Unverified Commit c15b8367 authored by OTABI Tomoya's avatar OTABI Tomoya Committed by GitHub
Browse files

Merge pull request #336924 from natsukium/openllm/update 

python312Packages.openllm: 0.5.7 -> 0.6.10; python312Packages.bentoml: 1.2.18 -> 1.3.3
parents 44982550 77707d80

pkgs/by-name/op/openllm/package.nix

0 → 100644
+76 −0
Original line number Diff line number Diff line 
{

  lib,

  fetchFromGitHub,

  python3,

}:

let

  python = python3.override {

    self = python;

    packageOverrides = _: super: {

      cattrs = super.cattrs.overridePythonAttrs (oldAttrs: rec {

        version = "23.1.2";

        build-system = [ super.poetry-core ];

        src = oldAttrs.src.override {

          rev = "refs/tags/v${version}";

          hash = "sha256-YO4Clbo5fmXbysxwwM2qCHJwO5KwDC05VctRVFruJcw=";

        };

      });

    };

  };

in



python.pkgs.buildPythonApplication rec {

  pname = "openllm";

  version = "0.6.10";

  pyproject = true;



  src = fetchFromGitHub {

    owner = "bentoml";

    repo = "openllm";

    rev = "refs/tags/v${version}";

    hash = "sha256-4KIpe6KjbBDDUj0IjzSccxjgZyBoaUVIQJYk1+W01Vo=";

  };



  pythonRemoveDeps = [

    "pathlib"

    "pip-requirements-parser"

  ];



  pythonRelaxDeps = [ "openai" ];



  build-system = with python.pkgs; [

    hatch-vcs

    hatchling

  ];



  dependencies = with python.pkgs; [

    accelerate

    bentoml

    dulwich

    nvidia-ml-py

    openai

    psutil

    pyaml

    questionary

    tabulate

    typer

    uv

  ];



  # no tests

  doCheck = false;



  pythonImportsCheck = [ "openllm" ];



  meta = with lib; {

    description = "Run any open-source LLMs, such as Llama 3.1, Gemma, as OpenAI compatible API endpoint in the cloud";

    homepage = "https://github.com/bentoml/OpenLLM";

    changelog = "https://github.com/bentoml/OpenLLM/releases/tag/v${version}";

    license = licenses.asl20;

    maintainers = with maintainers; [

      happysalada

      natsukium

    ];

    mainProgram = "openllm";

  };

}

pkgs/development/python-modules/bentoml/default.nix

+26 −22
Original line number Diff line number Diff line
@@ -7,6 +7,7 @@ 
  hatchling,

  hatch-vcs,

  aiohttp,

  aiosqlite,

  attrs,

  cattrs,

  circus,
@@ -15,25 +16,38 @@ 
  cloudpickle,

  deepmerge,

  fs,

  fs-s3fs,

  grpcio,

  grpcio-channelz,

  grpcio-health-checking,

  grpcio-reflection,

  httpx,

  httpx-ws,

  inflection,

  inquirerpy,

  jinja2,

  numpy,

  nvidia-ml-py,

  opentelemetry-api,

  opentelemetry-exporter-otlp,

  opentelemetry-exporter-otlp-proto-http,

  opentelemetry-instrumentation,

  opentelemetry-instrumentation-aiohttp-client,

  opentelemetry-instrumentation-asgi,

  opentelemetry-instrumentation-grpc,

  opentelemetry-sdk,

  opentelemetry-semantic-conventions,

  opentelemetry-util-http,

  packaging,

  pandas,

  pathspec,

  pillow,

  pip-requirements-parser,

  pip-tools,

  prometheus-client,

  protobuf,

  psutil,

  pyarrow,

  pydantic,

  python-dateutil,

  python-json-logger,

  python-multipart,
@@ -44,25 +58,10 @@ 
  starlette,

  tomli,

  tomli-w,

  tritonclient,

  uv,

  uvicorn,

  watchfiles,

  fs-s3fs,

  grpcio,

  grpcio-health-checking,

  opentelemetry-instrumentation-grpc,

  protobuf,

  grpcio-channelz,

  grpcio-reflection,

  pillow,

  pydantic,

  pandas,

  pyarrow,

  opentelemetry-exporter-otlp-proto-http,

  # https://pypi.org/project/opentelemetry-exporter-jaeger-proto-grpc/

  # , opentelemetry-exporter-jaeger # support for this exporter ends in july 2023

  opentelemetry-exporter-otlp,

  # , opentelemetry-exporter-zipkin

  tritonclient,

  # native check inputs

  pytestCheckHook,

  pytest-xdist,
@@ -75,7 +74,7 @@ 
}:



let

  version = "1.2.18";

  version = "1.3.3";

  aws = [ fs-s3fs ];

  grpc = [

    grpcio
@@ -90,7 +89,10 @@ let 
  ];

  grpc-reflection = grpc ++ [ grpcio-reflection ];

  grpc-channelz = grpc ++ [ grpcio-channelz ];

  monitor-otlp = [ opentelemetry-exporter-otlp-proto-http ];

  monitor-otlp = [

    opentelemetry-exporter-otlp-proto-http

    opentelemetry-instrumentation-grpc

  ];

