phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-12-28 04:41:58 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'main' into feat/litellm_sambanova_usage

Browse source
This commit is contained in:
Jorge Piedrahita Ortiz 2025-03-24 08:02:40 -05:00 • committed by GitHub
commit 8783dd8162
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
190 changed files with 8649 additions and 3304 deletions
Download patch file Download diff file Expand all files Collapse all files

8
llama_stack/templates/bedrock/bedrock.py
View file

@ -9,7 +9,11 @@ from pathlib import Path
from llama_stack.distribution.datatypes import Provider, ToolGroupInput
from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig
from llama_stack.providers.remote.inference.bedrock.models import MODEL_ENTRIES
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings, get_model_registry
from llama_stack.templates.template import (
DistributionTemplate,
RunConfigSettings,
get_model_registry,
)
def get_distribution_template() -> DistributionTemplate:
@ -76,7 +80,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
},

3
llama_stack/templates/bedrock/doc_template.md
View file

@ -47,9 +47,10 @@ You can do this via Conda (build code) or Docker which has a pre-built image.
This method allows you to get started quickly without having to build the distribution code.
```bash
LLAMA_STACK_PORT=5001
LLAMA_STACK_PORT=8321
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
llamastack/distribution-{{ name }} \
--port $LLAMA_STACK_PORT \

1
llama_stack/templates/bedrock/run.yaml
View file

@ -39,7 +39,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/bedrock/trace_store.db}
eval:

8
llama_stack/templates/cerebras/cerebras.py
View file

@ -14,7 +14,11 @@ from llama_stack.providers.inline.inference.sentence_transformers import (
from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig
from llama_stack.providers.remote.inference.cerebras import CerebrasImplConfig
from llama_stack.providers.remote.inference.cerebras.models import MODEL_ENTRIES
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings, get_model_registry
from llama_stack.templates.template import (
DistributionTemplate,
RunConfigSettings,
get_model_registry,
)
def get_distribution_template() -> DistributionTemplate:
@ -100,7 +104,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"CEREBRAS_API_KEY": (

5
llama_stack/templates/cerebras/doc_template.md
View file

@ -39,9 +39,10 @@ You can do this via Conda (build code) or Docker which has a pre-built image.
This method allows you to get started quickly without having to build the distribution code.
```bash
LLAMA_STACK_PORT=5001
LLAMA_STACK_PORT=8321
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
-v ./run.yaml:/root/my-run.yaml \
llamastack/distribution-{{ name }} \
@ -55,6 +56,6 @@ docker run \
```bash
llama stack build --template cerebras --image-type conda
llama stack run ./run.yaml \
--port 5001 \
--port 8321 \
--env CEREBRAS_API_KEY=$CEREBRAS_API_KEY
```

1
llama_stack/templates/cerebras/run.yaml
View file

@ -79,7 +79,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/cerebras/trace_store.db}
tool_runtime:

12
llama_stack/templates/ci-tests/ci_tests.py
View file

@ -15,10 +15,16 @@ from llama_stack.distribution.datatypes import (
from llama_stack.providers.inline.inference.sentence_transformers import (
SentenceTransformersInferenceConfig,
)
from llama_stack.providers.inline.vector_io.sqlite_vec.config import SQLiteVectorIOConfig
from llama_stack.providers.inline.vector_io.sqlite_vec.config import (
SQLiteVectorIOConfig,
)
from llama_stack.providers.remote.inference.fireworks.config import FireworksImplConfig
from llama_stack.providers.remote.inference.fireworks.models import MODEL_ENTRIES
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings, get_model_registry
from llama_stack.templates.template import (
DistributionTemplate,
RunConfigSettings,
get_model_registry,
)
def get_distribution_template() -> DistributionTemplate:
@ -104,7 +110,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"FIREWORKS_API_KEY": (

1
llama_stack/templates/ci-tests/run.yaml
View file

@ -42,7 +42,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/ci-tests/trace_store.db}
eval:

