llama-stack-0.2.23 and adjust to new APIs

pyproject.toml

@@ -21,8 +21,9 @@ dependencies = [
     "datasets>=3.6.0",
     "fastapi>=0.115.12",
     "httpx>=0.28.1",
-    "llama-stack==0.2.14",
+    "llama-stack==0.2.23",
     "mcp>=1.9.2",
+    "milvus-lite>=2.5.1",
     "numpy>=2.2.6",
     "openai>=1.82.0",
     "opentelemetry-exporter-otlp-proto-http>=1.33.1",
@@ -31,7 +32,7 @@ dependencies = [
     "psutil>=7.0.0",
     "pydantic>=2.11.5",
     "pymilvus>=2.5.10",
-    "ramalama==0.10.1",
+    "ramalama==0.17.1",
     "requests>=2.32.3",
     "sentence-transformers>=3.0.0",
     "six>=1.17.0",
@@ -40,6 +41,7 @@ dependencies = [
     "trl>=0.18.1",
     "urllib3>=2.4.0",
     "uvicorn>=0.34.2",
+    "setuptools<70",
 ]
 
 [dependency-groups]

requirements.txt

@@ -6,17 +6,19 @@ accelerate==1.7.0
     #   trl
 aiohappyeyeballs==2.6.1
     # via aiohttp
-aiohttp==3.12.7
+aiohttp==3.13.3
     # via
     #   fsspec
     #   llama-stack
     #   ramalama-stack
-aiosignal==1.3.2
+aiosignal==1.4.0
     # via aiohttp
 aiosqlite==0.21.0
     # via
     #   llama-stack
     #   ramalama-stack
+annotated-doc==0.0.4
+    # via fastapi
 annotated-types==0.7.0
     # via pydantic
 anyio==4.9.0
@@ -47,6 +49,8 @@ certifi==2025.4.26
     #   httpcore
     #   httpx
     #   requests
+cffi==2.0.0 ; platform_python_implementation != 'PyPy'
+    # via cryptography
 chardet==5.2.0
     # via ramalama-stack
 charset-normalizer==3.4.2
@@ -61,6 +65,10 @@ colorama==0.4.6 ; sys_platform == 'win32'
     # via
     #   click
     #   tqdm
+cryptography==45.0.7
+    # via
+    #   pyjwt
+    #   python-jose
 datasets==3.6.0
     # via
     #   ramalama-stack
@@ -80,7 +88,7 @@ distro==1.9.0
     #   openai
 ecdsa==0.19.1
     # via python-jose
-fastapi==0.115.12
+fastapi==0.129.0
     # via
     #   llama-stack
     #   ramalama-stack
@@ -115,7 +123,7 @@ h11==0.16.0
     #   httpcore
     #   llama-stack
     #   uvicorn
-hf-xet==1.1.2 ; platform_machine == 'aarch64' or platform_machine == 'amd64' or platform_machine == 'arm64' or platform_machine == 'x86_64'
+hf-xet==1.2.0 ; platform_machine == 'aarch64' or platform_machine == 'amd64' or platform_machine == 'arm64' or platform_machine == 'x86_64'
     # via huggingface-hub
 httpcore==1.0.9
     # via httpx
@@ -128,7 +136,7 @@ httpx==0.28.1
     #   ramalama-stack
 httpx-sse==0.4.0
     # via mcp
-huggingface-hub==0.32.4
+huggingface-hub==0.36.2
     # via
     #   accelerate
     #   datasets
@@ -148,6 +156,7 @@ importlib-metadata==8.6.1
 jinja2==3.1.6
     # via
     #   llama-stack
+    #   ramalama
     #   torch
 jiter==0.10.0
     # via openai
@@ -157,24 +166,28 @@ jsonschema==4.24.0
     # via
     #   autoevals
     #   llama-stack
+    #   mcp
+    #   ramalama
 jsonschema-specifications==2025.4.1
     # via jsonschema
-llama-stack==0.2.14
+llama-stack==0.2.23
     # via ramalama-stack
-llama-stack-client==0.2.14
+llama-stack-client==0.2.23
     # via llama-stack
 lxml==5.4.0
     # via blobfile
 markdown-it-py==3.0.0
     # via rich
 markupsafe==3.0.2
     # via jinja2
-mcp==1.9.2
+mcp==1.26.0
     # via ramalama-stack
 mdurl==0.1.2
     # via markdown-it-py
-milvus-lite==2.4.12 ; sys_platform != 'win32'
-    # via pymilvus
+milvus-lite==2.5.1
+    # via
