)
The Together Python library provides convenient access to the Together REST API from any Python 3.9+ application. The library includes type definitions for all request params and response fields, and offers both synchronous and asynchronous clients powered by httpx.
It is generated with Stainless.
The REST API documentation can be found on docs.together.ai. The full API of this library can be found in api.md.
# install from the production repo
pip install git+ssh://git@github.com/togethercomputer/together-py.gitNote
Once this package is published to PyPI, this will become: pip install --pre together
The full API of this library can be found in api.md.
import os
from together import Together
client = Together(
api_key=os.environ.get("TOGETHER_API_KEY"), # This is the default and can be omitted
)
chat_completion = client.chat.completions.create(
messages=[
{
"role": "user",
"content": "Say this is a test!",
}
],
model="meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
)
print(chat_completion.choices)While you can provide an api_key keyword argument,
we recommend using python-dotenv
to add TOGETHER_API_KEY="My API Key" to your .env file
so that your API Key is not stored in source control.
Simply import AsyncTogether instead of Together and use await with each API call:
import os
import asyncio
from together import AsyncTogether
client = AsyncTogether(
api_key=os.environ.get("TOGETHER_API_KEY"), # This is the default and can be omitted
)
async def main() -> None:
chat_completion = await client.chat.completions.create(
messages=[
{
"role": "user",
"content": "Say this is a test!",
}
],
model="meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
)
print(chat_completion.choices)
asyncio.run(main())Functionality between the synchronous and asynchronous clients is otherwise identical.
By default, the async client uses httpx for HTTP requests. However, for improved concurrency performance you may also use aiohttp as the HTTP backend.
You can enable this by installing aiohttp:
# install from the production repo
pip install 'together[aiohttp] @ git+ssh://git@github.com/togethercomputer/together-py.git'Then you can enable it by instantiating the client with http_client=DefaultAioHttpClient():
import asyncio
from together import DefaultAioHttpClient
from together import AsyncTogether
async def main() -> None:
async with AsyncTogether(
api_key="My API Key",
http_client=DefaultAioHttpClient(),
) as client:
chat_completion = await client.chat.completions.create(
messages=[
{
"role": "user",
"content": "Say this is a test!",
}
],
model="meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
)
print(chat_completion.choices)
asyncio.run(main())We provide support for streaming responses using Server Side Events (SSE).
from together import Together
client = Together()
stream = client.chat.completions.create(
messages=[
{
"role": "user",
"content": "Say this is a test",
}
],
model="meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
stream=True,
)
for chat_completion in stream:
print(chat_completion.choices)The async client uses the exact same interface.
from together import AsyncTogether
client = AsyncTogether()
stream = await client.chat.completions.create(
messages=[
{
"role": "user",
"content": "Say this is a test",
}
],
model="meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
stream=True,
)
async for chat_completion in stream:
print(chat_completion.choices)Nested request parameters are TypedDicts. Responses are Pydantic models which also provide helper methods for things like:
- Serializing back into JSON,
model.to_json() - Converting to a dictionary,
model.to_dict()
Typed requests and responses provide autocomplete and documentation within your editor. If you would like to see type errors in VS Code to help catch bugs earlier, set python.analysis.typeCheckingMode to basic.
Nested parameters are dictionaries, typed using TypedDict, for example:
from together import Together
client = Together()
chat_completion = client.chat.completions.create(
messages=[
{
"content": "content",
"role": "system",
}
],
model="meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
response_format={},
)
print(chat_completion.response_format)The async client uses the exact same interface. If you pass a PathLike instance, the file contents will be read asynchronously automatically.
When the library is unable to connect to the API (for example, due to network connection problems or a timeout), a subclass of together.APIConnectionError is raised.
When the API returns a non-success status code (that is, 4xx or 5xx
response), a subclass of together.APIStatusError is raised, containing status_code and response properties.