4
llama_stack/templates/dell/doc_template.md
View file

@ -43,6 +43,7 @@ export CUDA_VISIBLE_DEVICES=0
export LLAMA_STACK_PORT=8321
docker run --rm -it \
--pull always \
--network host \
-v $HOME/.cache/huggingface:/data \
-e HF_TOKEN=$HF_TOKEN \
@ -66,6 +67,7 @@ export SAFETY_MODEL=meta-llama/Llama-Guard-3-1B
export CUDA_VISIBLE_DEVICES=1
docker run --rm -it \
--pull always \
--network host \
-v $HOME/.cache/huggingface:/data \
-e HF_TOKEN=$HF_TOKEN \
@ -108,6 +110,7 @@ This method allows you to get started quickly without having to build the distri
```bash
docker run -it \
--pull always \
--network host \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
-v $HOME/.llama:/root/.llama \
@ -135,6 +138,7 @@ export SAFETY_MODEL=meta-llama/Llama-Guard-3-1B
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
-v $HOME/.llama:/root/.llama \
-v ./llama_stack/templates/tgi/run-with-safety.yaml:/root/my-run.yaml \

1
llama_stack/templates/dell/run-with-safety.yaml
View file

@ -45,7 +45,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/dell/trace_store.db}
eval:

1
llama_stack/templates/dell/run.yaml
View file

@ -41,7 +41,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/dell/trace_store.db}
eval:

40
llama_stack/templates/dev/dev.py
View file

@ -16,22 +16,42 @@ from llama_stack.distribution.datatypes import (
from llama_stack.providers.inline.inference.sentence_transformers import (
SentenceTransformersInferenceConfig,
)
from llama_stack.providers.inline.vector_io.sqlite_vec.config import SQLiteVectorIOConfig
from llama_stack.providers.inline.vector_io.sqlite_vec.config import (
SQLiteVectorIOConfig,
)
from llama_stack.providers.remote.inference.anthropic.config import AnthropicConfig
from llama_stack.providers.remote.inference.anthropic.models import MODEL_ENTRIES as ANTHROPIC_MODEL_ENTRIES
from llama_stack.providers.remote.inference.anthropic.models import (
MODEL_ENTRIES as ANTHROPIC_MODEL_ENTRIES,
)
from llama_stack.providers.remote.inference.fireworks.config import FireworksImplConfig
from llama_stack.providers.remote.inference.fireworks.models import MODEL_ENTRIES as FIREWORKS_MODEL_ENTRIES
from llama_stack.providers.remote.inference.fireworks.models import (
MODEL_ENTRIES as FIREWORKS_MODEL_ENTRIES,
)
from llama_stack.providers.remote.inference.gemini.config import GeminiConfig
from llama_stack.providers.remote.inference.gemini.models import MODEL_ENTRIES as GEMINI_MODEL_ENTRIES
from llama_stack.providers.remote.inference.gemini.models import (
MODEL_ENTRIES as GEMINI_MODEL_ENTRIES,
)
from llama_stack.providers.remote.inference.groq.config import GroqConfig
from llama_stack.providers.remote.inference.groq.models import MODEL_ENTRIES as GROQ_MODEL_ENTRIES
from llama_stack.providers.remote.inference.groq.models import (
MODEL_ENTRIES as GROQ_MODEL_ENTRIES,
)
from llama_stack.providers.remote.inference.openai.config import OpenAIConfig
from llama_stack.providers.remote.inference.openai.models import MODEL_ENTRIES as OPENAI_MODEL_ENTRIES
from llama_stack.providers.remote.inference.openai.models import (
MODEL_ENTRIES as OPENAI_MODEL_ENTRIES,
)
from llama_stack.providers.remote.inference.sambanova.config import SambaNovaImplConfig
from llama_stack.providers.remote.inference.sambanova.models import MODEL_ENTRIES as SAMBANOVA_MODEL_ENTRIES
from llama_stack.providers.remote.inference.sambanova.models import (
MODEL_ENTRIES as SAMBANOVA_MODEL_ENTRIES,
)
from llama_stack.providers.remote.vector_io.chroma.config import ChromaVectorIOConfig
from llama_stack.providers.remote.vector_io.pgvector.config import PGVectorVectorIOConfig
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings, get_model_registry
from llama_stack.providers.remote.vector_io.pgvector.config import (
PGVectorVectorIOConfig,
)
from llama_stack.templates.template import (
DistributionTemplate,
RunConfigSettings,
get_model_registry,
)
def get_inference_providers() -> Tuple[List[Provider], List[ModelInput]]:
@ -175,7 +195,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"FIREWORKS_API_KEY": (

1
llama_stack/templates/dev/run.yaml
View file

@ -76,7 +76,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/dev/trace_store.db}
eval:

18
llama_stack/templates/experimental-post-training/run.yaml
View file

@ -28,7 +28,11 @@ providers:
eval:
- provider_id: meta-reference
provider_type: inline::meta-reference
config: {}
config:
kvstore:
type: sqlite
namespace: null
db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/meta-reference-gpu}/meta_reference_eval.db
scoring:
- provider_id: basic
provider_type: inline::basic
@ -40,7 +44,11 @@ providers:
datasetio:
- provider_id: localfs
provider_type: inline::localfs
config: {}
config:
kvstore:
type: sqlite
namespace: null
db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/experimental-post-training}/localfs_datasetio.db
telemetry:
- provider_id: meta-reference
provider_type: inline::meta-reference
@ -58,7 +66,7 @@ providers:
persistence_store:
type: sqlite
namespace: null
db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/meta-reference-gpu}/agents_store.db
db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/experimental-post-training}/agents_store.db
safety:
- provider_id: llama-guard
provider_type: inline::llama-guard
@ -70,7 +78,7 @@ providers:
kvstore:
type: sqlite
namespace: null
db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/meta-reference-gpu}/faiss_store.db
db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/experimental-post-training}/faiss_store.db
tool_runtime:
- provider_id: brave-search
provider_type: remote::brave-search
@ -82,7 +90,7 @@ providers:
metadata_store:
namespace: null
type: sqlite
db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/meta-reference-gpu}/registry.db
db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/experimental-post-training}/registry.db
models: []
shields: []
vector_dbs: []

3
llama_stack/templates/fireworks/doc_template.md
View file

@ -49,9 +49,10 @@ You can do this via Conda (build code) or Docker which has a pre-built image.
This method allows you to get started quickly without having to build the distribution code.
```bash
LLAMA_STACK_PORT=5001
LLAMA_STACK_PORT=8321
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
llamastack/distribution-{{ name }} \
--port $LLAMA_STACK_PORT \