+    #   pymilvus
+    #   ramalama-stack
 mpmath==1.3.0
     # via sympy
 multidict==6.4.4
@@ -185,7 +198,7 @@ multiprocess==0.70.16
     # via datasets
 networkx==3.5
     # via torch
-numpy==2.2.6
+numpy==2.4.2
     # via
     #   accelerate
     #   datasets
@@ -232,7 +245,7 @@ nvidia-nvjitlink-cu12==12.6.85 ; platform_machine == 'x86_64' and sys_platform =
     #   torch
 nvidia-nvtx-cu12==12.6.77 ; platform_machine == 'x86_64' and sys_platform == 'linux'
     # via torch
-openai==1.84.0
+openai==2.21.0
     # via
     #   llama-stack
     #   ramalama-stack
@@ -296,17 +309,19 @@ psutil==7.0.0
     #   accelerate
     #   peft
     #   ramalama-stack
-pyaml==25.5.0
+pyaml==26.2.1
     # via llama-stack-client
 pyarrow==20.0.0
     # via datasets
-pyasn1==0.6.1
+pyasn1==0.6.2
     # via
     #   python-jose
     #   rsa
+pycparser==3.0 ; implementation_name != 'PyPy' and platform_python_implementation != 'PyPy'
+    # via cffi
 pycryptodomex==3.23.0
     # via blobfile
-pydantic==2.11.5
+pydantic==2.12.5
     # via
     #   fastapi
     #   llama-stack
@@ -315,12 +330,14 @@ pydantic==2.11.5
     #   openai
     #   pydantic-settings
     #   ramalama-stack
-pydantic-core==2.33.2
+pydantic-core==2.41.5
     # via pydantic
 pydantic-settings==2.9.1
     # via mcp
 pygments==2.19.1
     # via rich
+pyjwt==2.11.0
+    # via mcp
 pymilvus==2.5.10
     # via ramalama-stack
 python-dateutil==2.9.0.post0
@@ -338,6 +355,8 @@ python-multipart==0.0.20
     #   mcp
 pytz==2025.2
     # via pandas
+pywin32==311 ; sys_platform == 'win32'
+    # via mcp
 pyyaml==6.0.2
     # via
     #   accelerate
@@ -346,8 +365,9 @@ pyyaml==6.0.2
     #   huggingface-hub
     #   peft
     #   pyaml
+    #   ramalama
     #   transformers
-ramalama==0.10.1
+ramalama==0.17.1
     # via ramalama-stack
 referencing==0.36.2
     # via
@@ -389,9 +409,10 @@ scipy==1.15.3
     #   sentence-transformers
 sentence-transformers==4.1.0
     # via ramalama-stack
-setuptools==80.9.0
+setuptools==69.5.1
     # via
     #   pymilvus
+    #   ramalama-stack
     #   torch
     #   triton
 six==1.17.0
@@ -408,7 +429,7 @@ sqlalchemy==2.0.41
     # via ramalama-stack
 sse-starlette==2.3.6
     # via mcp
-starlette==0.46.2
+starlette==0.52.1
     # via
     #   fastapi
     #   llama-stack
@@ -451,31 +472,36 @@ triton==3.3.0 ; platform_machine == 'x86_64' and sys_platform == 'linux'
     # via torch
 trl==0.18.1
     # via ramalama-stack
-typing-extensions==4.14.0
+typing-extensions==4.15.0
     # via
+    #   aiosignal
     #   aiosqlite
     #   anyio
     #   fastapi
     #   huggingface-hub
     #   llama-stack-client
+    #   mcp
     #   openai
     #   opentelemetry-sdk
     #   pydantic
     #   pydantic-core
     #   referencing
     #   sentence-transformers
     #   sqlalchemy
+    #   starlette
     #   torch
     #   typing-inspection
-typing-inspection==0.4.1
+typing-inspection==0.4.2
     # via
+    #   fastapi
+    #   mcp
     #   pydantic
     #   pydantic-settings
 tzdata==2025.2
     # via pandas
 ujson==5.10.0
     # via pymilvus
-urllib3==2.4.0
+urllib3==2.6.3
     # via
     #   blobfile
     #   ramalama-stack

setup.py

@@ -9,28 +9,16 @@ def run(self):
         # Run the standard install
         super().run()
 
