phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-12-22 20:40:00 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix: post_training ci

Browse source 
test_perference_optimize was missing args for DPOAlignmentConfig. Add them in

Signed-off-by: Charlie Doern <cdoern@redhat.com>
This commit is contained in:
Charlie Doern 2025-07-31 10:39:40 -04:00
parent cf73146132
commit 5fc412695c
1 changed files with 5 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

7
tests/integration/post_training/test_post_training.py
View file

@ -194,9 +194,12 @@ class TestPostTraining:
# DPO algorithm configuration # DPO algorithm configuration
algorithm_config = DPOAlignmentConfig( algorithm_config = DPOAlignmentConfig(
beta=0.1, beta=0.1,
loss_type=DPOLossType.sigmoid, loss_type=DPOLossType.sigmoid, # Default loss type
reward_scale=1.0, # Scaling factor for reward signal (neutral scaling)
reward_clip=5.0, # Maximum absolute value for reward clipping (prevents extreme values)
epsilon=1e-8, # Small value for numerical stability
gamma=1.0,
) )
data_config = DataConfig( data_config = DataConfig(
dataset_id=dataset.identifier, dataset_id=dataset.identifier,
batch_size=1, batch_size=1,