8
llama_stack/templates/fireworks/fireworks.py
View file

@ -19,7 +19,11 @@ from llama_stack.providers.inline.inference.sentence_transformers import (
from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig
from llama_stack.providers.remote.inference.fireworks.config import FireworksImplConfig
from llama_stack.providers.remote.inference.fireworks.models import MODEL_ENTRIES
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings, get_model_registry
from llama_stack.templates.template import (
DistributionTemplate,
RunConfigSettings,
get_model_registry,
)
def get_distribution_template() -> DistributionTemplate:
@ -158,7 +162,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"FIREWORKS_API_KEY": (

1
llama_stack/templates/fireworks/run-with-safety.yaml
View file

@ -50,7 +50,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/fireworks/trace_store.db}
eval:

1
llama_stack/templates/fireworks/run.yaml
View file

@ -45,7 +45,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/fireworks/trace_store.db}
eval:

3
llama_stack/templates/groq/doc_template.md
View file

@ -49,9 +49,10 @@ You can do this via Conda (build code) or Docker which has a pre-built image.
This method allows you to get started quickly without having to build the distribution code.
```bash
LLAMA_STACK_PORT=5001
LLAMA_STACK_PORT=8321
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
llamastack/distribution-{{ name }} \
--port $LLAMA_STACK_PORT \

14
llama_stack/templates/groq/groq.py
View file

@ -7,17 +7,17 @@
from pathlib import Path
from llama_stack.apis.models.models import ModelType
from llama_stack.distribution.datatypes import (
ModelInput,
Provider,
ToolGroupInput,
)
from llama_stack.distribution.datatypes import ModelInput, Provider, ToolGroupInput
from llama_stack.providers.inline.inference.sentence_transformers import (
SentenceTransformersInferenceConfig,
)
from llama_stack.providers.remote.inference.groq import GroqConfig
from llama_stack.providers.remote.inference.groq.models import MODEL_ENTRIES
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings, get_model_registry
from llama_stack.templates.template import (
DistributionTemplate,
RunConfigSettings,
get_model_registry,
)
def get_distribution_template() -> DistributionTemplate:
@ -97,7 +97,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMASTACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"GROQ_API_KEY": (

1
llama_stack/templates/groq/run.yaml
View file

@ -45,7 +45,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/groq/trace_store.db}
eval:

2
llama_stack/templates/hf-endpoint/hf_endpoint.py
View file

@ -127,7 +127,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"HF_API_TOKEN": (

1
llama_stack/templates/hf-endpoint/run-with-safety.yaml
View file

@ -50,7 +50,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/hf-endpoint/trace_store.db}
eval:

1
llama_stack/templates/hf-endpoint/run.yaml
View file

@ -45,7 +45,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/hf-endpoint/trace_store.db}
eval:

2
llama_stack/templates/hf-serverless/hf_serverless.py
View file

@ -128,7 +128,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"HF_API_TOKEN": (

1
llama_stack/templates/hf-serverless/run-with-safety.yaml
View file

@ -50,7 +50,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/hf-serverless/trace_store.db}
eval:

1
llama_stack/templates/hf-serverless/run.yaml
View file

