RLVR-trained Llama-3.3-70B model for generating 5-room RPLAN floor plans from structured room and adjacency specifications.
Model Details
Base model: meta-llama/Llama-3.3-70B-Instruct
Training stages: supervised fine-tuning followed by RLVR/GRPO
Task: RPLAN 5-room floor-plan generation
Format: full merged causal language model checkpoint
Reported RLVR Metrics
Metric
Value
Compatibility
0.01 +/- 0.14
Diversity
8.96 +/- 0.00
Intended Use
Use this model to generate 5-room residential floor plans in the JSON format used by the associated RPLAN experiments. It is intended for research and evaluation on RPLAN-style floor-plan generation.
Limitations
The model is specialized to the 5-room RPLAN task and may not generalize to other datasets, room-count regimes, architectural standards, or safety-critical design workflows.
