together 1.5.35__py3-none-any.whl → 2.0.0a7__py3-none-any.whl

together/{cli → lib/cli}/api/models.py

@@ -1,10 +1,14 @@
 import json as json_lib
+from typing import Any, Dict, List, Literal, Optional
 
 import click
 from tabulate import tabulate
 
-from together import Together
-from together.types.models import ModelObject, ModelUploadResponse
+from together import Together, omit
+from together._models import BaseModel
+from together._response import APIResponse as APIResponse
+from together.lib.resources.models import filter_by_dedicated_models
+from together.types.model_upload_response import ModelUploadResponse
 
 
 @click.group()
@@ -26,16 +30,19 @@ def models(ctx: click.Context) -> None:
     help="Output in JSON format",
 )
 @click.pass_context
-def list(ctx: click.Context, type: str | None, json: bool) -> None:
+def list(ctx: click.Context, type: Optional[str], json: bool) -> None:
     """List models"""
     client: Together = ctx.obj
 
-    response = client.models.list(dedicated=(type == "dedicated"))
+    response = client.models.list()
+    models_list = response
 
-    display_list = []
+    if type == "dedicated":
+        models_list = filter_by_dedicated_models(client, models_list)
 
-    model: ModelObject
-    for model in response:
+    display_list: List[Dict[str, Any]] = []
+    model: BaseModel
+    for model in models_list:
         display_list.append(
             {
                 "ID": model.id,
@@ -44,8 +51,8 @@ def list(ctx: click.Context, type: str | None, json: bool) -> None:
                 "Type": model.type,
                 "Context Length": model.context_length,
                 "License": model.license,
-                "Input per 1M token": model.pricing.input,
-                "Output per 1M token": model.pricing.output,
+                "Input per 1M token": model.pricing.input if model.pricing else None,
+                "Output per 1M token": model.pricing.output if model.pricing else None,
             }
         )
 
@@ -98,12 +105,12 @@ def upload(
     ctx: click.Context,
     model_name: str,
     model_source: str,
-    model_type: str,
-    hf_token: str | None,
-    description: str | None,
-    base_model: str | None,
-    lora_model: str | None,
+    hf_token: Optional[str],
+    description: Optional[str],
+    base_model: Optional[str],
+    lora_model: Optional[str],
     json: bool,
+    model_type: Optional[Literal["model", "adapter"]] = "model",
 ) -> None:
     """Upload a custom model or adapter from Hugging Face or S3"""
     client: Together = ctx.obj
@@ -111,23 +118,23 @@ def upload(
     response: ModelUploadResponse = client.models.upload(
         model_name=model_name,
         model_source=model_source,
-        model_type=model_type,
-        hf_token=hf_token,
-        description=description,
-        base_model=base_model,
-        lora_model=lora_model,
+        model_type=model_type or omit,
+        hf_token=hf_token or omit,
+        description=description or omit,
+        base_model=base_model or omit,
+        lora_model=lora_model or omit,
     )
 
     if json:
         click.echo(json_lib.dumps(response.model_dump(), indent=2))
     else:
         click.echo(f"Model upload job created successfully!")
-        if response.job_id:
-            click.echo(f"Job ID: {response.job_id}")
-        if response.model_name:
-            click.echo(f"Model Name: {response.model_name}")
-        if response.model_id:
-            click.echo(f"Model ID: {response.model_id}")
-        if response.model_source:
-            click.echo(f"Model Source: {response.model_source}")
+        if response.data.job_id:
+            click.echo(f"Job ID: {response.data.job_id}")
+        if response.data.x_model_name:
+            click.echo(f"Model Name: {response.data.x_model_name}")
+        if response.data.x_model_id:
+            click.echo(f"Model ID: {response.data.x_model_id}")
+        if response.data.x_model_source:
+            click.echo(f"Model Source: {response.data.x_model_source}")
         click.echo(f"Message: {response.message}")