@ -45,7 +45,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/hf-serverless/trace_store.db}
eval:

8
llama_stack/templates/meta-reference-gpu/doc_template.md
View file

@ -65,9 +65,10 @@ You can do this via Conda (build code) or Docker which has a pre-built image.
This method allows you to get started quickly without having to build the distribution code.
```bash
LLAMA_STACK_PORT=5001
LLAMA_STACK_PORT=8321
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
-v ~/.llama:/root/.llama \
llamastack/distribution-{{ name }} \
@ -80,6 +81,7 @@ If you are using Llama Stack Safety / Shield APIs, use:
```bash
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
-v ~/.llama:/root/.llama \
llamastack/distribution-{{ name }} \
@ -95,7 +97,7 @@ Make sure you have done `uv pip install llama-stack` and have the Llama Stack CL
```bash
llama stack build --template {{ name }} --image-type conda
llama stack run distributions/{{ name }}/run.yaml \
--port 5001 \
--port 8321 \
--env INFERENCE_MODEL=meta-llama/Llama-3.2-3B-Instruct
```
@ -103,7 +105,7 @@ If you are using Llama Stack Safety / Shield APIs, use:
```bash
llama stack run distributions/{{ name }}/run-with-safety.yaml \
--port 5001 \
--port 8321 \
--env INFERENCE_MODEL=meta-llama/Llama-3.2-3B-Instruct \
--env SAFETY_MODEL=meta-llama/Llama-Guard-3-1B
```

2
llama_stack/templates/meta-reference-gpu/meta_reference.py
View file

@ -134,7 +134,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"INFERENCE_MODEL": (

1
llama_stack/templates/meta-reference-gpu/run-with-safety.yaml
View file

@ -52,7 +52,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/meta-reference-gpu/trace_store.db}
eval:

1
llama_stack/templates/meta-reference-gpu/run.yaml
View file

@ -46,7 +46,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/meta-reference-gpu/trace_store.db}
eval:

4
llama_stack/templates/meta-reference-quantized-gpu/doc_template.md
View file

@ -67,9 +67,10 @@ You can do this via Conda (build code) or Docker which has a pre-built image.
This method allows you to get started quickly without having to build the distribution code.
```bash
LLAMA_STACK_PORT=5001
LLAMA_STACK_PORT=8321
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
-v ~/.llama:/root/.llama \
llamastack/distribution-{{ name }} \
@ -82,6 +83,7 @@ If you are using Llama Stack Safety / Shield APIs, use:
```bash
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
-v ~/.llama:/root/.llama \
llamastack/distribution-{{ name }} \

2
llama_stack/templates/meta-reference-quantized-gpu/meta_reference.py
View file

@ -100,7 +100,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"INFERENCE_MODEL": (

1
llama_stack/templates/meta-reference-quantized-gpu/run.yaml
View file

@ -48,7 +48,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/meta-reference-quantized-gpu/trace_store.db}
eval:

5
llama_stack/templates/nvidia/doc_template.md
View file

@ -39,9 +39,10 @@ You can do this via Conda (build code) or Docker which has a pre-built image.
This method allows you to get started quickly without having to build the distribution code.
```bash
LLAMA_STACK_PORT=5001
LLAMA_STACK_PORT=8321
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
-v ./run.yaml:/root/my-run.yaml \
llamastack/distribution-{{ name }} \
@ -55,7 +56,7 @@ docker run \
```bash
llama stack build --template nvidia --image-type conda
llama stack run ./run.yaml \
--port 5001 \
--port 8321 \
--env NVIDIA_API_KEY=$NVIDIA_API_KEY
--env INFERENCE_MODEL=$INFERENCE_MODEL
```