-        # Write 'providers.d' to '~/.llama/providers.d'
-        # This allows users to see the remote providers
-        providers_dir = os.path.join(self.install_lib, "ramalama_stack", "providers.d")
-        target_dir_1 = os.path.expanduser("~/.llama/providers.d")
-        try:
-            os.makedirs(target_dir_1, exist_ok=True)
-            shutil.copytree(providers_dir, target_dir_1, dirs_exist_ok=True)
-            print(f"Copied {providers_dir} to {target_dir_1}")
-        except Exception as error:
-            print(f"Failed to copy {providers_dir} to {target_dir_1}. Error: {error}")
-            raise
-
         # Write `ramalama-run.yaml` to '~/.llama/distributions/ramalama'
         # This allows users to run the stack
         run_yaml = os.path.join(self.install_lib, "ramalama_stack", "ramalama-run.yaml")
-        target_dir_2 = os.path.expanduser("~/.llama/distributions/ramalama")
+        target_dir = os.path.expanduser("~/.llama/distributions/ramalama")
         try:
-            os.makedirs(target_dir_2, exist_ok=True)
-            shutil.copy(run_yaml, target_dir_2)
-            print(f"Copied {run_yaml} to {target_dir_2}")
+            os.makedirs(target_dir, exist_ok=True)
+            shutil.copy(run_yaml, target_dir)
+            print(f"Copied {run_yaml} to {target_dir}")
         except Exception as error:
-            print(f"Failed to copy {providers_dir} to {target_dir_1}. Error: {error}")
+            print(f"Failed to copy {providers_dir} to {target_dir}. Error: {error}")
             raise
 
 

src/ramalama_stack/models.py

@@ -9,66 +9,37 @@
 from llama_stack.providers.utils.inference.model_registry import (
     ProviderModelEntry,
     build_hf_repo_model_entry,
-    build_model_entry,
 )
 
 model_entries = [
     build_hf_repo_model_entry(
         "llama3.1:8b-instruct-fp16",
         CoreModelId.llama3_1_8b_instruct.value,
     ),
-    build_model_entry(
-        "llama3.1:8b",
-        CoreModelId.llama3_1_8b_instruct.value,
-    ),
     build_hf_repo_model_entry(
         "llama3.1:70b-instruct-fp16",
         CoreModelId.llama3_1_70b_instruct.value,
     ),
-    build_model_entry(
-        "llama3.1:70b",
-        CoreModelId.llama3_1_70b_instruct.value,
-    ),
     build_hf_repo_model_entry(
         "llama3.1:405b-instruct-fp16",
         CoreModelId.llama3_1_405b_instruct.value,
     ),
-    build_model_entry(
-        "llama3.1:405b",
-        CoreModelId.llama3_1_405b_instruct.value,
-    ),
     build_hf_repo_model_entry(
         "llama3.2:1b-instruct-fp16",
         CoreModelId.llama3_2_1b_instruct.value,
     ),
-    build_model_entry(
-        "llama3.2:1b",
-        CoreModelId.llama3_2_1b_instruct.value,
-    ),
     build_hf_repo_model_entry(
         "llama3.2:3b-instruct-fp16",
         CoreModelId.llama3_2_3b_instruct.value,
     ),
-    build_model_entry(
-        "llama3.2:3b",
-        CoreModelId.llama3_2_3b_instruct.value,
-    ),
     build_hf_repo_model_entry(
         "llama3.2-vision:11b-instruct-fp16",
         CoreModelId.llama3_2_11b_vision_instruct.value,
     ),
-    build_model_entry(
-        "llama3.2-vision:latest",
-        CoreModelId.llama3_2_11b_vision_instruct.value,
-    ),
     build_hf_repo_model_entry(
         "llama3.2-vision:90b-instruct-fp16",
         CoreModelId.llama3_2_90b_vision_instruct.value,
     ),
-    build_model_entry(
-        "llama3.2-vision:90b",
-        CoreModelId.llama3_2_90b_vision_instruct.value,
-    ),
     build_hf_repo_model_entry(
         "llama3.3:70b",
         CoreModelId.llama3_3_70b_instruct.value,