_wandb:
    value:
        cli_version: 0.19.11
        m: []
        python_version: 3.13.2
        t:
            "1":
                - 1
                - 11
                - 49
                - 51
                - 55
                - 95
            "2":
                - 1
                - 11
                - 49
                - 51
                - 55
                - 95
            "3":
                - 13
                - 16
                - 23
                - 55
            "4": 3.13.2
            "5": 0.19.11
            "6": 4.51.3
            "8":
                - 5
            "12": 0.19.11
            "13": linux-x86_64
batch_size:
    value: -1
data_path:
    value: humor_dataset.jsonl
data_path_to_save_groups:
    value: data/humor_1.jsonl
ensure_scores_are_not_same:
    value: false
eval_handling:
    value: STOP_TRAIN
eval_limit_ratio:
    value: 0.5
group_size:
    value: 2
include_messages:
    value: true
inference_weight:
    value: 1
max_batches_offpolicy:
    value: 3
max_eval_workers:
    value: 16
max_num_workers:
    value: -1
max_num_workers_per_node:
    value: 8
max_token_length:
    value: 2048
min_items_sent_before_logging:
    value: 2
num_rollouts_per_group_for_logging:
    value: 1
num_rollouts_to_keep:
    value: 32
rollout_server_url:
    value: http://localhost:8000
steps_per_eval:
    value: 100
tokenizer_name:
    value: NousResearch/DeepHermes-3-Llama-3-3B-Preview
total_steps:
    value: 2
use_wandb:
    value: true
wandb_name:
    value: null