1
llama_stack/templates/nvidia/run-with-safety.yaml
View file

@ -48,7 +48,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/nvidia/trace_store.db}
eval:

1
llama_stack/templates/nvidia/run.yaml
View file

@ -43,7 +43,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/nvidia/trace_store.db}
eval:

6
llama_stack/templates/ollama/doc_template.md
View file

@ -60,9 +60,10 @@ Now you are ready to run Llama Stack with Ollama as the inference provider. You
This method allows you to get started quickly without having to build the distribution code.
```bash
export LLAMA_STACK_PORT=5001
export LLAMA_STACK_PORT=8321
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
-v ~/.llama:/root/.llama \
llamastack/distribution-{{ name }} \
@ -80,6 +81,7 @@ cd /path/to/llama-stack
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
-v ~/.llama:/root/.llama \
-v ./llama_stack/templates/ollama/run-with-safety.yaml:/root/my-run.yaml \
@ -96,7 +98,7 @@ docker run \
Make sure you have done `uv pip install llama-stack` and have the Llama Stack CLI available.
```bash
export LLAMA_STACK_PORT=5001
export LLAMA_STACK_PORT=8321
llama stack build --template {{ name }} --image-type conda
llama stack run ./run.yaml \

2
llama_stack/templates/ollama/ollama.py
View file

@ -138,7 +138,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"OLLAMA_URL": (

1
llama_stack/templates/ollama/run-with-safety.yaml
View file

@ -43,7 +43,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/ollama/trace_store.db}
eval:

1
llama_stack/templates/ollama/run.yaml
View file

@ -41,7 +41,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/ollama/trace_store.db}
eval:

