phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-07-05 21:49:13 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

feat: associated models API with post_training

Browse source 
there are likely scenarios where admins of a stack only want to allow clients to fine-tune certain models, register certain models to be fine-tuned. etc
introduce the post_training router and post_training_models as the associated type. A different model type needs to be used for inference vs post_training due to the structure of the router currently.

Signed-off-by: Charlie Doern <cdoern@redhat.com>
This commit is contained in:
Charlie Doern 2025-05-30 12:05:33 -04:00
parent 63a9f08c9e
commit 71caa271ad
11 changed files with 393 additions and 23 deletions
Download patch file Download diff file Expand all files Collapse all files

7
llama_stack/apis/post_training/post_training.py
View file

@ -13,6 +13,7 @@ from pydantic import BaseModel, Field
from llama_stack.apis.common.content_types import URL
from llama_stack.apis.common.job_types import JobStatus
from llama_stack.apis.common.training_types import Checkpoint
from llama_stack.apis.models import Model
from llama_stack.schema_utils import json_schema_type, register_schema, webmethod
@ -168,7 +169,13 @@ class PostTrainingJobArtifactsResponse(BaseModel):
# TODO(ashwin): metrics, evals
class ModelStore(Protocol):
async def get_model(self, identifier: str) -> Model: ...
class PostTraining(Protocol):
model_store: ModelStore | None = None
@webmethod(route="/post-training/supervised-fine-tune", method="POST")
async def supervised_fine_tune(
self,