24 lines
799 B
JSON
24 lines
799 B
JSON
|
|
{
|
||
|
|
"bf16": true,
|
||
|
|
"dataset_eval_split": "validation",
|
||
|
|
"dataset_name": "vwxyzjn/summarize_from_feedback_tldr_3_filtered_oai_preprocessing_1706381144",
|
||
|
|
"ddp_find_unused_parameters": false,
|
||
|
|
"eval_steps": 0.2,
|
||
|
|
"evaluation_strategy": "steps",
|
||
|
|
"git": "7906781",
|
||
|
|
"gradient_accumulation_steps": 4,
|
||
|
|
"gradient_checkpointing": false,
|
||
|
|
"hub_model_id": "mnoukhov/pythia410m-sft-tldr",
|
||
|
|
"learning_rate": "3e-6",
|
||
|
|
"logging_steps": 100,
|
||
|
|
"lr_scheduler_type": "cosine",
|
||
|
|
"max_seq_length": 580,
|
||
|
|
"model_name": "EleutherAI/pythia-410m-deduped",
|
||
|
|
"name": "newsft_pythia410m_tldr.yml",
|
||
|
|
"num_train_epochs": 1,
|
||
|
|
"per_device_eval_batch_size": 8,
|
||
|
|
"per_device_train_batch_size": 32,
|
||
|
|
"push_to_hub": true,
|
||
|
|
"report_to": "wandb",
|
||
|
|
"task_type": "tldr"
|
||
|
|
}
|