26
llama_stack/templates/open-benchmark/open_benchmark.py
View file

@ -203,6 +203,20 @@ def get_distribution_template() -> DistributionTemplate:
uri="huggingface://datasets/llamastack/bfcl_v3?split=train",
),
),
DatasetInput(
dataset_id="ifeval",
purpose=DatasetPurpose.eval_messages_answer,
source=URIDataSource(
uri="huggingface://datasets/llamastack/IfEval?split=train",
),
),
DatasetInput(
dataset_id="docvqa",
purpose=DatasetPurpose.eval_messages_answer,
source=URIDataSource(
uri="huggingface://datasets/llamastack/docvqa?split=val",
),
),
]
default_benchmarks = [
@ -231,6 +245,16 @@ def get_distribution_template() -> DistributionTemplate:
dataset_id="bfcl",
scoring_functions=["basic::bfcl"],
),
BenchmarkInput(
benchmark_id="meta-reference-ifeval",
dataset_id="ifeval",
scoring_functions=["basic::ifeval"],
),
BenchmarkInput(
benchmark_id="meta-reference-docvqa",
dataset_id="docvqa",
scoring_functions=["basic::docvqa"],
),
]
return DistributionTemplate(
name=name,
@ -255,7 +279,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"TOGETHER_API_KEY": (

23
llama_stack/templates/open-benchmark/run.yaml
View file

@ -68,7 +68,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/open-benchmark/trace_store.db}
eval:
@ -188,6 +187,18 @@ datasets:
uri: huggingface://datasets/llamastack/bfcl_v3?split=train
metadata: {}
dataset_id: bfcl
- purpose: eval/messages-answer
source:
type: uri
uri: huggingface://datasets/llamastack/IfEval?split=train
metadata: {}
dataset_id: ifeval
- purpose: eval/messages-answer
source:
type: uri
uri: huggingface://datasets/llamastack/docvqa?split=val
metadata: {}
dataset_id: docvqa
scoring_fns: []
benchmarks:
- dataset_id: simpleqa
@ -215,6 +226,16 @@ benchmarks:
- basic::bfcl
metadata: {}
benchmark_id: meta-reference-bfcl
- dataset_id: ifeval
scoring_functions:
- basic::ifeval
metadata: {}
benchmark_id: meta-reference-ifeval
- dataset_id: docvqa
scoring_functions:
- basic::docvqa
metadata: {}
benchmark_id: meta-reference-docvqa
tool_groups:
- toolgroup_id: builtin::websearch
provider_id: tavily-search

7
llama_stack/templates/passthrough/passthrough.py
View file

@ -21,10 +21,7 @@ from llama_stack.providers.remote.inference.passthrough.config import (
PassthroughImplConfig,
)
from llama_stack.providers.utils.inference.model_registry import ProviderModelEntry
from llama_stack.templates.template import (
DistributionTemplate,
RunConfigSettings,
)
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings
def get_distribution_template() -> DistributionTemplate:
@ -186,7 +183,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"PASSTHROUGH_API_KEY": (

1
llama_stack/templates/passthrough/run-with-safety.yaml
View file

@ -50,7 +50,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/passthrough/trace_store.db}
eval:

1
llama_stack/templates/passthrough/run.yaml
View file

@ -45,7 +45,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/passthrough/trace_store.db}
eval:

10
llama_stack/templates/remote-vllm/doc_template.md
View file

@ -36,6 +36,7 @@ export INFERENCE_MODEL=meta-llama/Llama-3.2-3B-Instruct
export CUDA_VISIBLE_DEVICES=0
docker run \
--pull always \
--runtime nvidia \
--gpus $CUDA_VISIBLE_DEVICES \
-v ~/.cache/huggingface:/root/.cache/huggingface \
@ -48,6 +49,8 @@ docker run \
--port $INFERENCE_PORT
```
Note that you'll also need to set `--enable-auto-tool-choice` and `--tool-call-parser` to [enable tool calling in vLLM](https://docs.vllm.ai/en/latest/features/tool_calling.html).
If you are using Llama Stack Safety / Shield APIs, then you will need to also run another instance of a vLLM with a corresponding safety model like `meta-llama/Llama-Guard-3-1B` using a script like:
```bash
@ -56,6 +59,7 @@ export SAFETY_MODEL=meta-llama/Llama-Guard-3-1B
export CUDA_VISIBLE_DEVICES=1
docker run \
--pull always \
--runtime nvidia \
--gpus $CUDA_VISIBLE_DEVICES \
-v ~/.cache/huggingface:/root/.cache/huggingface \
@ -79,10 +83,11 @@ This method allows you to get started quickly without having to build the distri
```bash
export INFERENCE_PORT=8000
export INFERENCE_MODEL=meta-llama/Llama-3.2-3B-Instruct
export LLAMA_STACK_PORT=5001
export LLAMA_STACK_PORT=8321
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
-v ./run.yaml:/root/my-run.yaml \
llamastack/distribution-{{ name }} \
@ -104,6 +109,7 @@ cd /path/to/llama-stack
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
-v ~/.llama:/root/.llama \
-v ./llama_stack/templates/remote-vllm/run-with-safety.yaml:/root/my-run.yaml \
@ -124,7 +130,7 @@ Make sure you have done `uv pip install llama-stack` and have the Llama Stack CL
```bash
export INFERENCE_PORT=8000
export INFERENCE_MODEL=meta-llama/Llama-3.2-3B-Instruct
export LLAMA_STACK_PORT=5001
export LLAMA_STACK_PORT=8321
cd distributions/remote-vllm
llama stack build --template remote-vllm --image-type conda

