phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-08-01 16:24:44 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

temp commit

Browse source
This commit is contained in:
Botao Chen 2024-11-27 14:35:01 -08:00
parent c31a78dfcb
commit 18ae577760
2 changed files with 21 additions and 20 deletions
Download patch file Download diff file Expand all files Collapse all files

20
llama_stack/apis/post_training/post_training.py
View file

@ -183,16 +183,16 @@ class PostTraining(Protocol):
@webmethod(route="/post-training/supervised-fine-tune")
def supervised_fine_tune(
self,
job_uuid: str,
model: str,
dataset_id: str,
validation_dataset_id: str,
algorithm: FinetuningAlgorithm,
algorithm_config: LoraFinetuningConfig,
optimizer_config: OptimizerConfig,
training_config: TrainingConfig,
hyperparam_search_config: Dict[str, Any],
logger_config: Dict[str, Any],
job_uuid: Optional[str],
model: Optional[str],
dataset_id: Optional[str],
validation_dataset_id: Optional[str],
algorithm: Optional[FinetuningAlgorithm],
algorithm_config: Optional[LoraFinetuningConfig],
optimizer_config: Optional[OptimizerConfig],
training_config: Optional[TrainingConfig],
hyperparam_search_config: Optional[Dict[str, Any]],
logger_config: Optional[Dict[str, Any]],
) -> PostTrainingJob: ...
@webmethod(route="/post-training/preference-optimize")

21
llama_stack/providers/inline/post_training/meta_reference/post_training.py
View file

@ -30,6 +30,7 @@ class MetaReferencePostTrainingImpl:
)
OptimizerConfig(
optimizer_type=OptimizerType.adamw,
lr=3e-4,
lr_min=3e-5,
weight_decay=0.1,
@ -50,16 +51,16 @@ class MetaReferencePostTrainingImpl:
def supervised_fine_tune(
self,
job_uuid: str = "1234",
model: str = " meta-llama/Llama-3.2-3B-Instruct",
dataset_id: str = "alpaca",
validation_dataset_id: str = "alpaca",
algorithm: FinetuningAlgorithm = FinetuningAlgorithm.lora,
algorithm_config: LoraFinetuningConfig = LoraFinetuningConfig,
optimizer_config: OptimizerConfig = OptimizerConfig,
training_config: TrainingConfig = TrainingConfig,
hyperparam_search_config: Dict[str, Any] = {},
logger_config: Dict[str, Any] = {},
job_uuid: Optional[str] = "1234",
model: Optional[str] = " meta-llama/Llama-3.2-3B-Instruct",
dataset_id: Optional[str] = "alpaca",
validation_dataset_id: Optional[str] = "alpaca",
algorithm: Optional[FinetuningAlgorithm] = FinetuningAlgorithm.lora,
algorithm_config: Optional[LoraFinetuningConfig] = LoraFinetuningConfig,
optimizer_config: Optional[OptimizerConfig] = OptimizerConfig,
training_config: Optional[TrainingConfig] = TrainingConfig,
hyperparam_search_config: Optional[Dict[str, Any]] = {},
logger_config: Optional[Dict[str, Any]] = {},
) -> PostTrainingJob:
# wrapper request to make it easier to pass around (internal only, not exposed to API)
request = PostTrainingSFTRequest(