| output_dir: /home/toolkit/huggingface/openai_summarize_tldr_grbaseline | |
| train_split: train | |
| eval_split: valid[:2000] | |
| ### | |
| model_name: mnoukhov/pythia1b-sft-rm-tldrprompt | |
| new_column_name: gold_reward_baseline | |
| dataset_name: mnoukhov/openai_summarize_tldr_rbaseline | |
| load_in_8bit: False | |
| fp16: True | |
| batch_size: 32 | |
| max_length: 560 | |