together/lib/cli/api/utils.py

@@ -0,0 +1,50 @@
+from __future__ import annotations
+
+from typing import Literal
+from gettext import gettext as _
+from typing_extensions import override
+
+import click
+
+
+class AutoIntParamType(click.ParamType):
+    name = "integer_or_max"
+    _number_class = int
+
+    @override
+    def convert(
+        self, value: str, param: click.Parameter | None, ctx: click.Context | None
+    ) -> int | Literal["max"] | None:
+        if value == "max":
+            return "max"
+        try:
+            return int(value)
+        except ValueError:
+            self.fail(
+                _("{value!r} is not a valid {number_type}.").format(value=value, number_type=self.name),
+                param,
+                ctx,
+            )
+
+
+class BooleanWithAutoParamType(click.ParamType):
+    name = "boolean_or_auto"
+
+    @override
+    def convert(
+        self, value: str, param: click.Parameter | None, ctx: click.Context | None
+    ) -> bool | Literal["auto"] | None:
+        if value == "auto":
+            return "auto"
+        try:
+            return bool(value)
+        except ValueError:
+            self.fail(
+                _("{value!r} is not a valid {type}.").format(value=value, type=self.name),
+                param,
+                ctx,
+            )
+
+
+INT_WITH_MAX = AutoIntParamType()
+BOOL_WITH_AUTO = BooleanWithAutoParamType()

together/{cli → lib/cli}/cli.py

@@ -6,21 +6,19 @@ from typing import Any
 import click
 
 import together
-from together.cli.api.chat import chat, interactive
-from together.cli.api.completions import completions
-from together.cli.api.endpoints import endpoints
-from together.cli.api.evaluation import evaluation
-from together.cli.api.files import files
-from together.cli.api.finetune import fine_tuning
-from together.cli.api.images import images
-from together.cli.api.models import models
-from together.constants import MAX_RETRIES, TIMEOUT_SECS
+from together._version import __version__
+from together._constants import DEFAULT_TIMEOUT
+from together.lib.cli.api.evals import evals
+from together.lib.cli.api.files import files
+from together.lib.cli.api.models import models
+from together.lib.cli.api.endpoints import endpoints
+from together.lib.cli.api.fine_tuning import fine_tuning
 
 
-def print_version(ctx: click.Context, params: Any, value: Any) -> None:
+def print_version(ctx: click.Context, _params: Any, value: Any) -> None:
     if not value or ctx.resilient_parsing:
         return
-    click.echo(f"Version {together.version}")
+    click.echo(f"Version {__version__}")
     ctx.exit()
 
 
@@ -32,16 +30,12 @@ def print_version(ctx: click.Context, params: Any, value: Any) -> None:
     help="API Key. Defaults to environment variable `TOGETHER_API_KEY`",
     default=os.getenv("TOGETHER_API_KEY"),
 )
-@click.option(
-    "--base-url", type=str, help="API Base URL. Defaults to Together AI endpoint."
-)
-@click.option(
-    "--timeout", type=int, help=f"Request timeout. Defaults to {TIMEOUT_SECS} seconds"
-)
+@click.option("--base-url", type=str, help="API Base URL. Defaults to Together AI endpoint.")
+@click.option("--timeout", type=int, help=f"Request timeout. Defaults to {DEFAULT_TIMEOUT} seconds")
 @click.option(
     "--max-retries",
     type=int,
-    help=f"Maximum number of HTTP retries. Defaults to {MAX_RETRIES}.",
+    help=f"Maximum number of HTTP retries.",
 )
 @click.option(
     "--version",
@@ -57,25 +51,21 @@ def main(
     api_key: str | None,
     base_url: str | None,
     timeout: int | None,
-    max_retries: int | None,
     debug: bool | None,
+    max_retries: int | None,
 ) -> None:
     """This is a sample CLI tool."""
-    together.log = "debug" if debug else None
+    os.environ.setdefault("TOGETHER_LOG", "debug" if debug else "info")
     ctx.obj = together.Together(
-        api_key=api_key, base_url=base_url, timeout=timeout, max_retries=max_retries
+        api_key=api_key, base_url=base_url, timeout=timeout, max_retries=max_retries if max_retries is not None else 0
     )
 
