output_dir: summarize_from_feedback_tldr3_generated_relabel_20k_dpo_costa_1b_fp16.yml_3d94f50_b35a8
dataset_name: mnoukhov/summarize_from_feedback_tldr_3_filtered_oai_preprocessing_1706381144_unlabelled
model_name: mnoukhov/EleutherAI_pythia-1b-deduped__sft__tldr_dpo_costa_1b_fp16.yml_3d94f50_b35a8
base_model_name: vwxyzjn/EleutherAI_pythia-1b-deduped__sft__tldr
base_model_revision: sft__55513__1706646024
lora_model: True
tokenizer_name: EleutherAI/pythia-1b-deduped
train_split: train[:20000]
batch_size: 4
max_prompt_length: 512
max_new_tokens: 128
temperature: 0.7
top_p: 1.0
dtype: bfloat16