feat(distro): fork off a starter-gpu distribution

llama_stack/distributions/starter/build.yaml

@@ -1,6 +1,7 @@
 version: 2
 distribution_spec:
-  description: Quick start template for running Llama Stack with several popular providers
+  description: Quick start template for running Llama Stack with several popular providers.
+    This distribution is intended for CPU-only environments.
   providers:
     inference:
     - provider_type: remote::cerebras
@@ -34,7 +35,7 @@ distribution_spec:
     telemetry:
     - provider_type: inline::meta-reference
     post_training:
-    - provider_type: inline::huggingface
+    - provider_type: inline::huggingface-cpu
     eval:
     - provider_type: inline::meta-reference
     datasetio:

llama_stack/distributions/starter/run.yaml

@@ -156,8 +156,8 @@ providers:
       sqlite_db_path: ${env.SQLITE_STORE_DIR:=~/.llama/distributions/starter}/trace_store.db
       otel_exporter_otlp_endpoint: ${env.OTEL_EXPORTER_OTLP_ENDPOINT:=}
   post_training:
-  - provider_id: huggingface
-    provider_type: inline::huggingface
+  - provider_id: huggingface-cpu
+    provider_type: inline::huggingface-cpu
     config:
       checkpoint_format: huggingface
       distributed_backend: null

llama_stack/distributions/starter/starter.py

@@ -120,7 +120,7 @@ def get_distribution_template() -> DistributionTemplate:
         ],
         "agents": [BuildProvider(provider_type="inline::meta-reference")],
         "telemetry": [BuildProvider(provider_type="inline::meta-reference")],
-        "post_training": [BuildProvider(provider_type="inline::huggingface")],
+        "post_training": [BuildProvider(provider_type="inline::huggingface-cpu")],
         "eval": [BuildProvider(provider_type="inline::meta-reference")],
         "datasetio": [
             BuildProvider(provider_type="remote::huggingface"),
@@ -178,7 +178,7 @@ def get_distribution_template() -> DistributionTemplate:
     return DistributionTemplate(
         name=name,
         distro_type="self_hosted",
-        description="Quick start template for running Llama Stack with several popular providers",
+        description="Quick start template for running Llama Stack with several popular providers. This distribution is intended for CPU-only environments.",
         container_image=None,
         template_path=None,
         providers=providers,