Atomic Lattice Grid
Physical Alignment
HARDNESS0 / 0
CONDUCTIVITY0 / 0
THERMAL RES.0 / 0
ELASTICITY0 / 0
Stability: 0.000
Symmetry: 0.000
Reward Convergence Timeline
Synthesis Performance
Historical benchmark metadata for the current laboratory session.
Phase Distribution
Success Rate by Difficulty
Technical Integration
Overview
MaterialForge is a reinforcement learning benchmark focused on inverse materials discovery. Agents are tasked with positioning atomic species in an 8×8 crystalline lattice to achieve a set of target macro-physical properties.
REST API
POST /reset
POST /step
Websocket
ws://{host}/ws