Skip to content

🧠 两小时从 0 训出你的 64M 大模型。Train your own 64M LLM from scratch in 2 hours.

update trl

gongjy02adb7bc0de3
diff --git a/5-dpo_train.py b/5-dpo_train.py
index 3ae3df7..ef17e2c 100644
--- a/5-dpo_train.py
+++ b/5-dpo_train.py
@@ -3,7 +3,7 @@ import warnings
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
from transformers import TrainingArguments, AutoModelForCausalLM, AutoTokenizer
-from trl import DPOTrainer
+from trl import DPOConfig, DPOTrainer
from datasets import load_dataset
warnings.filterwarnings('ignore')
@@ -23,7 +23,7 @@ def init_model():
if __name__ == '__main__':
model, tokenizer = init_model()
- training_args = TrainingArguments(
+ training_config = DPOConfig(
output_dir="./minimind_dpo",
per_device_train_batch_size=1,
remove_unused_columns=False,
@@ -38,7 +38,7 @@ if __name__ == '__main__':
dpo_trainer = DPOTrainer(
model,
ref_model=None,
- args=training_args,
+ args=training_config,
beta=0.1,
train_dataset=train_dataset['train'],
tokenizer=tokenizer,
diff --git a/requirements.txt b/requirements.txt
index 26c8d29..342675d 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -23,6 +23,6 @@ torch==2.1.2
transformers==4.44.0
jinja2==3.1.2
jsonlines==4.0.0
-trl==0.8.6
+trl==0.11.3
ujson==5.1.0
wandb==0.18.3