mirror of
https://github.com/meta-llama/llama-stack.git
synced 2025-10-05 04:17:32 +00:00
feat: allow user to register model alias explicitly, tests
# What does this PR do? Context: https://github.com/llamastack/llama-stack/discussions/3483 This PR enables the registering `provider_model_id` as the model identifier without breaking backward compatibility. ## Test Plan todo # What does this PR do? ## Test Plan
This commit is contained in:
parent
ac1414b571
commit
83a229554b
20 changed files with 236 additions and 92 deletions
|
@ -75,7 +75,7 @@ def get_distribution_template() -> DistributionTemplate:
|
|||
|
||||
default_models = [
|
||||
ModelInput(
|
||||
model_id="${env.INFERENCE_MODEL}",
|
||||
provider_model_id="${env.INFERENCE_MODEL}",
|
||||
provider_id="vllm-inference",
|
||||
)
|
||||
]
|
||||
|
@ -85,7 +85,7 @@ def get_distribution_template() -> DistributionTemplate:
|
|||
config=SentenceTransformersInferenceConfig.sample_run_config(),
|
||||
)
|
||||
embedding_model = ModelInput(
|
||||
model_id="all-MiniLM-L6-v2",
|
||||
provider_model_id="all-MiniLM-L6-v2",
|
||||
provider_id=embedding_provider.provider_id,
|
||||
model_type=ModelType.embedding,
|
||||
metadata={
|
||||
|
|
|
@ -88,14 +88,16 @@ inference_store:
|
|||
password: ${env.POSTGRES_PASSWORD:=llamastack}
|
||||
models:
|
||||
- metadata: {}
|
||||
model_id: ${env.INFERENCE_MODEL}
|
||||
provider_model_id: ${env.INFERENCE_MODEL}
|
||||
provider_id: vllm-inference
|
||||
model_type: llm
|
||||
use_provider_model_id_as_id: false
|
||||
- metadata:
|
||||
embedding_dimension: 384
|
||||
model_id: all-MiniLM-L6-v2
|
||||
provider_model_id: all-MiniLM-L6-v2
|
||||
provider_id: sentence-transformers
|
||||
model_type: embedding
|
||||
use_provider_model_id_as_id: false
|
||||
shields:
|
||||
- shield_id: meta-llama/Llama-Guard-3-8B
|
||||
vector_dbs: []
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue