Atlas search
Find a concept, company, or signal.
Search runs entirely in your browser across the project's validated, Git-managed content. No query is sent to an external service.
20 results
Mercor
An expert-data platform supplying frontier training data, evaluation work, and RL environments.
AfterQuery
An applied research lab turning expert reasoning and real workflows into model training data.
Fleet
An applied research lab building high-fidelity training environments for AI agents.
Pretraining
The broad first phase in which a model learns patterns from a very large dataset.
Post-training
Training after pretraining that shapes a model into a useful, reliable assistant or agent.
Supervised fine-tuning
Training on curated examples that demonstrate desired inputs and outputs.
Data annotation
The process of adding labels, judgments, corrections, or other structure to data used for training or evaluation.
Preference data
Human or model judgments that compare outputs or score their quality.
RLHF
Reinforcement learning from human feedback used to align model behavior with human preferences.
Reward model
A learned scorer that predicts how desirable a model output is.
Reinforcement learning
Learning behavior through actions, feedback, and rewards in an environment.
Benchmark
A standardized set of tasks and scoring rules used to compare model performance.
Evaluation
The systematic measurement of a model's capabilities, quality, and failure modes.
Synthetic data
Artificially generated examples used for training or evaluation.
Agent environment
A controlled world in which an AI agent observes, acts, and receives feedback.
Inference
The process of running a trained model to produce an output or take an action.
From pretraining to agents: how the AI model-development lifecycle fits together
A system-level guide to how base models become instruction-following, evaluated, tool-using systems—and how evidence from operation feeds the next cycle.
APEX tests AI on economically valuable professional work
Mercor's APEX program uses expert-authored tasks to examine professional work in areas such as medicine, law, finance, and consulting.
AfterQuery reports research on agent training for Terminal-Bench 2.0
AfterQuery published work connecting expert-curated trajectories and tooling with improved results on an agent-oriented terminal benchmark.
Fleet centers its research thesis on high-fidelity agent environments
Fleet describes simulated worlds and real-world challenges designed to model work for the training and evaluation of AI agents.