phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-12-23 21:23:59 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'main' into enable-vector-stores-files-api-tests

Browse source
This commit is contained in:
Francisco Arceo 2025-07-31 12:32:43 -04:00 • committed by GitHub
commit 474671d462
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
12 changed files with 67 additions and 111 deletions
Download patch file Download diff file Expand all files Collapse all files

8
llama_stack/apis/post_training/post_training.py
View file

@ -193,18 +193,10 @@ class DPOLossType(Enum):
class DPOAlignmentConfig(BaseModel):
"""Configuration for Direct Preference Optimization (DPO) alignment.
:param reward_scale: Scaling factor for the reward signal
:param reward_clip: Maximum absolute value for reward clipping
:param epsilon: Small value added for numerical stability
:param gamma: Discount factor for future rewards
:param beta: Temperature parameter for the DPO loss
:param loss_type: The type of loss function to use for DPO
"""
reward_scale: float
reward_clip: float
epsilon: float
gamma: float
beta: float
loss_type: DPOLossType = DPOLossType.sigmoid