423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-12-16 20:00:36 +08:00
Code Issues Packages Projects Releases Wiki Activity

support galore

Browse Source
This commit is contained in:
hiyouga
2024-03-07 22:41:36 +08:00
parent 725f7cd70f
commit 28f7862188
12 changed files with 115 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
src/llmtuner/train/ppo/workflow.py
View File

@@ -13,8 +13,8 @@ from ...extras.callbacks import FixValueHeadModelCallback
from ...extras.misc import fix_valuehead_checkpoint
from ...extras.ploting import plot_loss
from ...model import load_model, load_tokenizer
from ...train.ppo.trainer import CustomPPOTrainer
from ...train.utils import create_ref_model, create_reward_model
from ..utils import create_custom_optimzer, create_ref_model, create_reward_model
from .trainer import CustomPPOTrainer
if TYPE_CHECKING:
@@ -64,7 +64,10 @@ def run_ppo(
)
# Create optimizer and scheduler
optimizer = AdamW(filter(lambda p: p.requires_grad, model.parameters()), lr=training_args.learning_rate)
optimizer = create_custom_optimzer(model, training_args, finetuning_args)
if optimizer is None:
optimizer = AdamW(filter(lambda p: p.requires_grad, model.parameters()), lr=training_args.learning_rate)
if training_args.max_steps > 0:
num_training_steps = training_args.max_steps
else: