together 2.0.0a17__py3-none-any.whl → 2.0.0a19__py3-none-any.whl

together/types/beta/jig/volume_update_params.py

@@ -0,0 +1,34 @@
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from __future__ import annotations
+
+from typing_extensions import Literal, TypedDict
+
+__all__ = ["VolumeUpdateParams", "Content"]
+
+
+class VolumeUpdateParams(TypedDict, total=False):
+    content: Content
+    """Content specifies the new content that will be preloaded to this volume"""
+
+    name: str
+    """Name is the new unique identifier for the volume within the project"""
+
+    type: Literal["readOnly"]
+    """Type is the new volume type (currently only "readOnly" is supported)"""
+
+
+class Content(TypedDict, total=False):
+    """Content specifies the new content that will be preloaded to this volume"""
+
+    source_prefix: str
+    """
+    SourcePrefix is the file path prefix for the content to be preloaded into the
+    volume
+    """
+
+    type: Literal["files"]
+    """
+    Type is the content type (currently only "files" is supported which allows
+    preloading files uploaded via Files API into the volume)
+    """

together/types/beta/jig_deploy_params.py

@@ -0,0 +1,150 @@
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from __future__ import annotations
+
+from typing import Dict, Iterable
+from typing_extensions import Literal, Required, TypedDict
+
+from ..._types import SequenceNotStr
+
+__all__ = ["JigDeployParams", "EnvironmentVariable", "Volume"]
+
+
+class JigDeployParams(TypedDict, total=False):
+    gpu_type: Required[Literal["h100-80gb", "a100-80gb"]]
+    """GPUType specifies the GPU hardware to use (e.g., "h100-80gb")."""
+
+    image: Required[str]
+    """Image is the container image to deploy from registry.together.ai."""
+
+    name: Required[str]
+    """Name is the unique identifier for your deployment.
+
+    Must contain only alphanumeric characters, underscores, or hyphens (1-100
+    characters)
+    """
+
+    args: SequenceNotStr[str]
+    """Args overrides the container's CMD.
+
+    Provide as an array of arguments (e.g., ["python", "app.py"])
+    """
+
+    autoscaling: Dict[str, str]
+    """Autoscaling configuration as key-value pairs.
+
+    Example: {"metric": "QueueBacklogPerWorker", "target": "10"} to scale based on
+    queue backlog
+    """
+
+    command: SequenceNotStr[str]
+    """Command overrides the container's ENTRYPOINT.
+
+    Provide as an array (e.g., ["/bin/sh", "-c"])
+    """
+
+    cpu: float
+    """
+    CPU is the number of CPU cores to allocate per container instance (e.g., 0.1 =
+    100 milli cores)
+    """
+
+    description: str
+    """Description is an optional human-readable description of your deployment"""
+
+    environment_variables: Iterable[EnvironmentVariable]
+    """EnvironmentVariables is a list of environment variables to set in the container.
+
+    Each must have a name and either a value or value_from_secret
+    """
+
+    gpu_count: int
+    """GPUCount is the number of GPUs to allocate per container instance.
+
+    Defaults to 0 if not specified
+    """
+
+    health_check_path: str
+    """HealthCheckPath is the HTTP path for health checks (e.g., "/health").
+
+    If set, the platform will check this endpoint to determine container health
+    """
+
+    max_replicas: int
+    """
+    MaxReplicas is the maximum number of container instances that can be scaled up
+    to. If not set, will be set to MinReplicas
+    """
+
+    memory: float
+    """
+    Memory is the amount of RAM to allocate per container instance in GiB (e.g., 0.5
+    = 512MiB)
+    """
+
+    min_replicas: int
+    """MinReplicas is the minimum number of container instances to run.
+
+    Defaults to 1 if not specified
+    """
+
+    port: int
+    """
+    Port is the container port your application listens on (e.g., 8080 for web
+    servers). Required if your application serves traffic
+    """
+
+    storage: int
+    """
+    Storage is the amount of ephemeral disk storage to allocate per container
+    instance (e.g., 10 = 10GiB)
+    """
+
+    termination_grace_period_seconds: int
+    """
+    TerminationGracePeriodSeconds is the time in seconds to wait for graceful
+    shutdown before forcefully terminating the replica
+    """
+
+    volumes: Iterable[Volume]
+    """Volumes is a list of volume mounts to attach to the container.
+
+    Each mount must reference an existing volume by name
+    """
+
+
+class EnvironmentVariable(TypedDict, total=False):
+    name: Required[str]
+    """Name is the environment variable name (e.g., "DATABASE_URL").
+
+    Must start with a letter or underscore, followed by letters, numbers, or
+    underscores
+    """
+
+    value: str
+    """Value is the plain text value for the environment variable.
+
+    Use this for non-sensitive values. Either Value or ValueFromSecret must be set,
+    but not both
+    """
+
+    value_from_secret: str
+    """ValueFromSecret references a secret by name or ID to use as the value.
+
+    Use this for sensitive values like API keys or passwords. Either Value or
+    ValueFromSecret must be set, but not both
+    """
+
+
+class Volume(TypedDict, total=False):
+    mount_path: Required[str]
+    """
+    MountPath is the path in the container where the volume will be mounted (e.g.,
+    "/data")
+    """
+
+    name: Required[str]
+    """Name is the name of the volume to mount.
+
+    Must reference an existing volume by name or ID
+    """

together/types/beta/jig_list_response.py

@@ -0,0 +1,16 @@
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from typing import List, Optional
+
+from ..._models import BaseModel
+from .deployment import Deployment
+
+__all__ = ["JigListResponse"]
+
+
+class JigListResponse(BaseModel):
+    data: Optional[List[Deployment]] = None
+    """Data is the array of deployment items"""
+
+    object: Optional[str] = None
+    """Object is the type identifier for this response (always "list")"""

together/types/beta/jig_retrieve_logs_params.py

@@ -0,0 +1,12 @@
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from __future__ import annotations
+
+from typing_extensions import TypedDict
+
+__all__ = ["JigRetrieveLogsParams"]
+
+
+class JigRetrieveLogsParams(TypedDict, total=False):
+    replica_id: str
+    """Replica ID to filter logs"""

together/types/beta/jig_update_params.py

@@ -0,0 +1,141 @@
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from __future__ import annotations
+
+from typing import Dict, Iterable
+from typing_extensions import Literal, Required, TypedDict
+
+from ..._types import SequenceNotStr
+
+__all__ = ["JigUpdateParams", "EnvironmentVariable", "Volume"]
+
+
+class JigUpdateParams(TypedDict, total=False):
+    args: SequenceNotStr[str]
+    """Args overrides the container's CMD.
+
+    Provide as an array of arguments (e.g., ["python", "app.py"])
+    """
+
+    autoscaling: Dict[str, str]
+    """Autoscaling configuration as key-value pairs.
+
+    Example: {"metric": "QueueBacklogPerWorker", "target": "10"} to scale based on
+    queue backlog
+    """
+
+    command: SequenceNotStr[str]
+    """Command overrides the container's ENTRYPOINT.
+
+    Provide as an array (e.g., ["/bin/sh", "-c"])
+    """
+
+    cpu: float
+    """
+    CPU is the number of CPU cores to allocate per container instance (e.g., 0.1 =
+    100 milli cores)
+    """
+
+    description: str
+    """Description is an optional human-readable description of your deployment"""
+
+    environment_variables: Iterable[EnvironmentVariable]
+    """EnvironmentVariables is a list of environment variables to set in the container.
+
+    This will replace all existing environment variables
+    """
+
+    gpu_count: int
+    """GPUCount is the number of GPUs to allocate per container instance"""
+
+    gpu_type: Literal["h100-80gb", " a100-80gb"]
+    """GPUType specifies the GPU hardware to use (e.g., "h100-80gb")"""
+
+    health_check_path: str
+    """HealthCheckPath is the HTTP path for health checks (e.g., "/health").
+
+    Set to empty string to disable health checks
+    """
+
+    image: str
+    """Image is the container image to deploy from registry.together.ai."""
+
+    max_replicas: int
+    """MaxReplicas is the maximum number of replicas that can be scaled up to."""
+
+    memory: float
+    """
+    Memory is the amount of RAM to allocate per container instance in GiB (e.g., 0.5
+    = 512MiB)
+    """
+
+    min_replicas: int
+    """MinReplicas is the minimum number of replicas to run"""
+
+    name: str
+    """Name is the new unique identifier for your deployment.
+
+    Must contain only alphanumeric characters, underscores, or hyphens (1-100
+    characters)
+    """
+
+    port: int
+    """
+    Port is the container port your application listens on (e.g., 8080 for web
+    servers)
+    """
+
+    storage: int
+    """
+    Storage is the amount of ephemeral disk storage to allocate per container
+    instance (e.g., 10 = 10GiB)
+    """
+
+    termination_grace_period_seconds: int
+    """
+    TerminationGracePeriodSeconds is the time in seconds to wait for graceful
+    shutdown before forcefully terminating the replica
+    """
+
+    volumes: Iterable[Volume]
+    """Volumes is a list of volume mounts to attach to the container.
+
+    This will replace all existing volumes
+    """
+
+
+class EnvironmentVariable(TypedDict, total=False):
+    name: Required[str]
+    """Name is the environment variable name (e.g., "DATABASE_URL").
+
+    Must start with a letter or underscore, followed by letters, numbers, or
+    underscores
+    """
+
+    value: str
+    """Value is the plain text value for the environment variable.
+
+    Use this for non-sensitive values. Either Value or ValueFromSecret must be set,
+    but not both
+    """
+
+    value_from_secret: str
+    """ValueFromSecret references a secret by name or ID to use as the value.
+
+    Use this for sensitive values like API keys or passwords. Either Value or
+    ValueFromSecret must be set, but not both
+    """
+
+
+class Volume(TypedDict, total=False):
+    mount_path: Required[str]
+    """
+    MountPath is the path in the container where the volume will be mounted (e.g.,
+    "/data")
+    """
+
+    name: Required[str]
+    """Name is the name of the volume to mount.
+
+    Must reference an existing volume by name or ID
+    """

together/types/chat/completion_create_params.py

@@ -28,6 +28,7 @@ __all__ = [
     "MessageChatCompletionFunctionMessageParam",
     "FunctionCall",
     "FunctionCallName",
+    "Reasoning",
     "ResponseFormat",
     "ResponseFormatText",
     "ResponseFormatJsonSchema",
@@ -111,6 +112,8 @@ class CompletionCreateParamsBase(TypedDict, total=False):
     a model talking about new topics.
     """
 
+    reasoning: Reasoning
+
     reasoning_effort: Literal["low", "medium", "high"]
     """
     Controls the level of reasoning effort the model should apply when generating
@@ -314,6 +317,14 @@ class FunctionCallName(TypedDict, total=False):
 FunctionCall: TypeAlias = Union[Literal["none", "auto"], FunctionCallName]
 
 
+class Reasoning(TypedDict, total=False):
+    enabled: bool
+    """
+    For models that support toggling reasoning functionality, this object can be
+    used to control that functionality.
+    """
+
+
 class ResponseFormatText(TypedDict, total=False):
     """Default response format. Used to generate text responses."""
 

together/types/{hardware_list_params.py → endpoint_list_hardware_params.py}

@@ -4,10 +4,10 @@ from __future__ import annotations
 
 from typing_extensions import TypedDict
 
-__all__ = ["HardwareListParams"]
+__all__ = ["EndpointListHardwareParams"]
 
 
-class HardwareListParams(TypedDict, total=False):
+class EndpointListHardwareParams(TypedDict, total=False):
     model: str
     """Filter hardware configurations by model compatibility.
 

together/types/{hardware_list_response.py → endpoint_list_hardware_response.py}

@@ -6,7 +6,7 @@ from typing_extensions import Literal
 
 from .._models import BaseModel
 
-__all__ = ["HardwareListResponse", "Data", "DataPricing", "DataSpecs", "DataAvailability"]
+__all__ = ["EndpointListHardwareResponse", "Data", "DataPricing", "DataSpecs", "DataAvailability"]
 
 
 class DataPricing(BaseModel):
@@ -60,7 +60,7 @@ class Data(BaseModel):
     """Indicates the current availability status of a hardware configuration"""
 
 
-class HardwareListResponse(BaseModel):
+class EndpointListHardwareResponse(BaseModel):
     data: List[Data]
 
     object: Literal["list"]

together/types/models/__init__.py

@@ -0,0 +1,5 @@
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from __future__ import annotations
+
+from .upload_status_response import UploadStatusResponse as UploadStatusResponse

together/types/{job_retrieve_response.py → models/upload_status_response.py}

@@ -6,9 +6,9 @@ from typing_extensions import Literal
 
 from pydantic import Field as FieldInfo
 
-from .._models import BaseModel
+from ..._models import BaseModel
 
-__all__ = ["JobRetrieveResponse", "Args", "StatusUpdate"]
+__all__ = ["UploadStatusResponse", "Args", "StatusUpdate"]
 
 
 class Args(BaseModel):
@@ -27,7 +27,7 @@ class StatusUpdate(BaseModel):
     timestamp: datetime
 
 
-class JobRetrieveResponse(BaseModel):
+class UploadStatusResponse(BaseModel):
     args: Args
 
     created_at: datetime

{together-2.0.0a17.dist-info → together-2.0.0a19.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: together
-Version: 2.0.0a17
+Version: 2.0.0a19
 Summary: The official Python library for the together API
 Project-URL: Homepage, https://github.com/togethercomputer/together-py
 Project-URL: Repository, https://github.com/togethercomputer/together-py
@@ -32,11 +32,15 @@ Requires-Dist: pydantic<3,>=1.9.0
 Requires-Dist: rich>=13.7.1
 Requires-Dist: sniffio
 Requires-Dist: tabulate>=0.9.0
+Requires-Dist: tomli>=2.0.0; python_version < '3.11'
 Requires-Dist: tqdm>=4.67.1
 Requires-Dist: types-pyyaml>=6.0.12.20250915
 Requires-Dist: types-tabulate>=0.9.0.20240106
 Requires-Dist: types-tqdm>=4.67.0.20250516
 Requires-Dist: typing-extensions<5,>=4.10
+Provides-Extra: aiofiles
+Requires-Dist: aiofiles>=25.0.0; extra == 'aiofiles'
+Requires-Dist: types-aiofiles; extra == 'aiofiles'
 Provides-Extra: aiohttp
 Requires-Dist: aiohttp; extra == 'aiohttp'
 Requires-Dist: httpx-aiohttp>=0.1.9; extra == 'aiohttp'
@@ -235,20 +239,17 @@ from together import Together
 
 client = Together()
 
-cluster = client.beta.clusters.create(
-    billing_type="RESERVED",
-    cluster_name="cluster_name",
-    driver_version="CUDA_12_5_555",
-    gpu_type="H100_SXM",
-    num_gpus=0,
-    region="us-central-8",
-    shared_volume={
-        "region": "region",
-        "size_tib": 0,
-        "volume_name": "volume_name",
-    },
+chat_completion = client.chat.completions.create(
+    messages=[
+        {
+            "content": "content",
+            "role": "system",
+        }
+    ],
+    model="meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
+    reasoning={},
 )
-print(cluster.shared_volume)
+print(chat_completion.reasoning)
 ```
 
 The async client uses the exact same interface. If you pass a [`PathLike`](https://docs.python.org/3/library/os.html#os.PathLike) instance, the file contents will be read asynchronously automatically.