 
-main.add_command(chat)
-main.add_command(interactive)
-main.add_command(completions)
-main.add_command(images)
 main.add_command(files)
 main.add_command(fine_tuning)
 main.add_command(models)
 main.add_command(endpoints)
-main.add_command(evaluation)
+main.add_command(evals)
 
 if __name__ == "__main__":
     main()

together/{constants.py → lib/constants.py}

@@ -1,16 +1,15 @@
 import enum
 
-
 # Session constants
-TIMEOUT_SECS = 600
-MAX_SESSION_LIFETIME_SECS = 180
-MAX_CONNECTION_RETRIES = 2
-MAX_RETRIES = 5
-INITIAL_RETRY_DELAY = 0.5
-MAX_RETRY_DELAY = 8.0
+# TIMEOUT_SECS = 600
+# MAX_SESSION_LIFETIME_SECS = 180
+# MAX_CONNECTION_RETRIES = 2
+# MAX_RETRIES = 5
+# INITIAL_RETRY_DELAY = 0.5
+# MAX_RETRY_DELAY = 8.0
 
 # API defaults
-BASE_URL = "https://api.together.xyz/v1"
+# BASE_URL = "https://api.together.xyz/v1"
 
 # Download defaults
 DOWNLOAD_BLOCK_SIZE = 10 * 1024 * 1024  # 10 MB
@@ -21,31 +20,19 @@ MAX_CONCURRENT_PARTS = 4  # Maximum concurrent parts for multipart upload
 
 # Multipart upload constants
 MIN_PART_SIZE_MB = 5  # Minimum part size (S3 requirement)
-TARGET_PART_SIZE_MB = 250  # Target part size
-MAX_MULTIPART_PARTS = 250  # Maximum parts per upload
+TARGET_PART_SIZE_MB = 250  # Target part size for optimal performance
+MAX_MULTIPART_PARTS = 250  # Maximum parts per upload (S3 limit)
 MULTIPART_UPLOAD_TIMEOUT = 300  # Timeout in seconds for uploading each part
 MULTIPART_THRESHOLD_GB = 5.0  # threshold for switching to multipart upload
 
-# maximum number of GB sized files we support finetuning for
-MAX_FILE_SIZE_GB = 50.1
-
-
-# Messages
-MISSING_API_KEY_MESSAGE = """TOGETHER_API_KEY not found.
-Please set it as an environment variable or set it as together.api_key
-Find your TOGETHER_API_KEY at https://api.together.xyz/settings/api-keys"""
-
 # Minimum number of samples required for fine-tuning file
 MIN_SAMPLES = 1
 
 # the number of bytes in a gigabyte, used to convert bytes to GB for readable comparison
 NUM_BYTES_IN_GB = 2**30
 
-# Multimodal limits
-MAX_IMAGES_PER_EXAMPLE = 10
-MAX_IMAGE_BYTES = 10 * 1024 * 1024  # 10MB
-# Max length = Header length + base64 factor (4/3) * image bytes
-MAX_BASE64_IMAGE_LENGTH = len("data:image/jpeg;base64,") + 4 * MAX_IMAGE_BYTES // 3
+# maximum number of GB sized files we support finetuning for
+MAX_FILE_SIZE_GB = 50.1
 
 # expected columns for Parquet files
 PARQUET_EXPECTED_COLUMNS = ["input_ids", "attention_mask", "labels"]

together/lib/resources/__init__.py

@@ -0,0 +1,11 @@
+from .files import (
+    UploadManager,
+    DownloadManager,
+    AsyncUploadManager,
+)
+
+__all__ = [
+    "DownloadManager",
+    "UploadManager",
+    "AsyncUploadManager",
+]