All errors inherit from together.APIError.
import together
from together import Together
client = Together()
try:
client.chat.completions.create(
messages=[
{
"role": "user",
"content": "Say this is a test",
}
],
model="meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
)
except together.APIConnectionError as e:
print("The server could not be reached")
print(e.__cause__) # an underlying Exception, likely raised within httpx.
except together.RateLimitError as e:
print("A 429 status code was received; we should back off a bit.")
except together.APIStatusError as e:
print("Another non-200-range status code was received")
print(e.status_code)
print(e.response)Error codes are as follows:
| Status Code | Error Type |
|---|---|
| 400 | BadRequestError |
| 401 | AuthenticationError |
| 403 | PermissionDeniedError |
| 404 | NotFoundError |
| 422 | UnprocessableEntityError |
| 429 | RateLimitError |
| >=500 | InternalServerError |
| N/A | APIConnectionError |
Certain errors are automatically retried 2 times by default, with a short exponential backoff. Connection errors (for example, due to a network connectivity problem), 408 Request Timeout, 409 Conflict, 429 Rate Limit, and >=500 Internal errors are all retried by default.
You can use the max_retries option to configure or disable retry settings:
from together import Together
# Configure the default for all requests:
client = Together(
# default is 2
max_retries=0,
)
# Or, configure per-request:
client.with_options(max_retries=5).chat.completions.create(
messages=[
{
"role": "user",
"content": "Say this is a test",
}
],
model="meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
)By default requests time out after 1 minute. You can configure this with a timeout option,
which accepts a float or an httpx.Timeout object:
from together import Together
# Configure the default for all requests:
client = Together(
# 20 seconds (default is 1 minute)
timeout=20.0,
)
# More granular control:
client = Together(
timeout=httpx.Timeout(60.0, read=5.0, write=10.0, connect=2.0),
)
# Override per-request:
client.with_options(timeout=5.0).chat.completions.create(
messages=[
{
"role": "user",
"content": "Say this is a test",
}
],
model="meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
)On timeout, an APITimeoutError is thrown.
Note that requests that time out are retried twice by default.
We use the standard library logging module.
You can enable logging by setting the environment variable TOGETHER_LOG to info.
$ export TOGETHER_LOG=infoOr to debug for more verbose logging.
In an API response, a field may be explicitly null, or missing entirely; in either case, its value is None in this library. You can differentiate the two cases with .model_fields_set:
if response.my_field is None:
if 'my_field' not in response.model_fields_set:
print('Got json like {}, without a "my_field" key present at all.')
else:
print('Got json like {"my_field": null}.')The "raw" Response object can be accessed by prefixing .with_raw_response. to any HTTP method call, e.g.,
from together import Together
client = Together()
response = client.chat.completions.with_raw_response.create(
messages=[{
"role": "user",
"content": "Say this is a test",
}],
model="meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
)
print(response.headers.get('X-My-Header'))
completion = response.parse() # get the object that `chat.completions.create()` would have returned
print(completion.choices)These methods return an APIResponse object.
The async client returns an AsyncAPIResponse with the same structure, the only difference being awaitable methods for reading the response content.
The above interface eagerly reads the full response body when you make the request, which may not always be what you want.
To stream the response body, use .with_streaming_response instead, which requires a context manager and only reads the response body once you call .read(), .text(), .json(), .iter_bytes(), .iter_text(), .iter_lines() or .parse(). In the async client, these are async methods.
with client.chat.completions.with_streaming_response.create(
messages=[
{
"role": "user",
"content": "Say this is a test",
}
],
model="meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
) as response:
print(response.headers.get("X-My-Header"))
for line in response.iter_lines():
print(line)The context manager is required so that the response will reliably be closed.
This library is typed for convenient access to the documented API.
If you need to access undocumented endpoints, params, or response properties, the library can still be used.
To make requests to undocumented endpoints, you can make requests using client.get, client.post, and other
http verbs. Options on the client will be respected (such as retries) when making this request.
import httpx
response = client.post(
"/foo",
cast_to=httpx.Response,
body={"my_param": True},
)
print(response.headers.get("x-foo"))If you want to explicitly send an extra param, you can do so with the extra_query, extra_body, and extra_headers request
options.
To access undocumented response properties, you can access the extra fields like response.unknown_prop. You
can also get all the extra fields on the Pydantic model as a dict with
response.model_extra.
