feat: Add async support #76

creatorrr · 2023-03-03T15:40:33Z

Closes #74

Example:

In [2]: import replicate

In [3]: model = replicate.models.get("creatorrr/instructor-large")

In [4]: version = await model.versions.get_async("bd2701dac1aea9d598bda71e6ae56b204287c0
   ...: a79e2cadf96b1393127d044495")

In [5]: inputs = {
   ...:     # Text to embed
   ...:     'text': "Hello world! How are you doing?",
   ...:
   ...:     # Embedding instruction
   ...:     'instruction': "Represent the following text",
   ...: }

In [6]: output = await version.predict_async(**inputs)

In [7]: output
Out[7]:
{'result': [[-0.008334644138813019, ... ]]}

Signed-off-by: Diwank Singh Tomer <diwank.singh@gmail.com>

proksi21 · 2023-03-06T07:20:51Z

hey guys, approve it please, we need async api

creatorrr · 2023-03-24T07:00:07Z

@bfirsh request for a review

ZentixUA · 2023-03-24T10:53:09Z

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/usr/local/lib/python3.10/dist-packages/httpx/_transports/default.py", line 60, in map_httpcore_exceptions
    yield
  File "/usr/local/lib/python3.10/dist-packages/httpx/_transports/default.py", line 353, in handle_async_request
    resp = await self._pool.handle_async_request(req)
  File "/usr/local/lib/python3.10/dist-packages/httpcore/_async/connection_pool.py", line 253, in handle_async_request
    raise exc
  File "/usr/local/lib/python3.10/dist-packages/httpcore/_async/connection_pool.py", line 237, in handle_async_request
    response = await connection.handle_async_request(request)
  File "/usr/local/lib/python3.10/dist-packages/httpcore/_async/connection.py", line 90, in handle_async_request
    return await self._connection.handle_async_request(request)
  File "/usr/local/lib/python3.10/dist-packages/httpcore/_async/http11.py", line 112, in handle_async_request
    raise exc
  File "/usr/local/lib/python3.10/dist-packages/httpcore/_async/http11.py", line 91, in handle_async_request
    ) = await self._receive_response_headers(**kwargs)
  File "/usr/local/lib/python3.10/dist-packages/httpcore/_async/http11.py", line 155, in _receive_response_headers
    event = await self._receive_event(timeout=timeout)
  File "/usr/local/lib/python3.10/dist-packages/httpcore/_async/http11.py", line 191, in _receive_event
    data = await self._network_stream.read(
  File "/usr/local/lib/python3.10/dist-packages/httpcore/backends/asyncio.py", line 31, in read
    with map_exceptions(exc_map):
  File "/usr/lib/python3.10/contextlib.py", line 153, in __exit__
    self.gen.throw(typ, value, traceback)
  File "/usr/local/lib/python3.10/dist-packages/httpcore/_exceptions.py", line 14, in map_exceptions
    raise to_exc(exc)
httpcore.ReadError

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/home/host/servers/bots/adventgpt/main.py", line 437, in improve
    output = await improve_ver.predict_async(**inputs)
  File "/home/host/servers/bots/adventgpt/replicate-python/replicate/version.py", line 37, in predict_async
    prediction = await self._client.predictions.create_async(version=self, input=kwargs)
  File "/home/host/servers/bots/adventgpt/replicate-python/replicate/prediction.py", line 144, in create_async
    resp = await self._client._request_async(
  File "/home/host/servers/bots/adventgpt/replicate-python/replicate/client.py", line 101, in _request_async
    resp = await client.request(method, self.base_url + path, **kwargs)
  File "/usr/local/lib/python3.10/dist-packages/httpx/_client.py", line 1533, in request
    return await self.send(request, auth=auth, follow_redirects=follow_redirects)
  File "/usr/local/lib/python3.10/dist-packages/httpx/_client.py", line 1620, in send
    response = await self._send_handling_auth(
  File "/usr/local/lib/python3.10/dist-packages/httpx/_client.py", line 1648, in _send_handling_auth
    response = await self._send_handling_redirects(
  File "/usr/local/lib/python3.10/dist-packages/httpx/_client.py", line 1685, in _send_handling_redirects
    response = await self._send_single_request(request)
  File "/usr/local/lib/python3.10/dist-packages/httpx/_client.py", line 1722, in _send_single_request
    response = await transport.handle_async_request(request)
  File "/usr/local/lib/python3.10/dist-packages/httpx/_transports/default.py", line 352, in handle_async_request
    with map_httpcore_exceptions():
  File "/usr/lib/python3.10/contextlib.py", line 153, in __exit__
    self.gen.throw(typ, value, traceback)
  File "/usr/local/lib/python3.10/dist-packages/httpx/_transports/default.py", line 77, in map_httpcore_exceptions
    raise mapped_exc(message) from exc
httpx.ReadError