3
llama_stack/templates/remote-vllm/run-with-safety.yaml
View file

@ -15,7 +15,7 @@ providers:
- provider_id: vllm-inference
provider_type: remote::vllm
config:
url: ${env.VLLM_URL}
url: ${env.VLLM_URL:http://localhost:8000/v1}
max_tokens: ${env.VLLM_MAX_TOKENS:4096}
api_token: ${env.VLLM_API_TOKEN:fake}
tls_verify: ${env.VLLM_TLS_VERIFY:true}
@ -88,7 +88,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/remote-vllm/trace_store.db}
tool_runtime:

3
llama_stack/templates/remote-vllm/run.yaml
View file

@ -15,7 +15,7 @@ providers:
- provider_id: vllm-inference
provider_type: remote::vllm
config:
url: ${env.VLLM_URL}
url: ${env.VLLM_URL:http://localhost:8000/v1}
max_tokens: ${env.VLLM_MAX_TOKENS:4096}
api_token: ${env.VLLM_API_TOKEN:fake}
tls_verify: ${env.VLLM_TLS_VERIFY:true}
@ -81,7 +81,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/remote-vllm/trace_store.db}
tool_runtime:

4
llama_stack/templates/remote-vllm/vllm.py
View file

@ -45,7 +45,7 @@ def get_distribution_template() -> DistributionTemplate:
provider_id="vllm-inference",
provider_type="remote::vllm",
config=VLLMInferenceAdapterConfig.sample_run_config(
url="${env.VLLM_URL}",
url="${env.VLLM_URL:http://localhost:8000/v1}",
),
)
embedding_provider = Provider(
@ -135,7 +135,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"INFERENCE_MODEL": (

