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

7
src/llamafactory/train/sft/workflow.py
View File

@@ -17,7 +17,7 @@
from typing import TYPE_CHECKING, List, Optional
from ...data import SFTDataCollatorWith4DAttentionMask, get_dataset
from ...data import SFTDataCollatorWith4DAttentionMask, get_dataset, get_template_and_fix_tokenizer
from ...extras.constants import IGNORE_INDEX
from ...extras.misc import get_logits_processor
from ...extras.ploting import plot_loss
@@ -43,7 +43,8 @@ def run_sft(
):
tokenizer_module = load_tokenizer(model_args)
tokenizer = tokenizer_module["tokenizer"]
dataset_module, template = get_dataset(model_args, data_args, training_args, stage="sft", **tokenizer_module)
template = get_template_and_fix_tokenizer(tokenizer, data_args)
dataset_module = get_dataset(template, model_args, data_args, training_args, stage="sft", **tokenizer_module)
model = load_model(tokenizer, model_args, finetuning_args, training_args.do_train)
if getattr(model, "is_quantized", False) and not training_args.do_train:
@@ -62,7 +63,7 @@ def run_sft(
# Override the decoding parameters of Seq2SeqTrainer
training_args.generation_max_length = training_args.generation_max_length or data_args.cutoff_len
training_args.generation_num_beams = data_args.eval_num_beams or training_args.generation_num_beams
training_args.remove_unused_columns = False # important for multimodal and pairwise dataset
training_args.remove_unused_columns = False # important for multimodal dataset
# Metric utils
metric_module = {}