You can directly override the httpx client to customize it for your use case, including:
- Support for proxies
- Custom transports
- Additional advanced functionality
import httpx
from together import Together, DefaultHttpxClient
client = Together(
# Or use the `TOGETHER_BASE_URL` env var
base_url="http://my.test.server.example.com:8083",
http_client=DefaultHttpxClient(
proxy="http://my.test.proxy.example.com",
transport=httpx.HTTPTransport(local_address="0.0.0.0"),
),
)You can also customize the client on a per-request basis by using with_options():
client.with_options(http_client=DefaultHttpxClient(...))By default the library closes underlying HTTP connections whenever the client is garbage collected. You can manually close the client using the .close() method if desired, or with a context manager that closes when exiting.
from together import Together
with Together() as client:
# make requests here
...
# HTTP client is now closed# Help
together files --help
# Check file
together files check example.jsonl
# Upload file
together files upload example.jsonl
# List files
together files list
# Retrieve file metadata
together files retrieve file-6f50f9d1-5b95-416c-9040-0799b2b4b894
# Retrieve file content
together files retrieve-content file-6f50f9d1-5b95-416c-9040-0799b2b4b894
# Delete remote file
together files delete file-6f50f9d1-5b95-416c-9040-0799b2b4b894# Help
together fine-tuning --help
# Create fine-tune job
together fine-tuning create \
--model togethercomputer/llama-2-7b-chat \
--training-file file-711d8724-b3e3-4ae2-b516-94841958117d
# List fine-tune jobs
together fine-tuning list
# Retrieve fine-tune job details
together fine-tuning retrieve ft-c66a5c18-1d6d-43c9-94bd-32d756425b4b
# List fine-tune job events
together fine-tuning list-events ft-c66a5c18-1d6d-43c9-94bd-32d756425b4b
# Cancel running job
together fine-tuning cancel ft-c66a5c18-1d6d-43c9-94bd-32d756425b4b
# Download fine-tuned model weights
together fine-tuning download ft-c66a5c18-1d6d-43c9-94bd-32d756425b4b# Help
together models --help
# List models
together models listThis package generally follows SemVer conventions, though certain backwards-incompatible changes may be released as minor versions:
- Changes that only affect static types, without breaking runtime behavior.
- Changes to library internals which are technically public but not intended or documented for external use. (Please open a GitHub issue to let us know if you are relying on such internals.)
- Changes that we do not expect to impact the vast majority of users in practice.
We take backwards-compatibility seriously and work hard to ensure you can rely on a smooth upgrade experience.
We are keen for your feedback; please open an issue with questions, bugs, or suggestions.
If you've upgraded to the latest version but aren't seeing any new features you were expecting then your python environment is likely still using an older version.
You can determine the version that is being used at runtime with:
import together
print(together.__version__)Python 3.9 or higher.
# Help
together files --help
# Check file
together files check example.jsonl
# Upload file
together files upload example.jsonl
# List files
together files list
# Retrieve file metadata
together files retrieve file-6f50f9d1-5b95-416c-9040-0799b2b4b894
# Retrieve file content
together files retrieve-content file-6f50f9d1-5b95-416c-9040-0799b2b4b894
# Delete remote file
together files delete file-6f50f9d1-5b95-416c-9040-0799b2b4b894# Help
together fine-tuning --help
# Create fine-tune job
together fine-tuning create \
--model togethercomputer/llama-2-7b-chat \
--training-file file-711d8724-b3e3-4ae2-b516-94841958117d
# List fine-tune jobs
together fine-tuning list
# Retrieve fine-tune job details
together fine-tuning retrieve ft-c66a5c18-1d6d-43c9-94bd-32d756425b4b
# List fine-tune job events
together fine-tuning list-events ft-c66a5c18-1d6d-43c9-94bd-32d756425b4b
# List fine-tune checkpoints
together fine-tuning list-checkpoints ft-c66a5c18-1d6d-43c9-94bd-32d756425b4b
# Cancel running job
together fine-tuning cancel ft-c66a5c18-1d6d-43c9-94bd-32d756425b4b
# Download fine-tuned model weights
together fine-tuning download ft-c66a5c18-1d6d-43c9-94bd-32d756425b4b
# Delete fine-tuned model weights
together fine-tuning delete ft-c66a5c18-1d6d-43c9-94bd-32d756425b4b# Help
together models --help
# List models
together models list
# Upload a model
together models upload --model-name my-org/my-model --model-source s3-or-hugging-face