3
llama_stack/templates/sambanova/doc_template.md
View file

@ -49,9 +49,10 @@ You can do this via Conda (build code) or Docker which has a pre-built image.
This method allows you to get started quickly without having to build the distribution code.
```bash
LLAMA_STACK_PORT=5001
LLAMA_STACK_PORT=8321
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
llamastack/distribution-{{ name }} \
--port $LLAMA_STACK_PORT \

1
llama_stack/templates/sambanova/run.yaml
View file

@ -54,7 +54,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/sambanova/trace_store.db}
tool_runtime:

12
llama_stack/templates/sambanova/sambanova.py
View file

@ -20,8 +20,14 @@ from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOCon
from llama_stack.providers.remote.inference.sambanova import SambaNovaImplConfig
from llama_stack.providers.remote.inference.sambanova.models import MODEL_ENTRIES
from llama_stack.providers.remote.vector_io.chroma.config import ChromaVectorIOConfig
from llama_stack.providers.remote.vector_io.pgvector.config import PGVectorVectorIOConfig
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings, get_model_registry
from llama_stack.providers.remote.vector_io.pgvector.config import (
PGVectorVectorIOConfig,
)
from llama_stack.templates.template import (
DistributionTemplate,
RunConfigSettings,
get_model_registry,
)
def get_distribution_template() -> DistributionTemplate:
@ -127,7 +133,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMASTACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"SAMBANOVA_API_KEY": (

6
llama_stack/templates/tgi/doc_template.md
View file

@ -38,6 +38,7 @@ export INFERENCE_MODEL=meta-llama/Llama-3.2-3B-Instruct
export CUDA_VISIBLE_DEVICES=0
docker run --rm -it \
--pull always \
-v $HOME/.cache/huggingface:/data \
-p $INFERENCE_PORT:$INFERENCE_PORT \
--gpus $CUDA_VISIBLE_DEVICES \
@ -58,6 +59,7 @@ export SAFETY_MODEL=meta-llama/Llama-Guard-3-1B
export CUDA_VISIBLE_DEVICES=1
docker run --rm -it \
--pull always \
-v $HOME/.cache/huggingface:/data \
-p $SAFETY_PORT:$SAFETY_PORT \
--gpus $CUDA_VISIBLE_DEVICES \
@ -78,9 +80,10 @@ Now you are ready to run Llama Stack with TGI as the inference provider. You can
This method allows you to get started quickly without having to build the distribution code.
```bash
LLAMA_STACK_PORT=5001
LLAMA_STACK_PORT=8321
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
llamastack/distribution-{{ name }} \
--port $LLAMA_STACK_PORT \
@ -97,6 +100,7 @@ cd /path/to/llama-stack
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
-v ~/.llama:/root/.llama \
-v ./llama_stack/templates/tgi/run-with-safety.yaml:/root/my-run.yaml \

1
llama_stack/templates/tgi/run-with-safety.yaml
View file

@ -45,7 +45,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/tgi/trace_store.db}
eval:

1
llama_stack/templates/tgi/run.yaml
View file

@ -44,7 +44,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/tgi/trace_store.db}
eval:

2
llama_stack/templates/tgi/tgi.py
View file

@ -129,7 +129,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"INFERENCE_MODEL": (

3
llama_stack/templates/together/doc_template.md
View file

@ -49,9 +49,10 @@ You can do this via Conda (build code) or Docker which has a pre-built image.
This method allows you to get started quickly without having to build the distribution code.
```bash
LLAMA_STACK_PORT=5001
LLAMA_STACK_PORT=8321
docker run \
-it \
--pull always \
-p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
llamastack/distribution-{{ name }} \
--port $LLAMA_STACK_PORT \

1
llama_stack/templates/together/run-with-safety.yaml
View file

@ -50,7 +50,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/together/trace_store.db}
eval:

1
llama_stack/templates/together/run.yaml
View file

@ -45,7 +45,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/together/trace_store.db}
eval:

8
llama_stack/templates/together/together.py
View file

@ -19,7 +19,11 @@ from llama_stack.providers.inline.inference.sentence_transformers import (
from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig
from llama_stack.providers.remote.inference.together import TogetherImplConfig
from llama_stack.providers.remote.inference.together.models import MODEL_ENTRIES
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings, get_model_registry
from llama_stack.templates.template import (
DistributionTemplate,
RunConfigSettings,
get_model_registry,
)
def get_distribution_template() -> DistributionTemplate:
@ -154,7 +158,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"TOGETHER_API_KEY": (

1
llama_stack/templates/vllm-gpu/run.yaml
View file

@ -49,7 +49,6 @@ providers:
- provider_id: meta-reference
provider_type: inline::meta-reference
config:
service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
sinks: ${env.TELEMETRY_SINKS:console,sqlite}
sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/vllm-gpu/trace_store.db}
eval:

2
llama_stack/templates/vllm-gpu/vllm.py
View file

@ -100,7 +100,7 @@ def get_distribution_template() -> DistributionTemplate:
},
run_config_env_vars={
"LLAMA_STACK_PORT": (
"5001",
"8321",
"Port for the Llama Stack distribution server",
),
"INFERENCE_MODEL": (