At the moment, even 2 tasks at the same time, it can not withstand.

I'm not sure what I'm saying, so I'm attaching my code:

CODE:

os.environ[str("REPLICATE_API_TOKEN")] = "..."
...

async def replicate_load():
    global improve_ver
    improve_model = replicate.models.get("xinntao/realesrgan")
    improve_ver = await improve_model.versions.get_async("1b976a4d456ed9e4d1a846597b7614e79eadad3032e9124fa63859db0fd59b56")

@bot.event
async def on_ready():
    await replicate_load()
    
async def....
    try:
        inputs = {
            # Input
            'img': f'{attachment.url}',

            'version': "General - RealESRGANplus",

            'scale': 8,

            'face_enhance': True,

            'tile': 0,
        }

        output = await improve_ver.predict_async(**inputs)
        embed = discord.Embed(title="Photo:", color=discord.Colour.blurple()).set_image(url=output)
        await msg.edit(embed=embed)
    except Exception as e:
        print(traceback.format_exc())

ZentixUA · 2023-03-24T11:34:40Z

I reinstalled all dependencies - the problem remained

Signed-off-by: Diwank Singh Tomer <diwank.singh@gmail.com>

creatorrr · 2023-03-24T18:43:37Z

@GenifeG can you please share a github gist to reproduce the error? I wrote a simple test and that seems to be passing

ZentixUA · 2023-03-24T19:24:09Z

At the moment I can't.

Please use the model from my code in the test, and run the function twice with an interval of, for example, 2 seconds (when the first one is not yet ready)

ZentixUA · 2023-03-25T16:54:40Z

@creatorrr

import asyncio
import os
import traceback
import replicate

os.environ[str("REPLICATE_API_TOKEN")] = "YOUR_API_KEY"


async def replicate_load():
    global improve_ver
    improve_model = replicate.models.get("xinntao/realesrgan")
    improve_ver = await improve_model.versions.get_async(
        "1b976a4d456ed9e4d1a846597b7614e79eadad3032e9124fa63859db0fd59b56")


asyncio.run(replicate_load())


async def process():
    await asyncio.sleep(3)
    print(66)
    try:
        inputs = {
            # Input
            'img': f'https://media.discordapp.net/attachments/640960911204548648/1089223567746551889/image.png',

            'version': "General - RealESRGANplus",

            'scale': 3,
            'face_enhance': True,
            'tile': 0,
        }

        output = await improve_ver.predict_async(**inputs)
        print(output)
    except Exception:
        print(traceback.format_exc())


async def main():
    tasks = []
    for i in range(2):
        tasks.append(asyncio.create_task(process()))
        await asyncio.sleep(2)
    await asyncio.gather(*tasks)


asyncio.run(main())

Use it for test. Thanks!

