mirror of
https://github.com/hiyouga/LLaMA-Factory.git
synced 2025-10-16 00:28:10 +08:00
172 lines
8.4 KiB
Python
172 lines
8.4 KiB
Python
import tiktoken
|
|
from typing import TYPE_CHECKING, Any, Dict, Generator, List, Literal
|
|
from itertools import chain
|
|
|
|
from llmtuner.extras.constants import IGNORE_INDEX
|
|
from llmtuner.extras.template import get_template_and_fix_tokenizer
|
|
|
|
if TYPE_CHECKING:
|
|
from datasets import Dataset
|
|
from transformers import Seq2SeqTrainingArguments
|
|
from transformers.tokenization_utils import PreTrainedTokenizer
|
|
from llmtuner.hparams import DataArguments
|
|
|
|
|
|
def preprocess_dataset(
|
|
dataset: "Dataset",
|
|
tokenizer: "PreTrainedTokenizer",
|
|
data_args: "DataArguments",
|
|
training_args: "Seq2SeqTrainingArguments",
|
|
stage: Literal["pt", "sft", "rm", "ppo"]
|
|
) -> "Dataset":
|
|
column_names = list(dataset.column_names)
|
|
template = get_template_and_fix_tokenizer(data_args.template, tokenizer)
|
|
|
|
def construct_example(examples: Dict[str, List[Any]]) -> Generator[Any, None, None]:
|
|
for i in range(len(examples["prompt"])):
|
|
query, response = examples["prompt"][i], examples["response"][i]
|
|
query = query + "\n" + examples["query"][i] if "query" in examples and examples["query"][i] else query
|
|
history = examples["history"][i] if "history" in examples else None
|
|
system = examples["system"][i] if "system" in examples else None
|
|
yield query, response, history, system
|
|
|
|
def preprocess_pretrain_dataset(examples: Dict[str, List[Any]]) -> Dict[str, Any]:
|
|
# build grouped texts with format `X1 X2 X3 ...` (without <eos>)
|
|
if isinstance(getattr(tokenizer, "tokenizer", None), tiktoken.Encoding): # for tiktoken tokenizer (Qwen)
|
|
kwargs = dict(allowed_special="all")
|
|
else:
|
|
kwargs = dict(add_special_tokens=False)
|
|
|
|
tokenized_examples = tokenizer(examples["prompt"], **kwargs)
|
|
concatenated_examples = {k: list(chain(*tokenized_examples[k])) for k in tokenized_examples.keys()}
|
|
total_length = len(concatenated_examples[list(concatenated_examples.keys())[0]])
|
|
block_size = data_args.max_source_length
|
|
# we drop the small remainder, and if the total_length < block_size, we exclude this batch
|
|
total_length = (total_length // block_size) * block_size
|
|
# split by chunks of max_source_length
|
|
result = {
|
|
k: [t[i: i + block_size] for i in range(0, total_length, block_size)]
|
|
for k, t in concatenated_examples.items()
|
|
}
|
|
return result
|
|
|
|
def preprocess_supervised_dataset(examples: Dict[str, List[Any]]) -> Dict[str, Any]:
|
|
# build inputs with format `<bos> X Y <eos>` and labels with format `<ignore> ... <ignore> Y <eos>`
|
|
# for multiturn examples, we only mask the prompt part in each prompt-response pair.
|
|
model_inputs = {"input_ids": [], "attention_mask": [], "labels": []}
|
|
max_length = data_args.max_source_length + data_args.max_target_length
|
|
|
|
for query, response, history, system in construct_example(examples):
|
|
input_ids, labels = [], []
|
|
|
|
for source_ids, target_ids in template.encode_multiturn(tokenizer, query, response, history, system):
|
|
if len(source_ids) > data_args.max_source_length:
|
|
source_ids = source_ids[:data_args.max_source_length]
|
|
if len(target_ids) > data_args.max_target_length:
|
|
target_ids = target_ids[:data_args.max_target_length]
|
|
|
|
if len(input_ids) + len(source_ids) + len(target_ids) > max_length:
|
|
break
|
|
|
|
input_ids += source_ids + target_ids
|
|
labels += [IGNORE_INDEX] * len(source_ids) + target_ids
|
|
|
|
model_inputs["input_ids"].append(input_ids)
|
|
model_inputs["attention_mask"].append([1] * len(input_ids))
|
|
model_inputs["labels"].append(labels)
|
|
|
|
return model_inputs
|
|
|
|
def preprocess_unsupervised_dataset(examples: Dict[str, List[Any]]) -> Dict[str, Any]:
|
|
# build inputs with format `<bos> X` and labels with format `Y <eos>`
|
|
model_inputs = {"input_ids": [], "attention_mask": [], "labels": []}
|
|
|
|
for query, response, history, system in construct_example(examples):
|
|
