423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-08-03 04:02:49 +08:00
Code Issues Packages Projects Releases Wiki Activity

feat: add batch size of map function in the preprocessed dataset

Browse Source 
Former-commit-id: 209313eeeab8d1a7c320bd9aa90a5f4656082b7c
This commit is contained in:
naem1023 2024-09-02 13:52:47 +09:00
parent 5af92971bc
commit 46695e42cc
2 changed files with 7 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
src/llamafactory/data/loader.py
View File

@ -179,6 +179,9 @@ def _get_preprocessed_dataset(
load_from_cache_file=(not data_args.overwrite_cache) or (training_args.local_process_index != 0), load_from_cache_file=(not data_args.overwrite_cache) or (training_args.local_process_index != 0),
desc="Running tokenizer on dataset", desc="Running tokenizer on dataset",
) )
if data_args.dataset_map_batch_size:
# Set the batch size conditionally without considering the default variable of the batch size in the map function
kwargs.update(batch_size=data_args.dataset_map_batch_size)
dataset = dataset.map(preprocess_func, batched=True, remove_columns=column_names, **kwargs) dataset = dataset.map(preprocess_func, batched=True, remove_columns=column_names, **kwargs)

4
src/llamafactory/hparams/data_args.py
View File

@ -109,6 +109,10 @@ class DataArguments:
default=None, default=None,
metadata={"help": "Path to save or load the tokenized datasets."}, metadata={"help": "Path to save or load the tokenized datasets."},
) )
dataset_map_batch_size: Optional[int] = field(
default=None,
metadata={"help": "Batch size for dataset mapping."},
)
def __post_init__(self): def __post_init__(self):
def split_arg(arg): def split_arg(arg):