Act-PRM Rollout Dataset
Run Metadata
run_name: act-prm-v2-ia=0-rr=0-ec=insurance_default_gpt5m-gc=default_openai-tc=rollout_only-rbc=default-mc=oai_gpt5m_med-nf=0-gs=4-s=0-r=0
run_cmd: main_pytorch.py --env_config insurance/default_gpt5m --model_config oai_gpt5m_med --generator_config default_openai --trainer_config rollout_only --replay_buffer_config default --log_path ./logs --group_size 4 --seed 0 --replicate 0 --verbose --eval_every 0
split: eval
num_trajectories: 20… See the full description on the dataset page:
https://huggingface.co/datasets/mzio/aprm-sft_rollouts-Eact-Gact-Sact-Ract-eval.