source_ids, target_ids = template.encode_oneturn(tokenizer, query, response, history, system)
|
|
|
|
if len(source_ids) > data_args.max_source_length:
|
|
source_ids = source_ids[:data_args.max_source_length]
|
|
if len(target_ids) > data_args.max_target_length:
|
|
target_ids = target_ids[:data_args.max_target_length]
|
|
|
|
model_inputs["input_ids"].append(source_ids)
|
|
model_inputs["attention_mask"].append([1] * len(source_ids))
|
|
model_inputs["labels"].append(target_ids)
|
|
|
|
return model_inputs
|
|
|
|
def preprocess_pairwise_dataset(examples):
|
|
# build input pairs with format `<bos> X`, `Y1 <eos>` and `Y2 <eos>`
|
|
model_inputs = {"prompt_ids": [], "chosen_ids": [], "rejected_ids": []}
|
|
for query, response, history, system in construct_example(examples):
|
|
prompt_ids, chosen_ids = template.encode_oneturn(tokenizer, query, response[0], history, system)
|
|
_, rejected_ids = template.encode_oneturn(tokenizer, query, response[1], history, system)
|
|
|
|
if len(prompt_ids) > data_args.max_source_length:
|
|
prompt_ids = prompt_ids[:data_args.max_source_length]
|
|
if len(chosen_ids) > data_args.max_target_length:
|
|
chosen_ids = chosen_ids[:data_args.max_target_length]
|
|
if len(rejected_ids) > data_args.max_target_length:
|
|
rejected_ids = rejected_ids[:data_args.max_target_length]
|
|
|
|
model_inputs["prompt_ids"].append(prompt_ids)
|
|
model_inputs["chosen_ids"].append(chosen_ids)
|
|
model_inputs["rejected_ids"].append(rejected_ids)
|
|
return model_inputs
|
|
|
|
def print_supervised_dataset_example(example):
|
|
print("input_ids:\n{}".format(example["input_ids"]))
|
|
print("inputs:\n{}".format(tokenizer.decode(example["input_ids"], skip_special_tokens=False)))
|
|
print("label_ids:\n{}".format(example["labels"]))
|
|
print("labels:\n{}".format(tokenizer.decode([
|
|
token_id if token_id != IGNORE_INDEX else tokenizer.pad_token_id for token_id in example["labels"]
|
|
], skip_special_tokens=False)))
|
|
|
|
def print_pairwise_dataset_example(example):
|
|
print("prompt_ids:\n{}".format(example["prompt_ids"]))
|
|
print("prompt:\n{}".format(tokenizer.decode(example["prompt_ids"], skip_special_tokens=False)))
|
|
print("chosen_ids:\n{}".format(example["chosen_ids"]))
|
|
print("chosen:\n{}".format(tokenizer.decode(example["chosen_ids"], skip_special_tokens=False)))
|
|
print("rejected_ids:\n{}".format(example["rejected_ids"]))
|
|
print("rejected:\n{}".format(tokenizer.decode(example["rejected_ids"], skip_special_tokens=False)))
|
|
|
|
def print_unsupervised_dataset_example(example):
|
|
print("input_ids:\n{}".format(example["input_ids"]))
|
|
print("inputs:\n{}".format(tokenizer.decode(example["input_ids"], skip_special_tokens=False)))
|
|
|
|
if stage == "pt":
|
|
dataset = dataset.filter(lambda example: example["prompt"])
|
|
preprocess_function = preprocess_pretrain_dataset
|
|
print_function = print_unsupervised_dataset_example
|
|
elif stage == "sft" and not training_args.predict_with_generate:
|
|
dataset = dataset.filter(lambda example: example["prompt"] and example["response"])
|
|
preprocess_function = preprocess_supervised_dataset
|
|
print_function = print_supervised_dataset_example
|
|
elif stage == "rm":
|
|
dataset = dataset.filter(lambda example: example["prompt"] and len(example["response"]) > 1)
|
|
preprocess_function = preprocess_pairwise_dataset
|
|
print_function = print_pairwise_dataset_example
|
|
else:
|
|
dataset = dataset.filter(lambda example: example["prompt"])
|
|
preprocess_function = preprocess_unsupervised_dataset
|
|
print_function = print_unsupervised_dataset_example
|
|
|
|
with training_args.main_process_first(desc="dataset map pre-processing"):
|
|
kwargs = {}
|
|
if not data_args.streaming:
|
|
kwargs = dict(
|
|
num_proc=data_args.preprocessing_num_workers,
|
|
load_from_cache_file=not data_args.overwrite_cache,
|
|
desc="Running tokenizer on dataset"
|
|
)
|
|
|
|
dataset = dataset.map(
|
|
preprocess_function,
|
|
batched=True,
|
|
remove_columns=column_names,
|
|
**kwargs
|
|
)
|
|
|
|
print_function(next(iter(dataset)))
|
|
return dataset
|