ZentixUA · 2023-03-30T12:41:21Z

Any changes?

issafuad · 2023-05-11T14:41:49Z

Hi is this going to be merged soon ? could really use the async support . many thanks

haoming29 · 2023-09-28T02:44:22Z

Hey @creatorrr Any updates to this PR? Is it idle? Many thanks.

creatorrr · 2023-09-28T06:04:18Z

@Techming idle unfortunately. I'm not using this lib anymore so not able to find time to update this. Any takers?

mattt · 2023-09-28T09:46:17Z

Thanks for your contribution, @creatorrr, and to @Techming and everyone else for your interest. I apologize for not responding sooner to this.

I have an alternative implementation for async support in the async branch. It's effectively a rewrite of the current library, so I'm working to incrementally pull in functionality to minimize breaking changes. The first step towards that is #147, which replaces requests with httpx, which has async support.

Please take a look and let me know what you think. I hope to get this all wrapped up soon.

haoming29 · 2023-09-28T14:42:32Z

Thanks @creatorrr for your previous work! and thank you @mattt for the updates and taking over. It seems that #147 has been ready for review for a couple of weeks and I'm wondering if there will be new features added to that PR or if we are simply waiting for review and approval. Thanks.

This PR adds support for async operations to the Replicate client. Namespace operations like `predictions.list` and `models.create` now have async variants with the `async_` prefix (`predictions.async_list` and `models.async_create`). Here's an example of what that looks like in practice: ```python import replicate model = await replicate.models.async_get("stability-ai/sdxl") input = { "prompt": "A chariot pulled by a team of rainbow unicorns, driven by an astronaut, dramatic lighting", } output = await replicate.async_run(f"stability-ai/sdxl:{model.latest_version.id}", input) ``` <details> <summary>Output</summary> <img src="https://github.com/replicate/replicate-python/assets/7659/6927f8b4-5f92-495d-a87c-135f31aa1847"/> </details> One of the most common questions I hear is how to run a bunch of predictions in parallel. The async functionality provided by this PR makes it really straightforward: ```python import asyncio import replicate # https://replicate.com/stability-ai/sdxl model_version = "stability-ai/sdxl:39ed52f2a78e934b3ba6e2a89f5b1c712de7dfea535525255b1aa35c5565e08b" prompts = [ f"A chariot pulled by a team of {count} rainbow unicorns" for count in ["two", "four", "six", "eight"] ] async with asyncio.TaskGroup() as tg: tasks = [ tg.create_task(replicate.async_run(model_version, input={"prompt": prompt})) for prompt in prompts ] results = await asyncio.gather(*tasks) print(results) ``` Under the hood, `Client` manages an `httpx.Client` and an `httpx.AsyncClient`, which handle calls to `_request` and `_async_request`, respectively. Both are created lazily, so API consumers using only sync or only async functionality won't be affected by functionality they aren't using. Implementation-wise, sync and async variants have separate code paths. This creates nontrivial amounts of duplication, but its benefits to clarity and performance justify those costs. For instance, it'd have been nice if the sync variant were implemented as a blocking call to the async variant, but that would require starting an event loop, which has additional overhead and causes problems if done within an existing event loop. Alternative to #76 Resolves #145 Resolves #107 Resolves #74 --------- Signed-off-by: Mattt Zmuda <mattt@replicate.com>

mattt · 2023-11-09T11:14:00Z

Hi @creatorrr. Following up on #76 (comment) —

I just merged #193, which implements async support. One notable difference in this API is that the naming convention uses async as a method name prefix, rather than a suffix. Otherwise, it should be a drop-in replacement for your fork or existing synchronous usage.

Thank you again for contributing this PR. Please let me know if you see any ways we can improve our implementation of async.

