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

support DPO training (2305.18290)

Browse Source 
Former-commit-id: 6d98de148e4af63a7028dfaeb6cf86eb56a4488f
This commit is contained in:
hiyouga
2023-08-11 03:02:53 +08:00
parent 72dfd74005
commit ca719a8697
33 changed files with 513 additions and 192 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
src/llmtuner/tuner/pt/workflow.py
View File

@@ -2,10 +2,9 @@
import math
from typing import TYPE_CHECKING, Optional, List
from transformers import DataCollatorForSeq2Seq
from transformers import DataCollatorForLanguageModeling
from llmtuner.dsets import get_dataset, preprocess_dataset, split_dataset
from llmtuner.extras.constants import IGNORE_INDEX
from llmtuner.extras.ploting import plot_loss
from llmtuner.tuner.core import load_model_and_tokenizer
from llmtuner.tuner.core.trainer import PeftTrainer
@@ -25,10 +24,7 @@ def run_pt(
dataset = get_dataset(model_args, data_args)
model, tokenizer = load_model_and_tokenizer(model_args, finetuning_args, training_args.do_train, stage="pt")
dataset = preprocess_dataset(dataset, tokenizer, data_args, training_args, stage="pt")
data_collator = DataCollatorForSeq2Seq(
tokenizer=tokenizer,
label_pad_token_id=IGNORE_INDEX if data_args.ignore_pad_token_for_loss else tokenizer.pad_token_id
)
data_collator = DataCollatorForLanguageModeling(tokenizer=tokenizer, mlm=False)
# Initialize our Trainer
trainer = PeftTrainer(