Download evaluation/run_simulator.py from davidwdw/b1k_model: direct link, hf CLI and curl.
- Browser
- Download file 431 Bytes
-
https://huggingface.co/davidwdw/b1k_model/resolve/main/evaluation/run_simulator.py
- Command line
-
hf download hf://davidwdw/b1k_model/evaluation/run_simulator.py
-
curl -L -o run_simulator.py https://huggingface.co/davidwdw/b1k_model/resolve/main/evaluation/run_simulator.py
431 Bytes
| """Official evaluator, with a fixed seed before each reset for paired comparisons.""" | |
| from omnigibson.eval import eval as official | |
| from omnigibson.eval.utils.eval_utils import seed_everything, DEFAULT_EVAL_SEED | |
| class PairedEvaluator(official.Evaluator): | |
| def reset(self): | |
| seed_everything(DEFAULT_EVAL_SEED) | |
| return super().reset() | |
| official.Evaluator=PairedEvaluator | |
| if __name__=='__main__': | |
| official.main() | |