ros/LLaMA-Factory
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

update get template

Browse Source 
Former-commit-id: 21ea0d0786f91c0bce79630963e66b815a6792a0
This commit is contained in:
hiyouga
2024-09-04 22:36:20 +08:00
parent 5d85be31ca
commit af178cbcd1
17 changed files with 57 additions and 56 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
scripts/cal_lr.py
View File

@@ -25,7 +25,7 @@ from torch.utils.data import DataLoader
from tqdm import tqdm
from transformers import DataCollatorForLanguageModeling, DataCollatorForSeq2Seq
from llamafactory.data import get_dataset
from llamafactory.data import get_dataset, get_template_and_fix_tokenizer
from llamafactory.extras.constants import IGNORE_INDEX
from llamafactory.hparams import get_train_args
from llamafactory.model import load_tokenizer
@@ -66,7 +66,8 @@ def calculate_lr(
)
tokenizer_module = load_tokenizer(model_args)
tokenizer = tokenizer_module["tokenizer"]
trainset = get_dataset(model_args, data_args, training_args, stage, **tokenizer_module)["train_dataset"]
template = get_template_and_fix_tokenizer(tokenizer, data_args)
trainset = get_dataset(template, model_args, data_args, training_args, stage, **tokenizer_module)["train_dataset"]
if stage == "pt":
data_collator = DataCollatorForLanguageModeling(tokenizer=tokenizer, mlm=False)
elif stage == "sft":

5
scripts/cal_ppl.py
View File

@@ -23,7 +23,7 @@ from torch.utils.data import DataLoader
from tqdm import tqdm
from transformers import DataCollatorForLanguageModeling, DataCollatorForSeq2Seq
from llamafactory.data import get_dataset
from llamafactory.data import get_dataset, get_template_and_fix_tokenizer
from llamafactory.extras.constants import IGNORE_INDEX
from llamafactory.hparams import get_train_args
from llamafactory.model import load_model, load_tokenizer
@@ -88,7 +88,8 @@ def cal_ppl(
)
tokenizer_module = load_tokenizer(model_args)
tokenizer = tokenizer_module["tokenizer"]
trainset = get_dataset(model_args, data_args, training_args, stage, **tokenizer_module)["train_dataset"]
template = get_template_and_fix_tokenizer(tokenizer, data_args)
trainset = get_dataset(template, model_args, data_args, training_args, stage, **tokenizer_module)["train_dataset"]
model = load_model(tokenizer, model_args, finetuning_args, is_trainable=False)
if stage == "pt":
data_collator = DataCollatorForLanguageModeling(tokenizer=tokenizer, mlm=False)

5
scripts/length_cdf.py
View File

@@ -18,7 +18,7 @@ from collections import defaultdict
import fire
from tqdm import tqdm
from llamafactory.data import get_dataset
from llamafactory.data import get_dataset, get_template_and_fix_tokenizer
from llamafactory.hparams import get_train_args
from llamafactory.model import load_tokenizer
@@ -48,7 +48,8 @@ def length_cdf(
)
)
tokenizer_module = load_tokenizer(model_args)
trainset = get_dataset(model_args, data_args, training_args, stage="sft", **tokenizer_module)["train_dataset"]
template = get_template_and_fix_tokenizer(tokenizer_module["tokenizer"], data_args)
trainset = get_dataset(template, model_args, data_args, training_args, "sft", **tokenizer_module)["train_dataset"]
total_num = len(trainset)
length_dict = defaultdict(int)
for sample in tqdm(trainset["input_ids"]):