains on our map. On the map, each entry here sits in the domain its own site or paper leads with.
Steering model behavior toward human intent and preferences, through preference data, reward models and evaluations of safe, helpful responses.
- RL environments
- 3
- AI benchmarks
- 29
- Shares entries with
- 14 categories
Snapshot 2026-11-05. How categories are built · Where the data comes from
In short
- What are RL environments for alignment?
- A simulated workplace where an AI agent practises a task (a code repository, a browser, business software) and an automatic grader scores each attempt. Labs train on thousands of these attempts. Bought to improve models. For alignment, that means steering model behavior toward human intent and preferences, through preference data, reward models and evaluations of safe, helpful responses.
- Who builds RL environments for alignment?
- Surge AI, dmodel and Trajectory Labs.
- What are the main alignment benchmarks?
- RewardBench, MOOD, CommunityBench, AuAu, Stick To Your Role! and Agent-ValueBench, among 25 confirmed below.
RL environments 3
-
Surge AI
Sells RL environments“We create rich, complex RL environments that challenge agentic models in novel ways, and design verifiers that reward their behavior.” source ↗
-
dmodel
Sells RL environments“We build RL environments for open-ended interpretability tasks.” source ↗
-
Trajectory Labs
Sells RL environments“Create tasks, environments, and adversarial examples that expose consequential agent failures.” source ↗
AI benchmarks 29
-
RewardBench
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
MOOD
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
CommunityBench
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
AuAu
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
Stick To Your Role!
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
Agent-ValueBench
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
Beyond Value Benchmarks
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
Faithfulness Metrics Don't Measure Faithfulness
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
COMPASS
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
How Well Do Models Follow Their Constitutions?
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
JuICE
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
Misaligned by Reward
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
SNN Fairness Benchmark
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
The Compliance Gap
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
MUSE
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
A Human-Centric Pipeline for Aligning Large Language Models with Chinese Medical Ethics
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
Aristotelian Virtue Profiling of LLMs through Ethical Dilemmas
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
Evaluating Reward Model Generalization via Pairwise Maximum Discrepancy Competitions
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
Extending Beacon to Hindi
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
How Implicit Bias Accumulates and Propagates in LLM Long-term Memory
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
Pseudo-Deliberation in Language Models
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
RMGAP
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
The Wrong Kind of Right
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
Truthful AI Advisors
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
-
Uncovering Political Bias in Large Language Models using Parliamentary Voting Records
No line from its own site or paper places it in Alignment yet; the catalog lists it here.
4 not yet confirmedListed in the catalog; no line from its own site or paper confirms it yet.
Next: pick another domain. Tags come from the RL Research daily scrape, canonicalized nightly.