This PR adds support for async operations to the Replicate client. Namespace operations like `predictions.list` and `models.create` now have async variants with the `async_` prefix (`predictions.async_list` and `models.async_create`). Here's an example of what that looks like in practice: ```python import replicate model = await replicate.models.async_get("stability-ai/sdxl") input = { "prompt": "A chariot pulled by a team of rainbow unicorns, driven by an astronaut, dramatic lighting", } output = await replicate.async_run(f"stability-ai/sdxl:{model.latest_version.id}", input) ``` <details> <summary>Output</summary> <img src="https://github.com/replicate/replicate-python/assets/7659/6927f8b4-5f92-495d-a87c-135f31aa1847"/> </details> One of the most common questions I hear is how to run a bunch of predictions in parallel. The async functionality provided by this PR makes it really straightforward: ```python import asyncio import replicate # https://replicate.com/stability-ai/sdxl model_version = "stability-ai/sdxl:39ed52f2a78e934b3ba6e2a89f5b1c712de7dfea535525255b1aa35c5565e08b" prompts = [ f"A chariot pulled by a team of {count} rainbow unicorns" for count in ["two", "four", "six", "eight"] ] async with asyncio.TaskGroup() as tg: tasks = [ tg.create_task(replicate.async_run(model_version, input={"prompt": prompt})) for prompt in prompts ] results = await asyncio.gather(*tasks) print(results) ``` Under the hood, `Client` manages an `httpx.Client` and an `httpx.AsyncClient`, which handle calls to `_request` and `_async_request`, respectively. Both are created lazily, so API consumers using only sync or only async functionality won't be affected by functionality they aren't using. Implementation-wise, sync and async variants have separate code paths. This creates nontrivial amounts of duplication, but its benefits to clarity and performance justify those costs. For instance, it'd have been nice if the sync variant were implemented as a blocking call to the async variant, but that would require starting an event loop, which has additional overhead and causes problems if done within an existing event loop. Alternative to replicate/replicate-python#76 Resolves replicate/replicate-python#145 Resolves replicate/replicate-python#107 Resolves replicate/replicate-python#74 --------- Signed-off-by: Mattt Zmuda <mattt@replicate.com>

feat: Add async support

363c8f5

Signed-off-by: Diwank Singh Tomer <diwank.singh@gmail.com>

0dminnimda mentioned this pull request Mar 11, 2023

Request: Add async support #74

Closed

Merge branch 'main' into f/async-support

c555a52

feat: Add a simple test

648e351

Signed-off-by: Diwank Singh Tomer <diwank.singh@gmail.com>

antont mentioned this pull request Oct 19, 2023

Is there an async option while calling predict in the client? #107

Closed

mattt mentioned this pull request Nov 8, 2023

Add async support #193

Merged

mattt closed this Nov 9, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: Add async support #76

feat: Add async support #76

creatorrr commented Mar 3, 2023

proksi21 commented Mar 6, 2023

creatorrr commented Mar 24, 2023

ZentixUA commented Mar 24, 2023 •

edited

ZentixUA commented Mar 24, 2023

creatorrr commented Mar 24, 2023

ZentixUA commented Mar 24, 2023

ZentixUA commented Mar 25, 2023

ZentixUA commented Mar 30, 2023

issafuad commented May 11, 2023

haoming29 commented Sep 28, 2023

creatorrr commented Sep 28, 2023

mattt commented Sep 28, 2023 •

edited

haoming29 commented Sep 28, 2023

mattt commented Nov 9, 2023

feat: Add async support #76

feat: Add async support #76

Conversation

creatorrr commented Mar 3, 2023

Example:

proksi21 commented Mar 6, 2023

creatorrr commented Mar 24, 2023

ZentixUA commented Mar 24, 2023 • edited

ZentixUA commented Mar 24, 2023

creatorrr commented Mar 24, 2023

ZentixUA commented Mar 24, 2023

ZentixUA commented Mar 25, 2023

ZentixUA commented Mar 30, 2023

issafuad commented May 11, 2023

haoming29 commented Sep 28, 2023

creatorrr commented Sep 28, 2023

mattt commented Sep 28, 2023 • edited

haoming29 commented Sep 28, 2023

mattt commented Nov 9, 2023

ZentixUA commented Mar 24, 2023 •

edited

mattt commented Sep 28, 2023 •

edited