  # tracing-jaeger = [ opentelemetry-exporter-jaeger ];

  tracing-otlp = [ opentelemetry-exporter-otlp ];

  # tracing-zipkin = [ opentelemetry-exporter-zipkin ];
@@ -126,7 +128,7 @@ buildPythonPackage { 
    owner = "bentoml";

    repo = "BentoML";

    rev = "refs/tags/v${version}";

    hash = "sha256-giZteSikwS9YEcVMPCC9h2khbBgvUPRW1biAyixO13Y=";

    hash = "sha256-PjmXPSPukLJ+iCpBdUynhcWCfFqplmdsgj0LYpodE/c=";

  };



  pythonRelaxDeps = [
@@ -148,6 +150,7 @@ buildPythonPackage { 


  dependencies = [

    aiohttp

    aiosqlite

    attrs

    cattrs

    circus
@@ -159,6 +162,7 @@ buildPythonPackage { 
    httpx

    httpx-ws

    inflection

    inquirerpy

    jinja2

    numpy

    nvidia-ml-py
@@ -172,7 +176,6 @@ buildPythonPackage { 
    packaging

    pathspec

    pip-requirements-parser

    pip-tools

    prometheus-client

    psutil

    pydantic
@@ -185,6 +188,7 @@ buildPythonPackage { 
    simple-di

    starlette

    tomli-w

    uv

    uvicorn

    watchfiles

  ] ++ lib.optionals (pythonOlder "3.11") [ tomli ];

pkgs/development/python-modules/openllm-client/default.nix

deleted100644 → 0
+0 −70
Original line number Diff line number Diff line 
{

  lib,

  buildPythonPackage,

  pythonOlder,

  bentoml,

  hatch-fancy-pypi-readme,

  hatch-vcs,

  hatchling,

  anyio,

  distro,

  httpx,

  httpx-auth,

  openllm-core,

  soundfile,

  transformers,

}:



buildPythonPackage rec {

  inherit (openllm-core) src version;

  pname = "openllm-client";

  pyproject = true;



  disabled = pythonOlder "3.8";



  sourceRoot = "${src.name}/openllm-client";



  postPatch = ''

    substituteInPlace pyproject.toml \

      --replace-fail "hatchling==1.18.0" "hatchling" \

      --replace-fail "hatch-vcs==0.3.0" "hatch-vcs" \

      --replace-fail "hatch-fancy-pypi-readme==23.1.0" "hatch-fancy-pypi-readme"

  '';



  build-system = [

    hatch-fancy-pypi-readme

    hatch-vcs

    hatchling

  ];



  dependencies = [

    anyio

    distro

    httpx

    openllm-core

  ];



  optional-dependencies = {

    grpc = [ bentoml ] ++ bentoml.optional-dependencies.grpc;

    auth = [ httpx-auth ];

    agents = [

      transformers

      # diffusers

      soundfile

    ] ++ transformers.optional-dependencies.agents;

    full = optional-dependencies.grpc ++ optional-dependencies.agents;

  };



  # there is no tests

  doCheck = false;



  pythonImportsCheck = [ "openllm_client" ];



  meta = with lib; {

    description = "Interacting with OpenLLM HTTP/gRPC server, or any BentoML server";

    homepage = "https://github.com/bentoml/OpenLLM/tree/main/openllm-client";

    changelog = "https://github.com/bentoml/OpenLLM/blob/${src.rev}/CHANGELOG.md";

    license = licenses.asl20;

    maintainers = with maintainers; [ natsukium ];

  };

}

pkgs/development/python-modules/openllm-core/default.nix

deleted100644 → 0
+0 −106
Original line number Diff line number Diff line 
{

  lib,

  buildPythonPackage,

  fetchFromGitHub,

  pythonOlder,

  accelerate,

  attrs,

  bitsandbytes,

  bentoml,

  cattrs,

  click-option-group,

  datasets,

  deepmerge,

  hatch-fancy-pypi-readme,

  hatch-vcs,

  hatchling,

  inflection,

  mypy-extensions,

  orjson,

  peft,

  pydantic,

  transformers,

  typing-extensions,

}:



buildPythonPackage rec {

  pname = "openllm-core";

  version = "0.5.7";

  pyproject = true;



  disabled = pythonOlder "3.8";



  src = fetchFromGitHub {

    owner = "bentoml";

    repo = "OpenLLM";

    rev = "refs/tags/v${version}";

    hash = "sha256-sEZLszzoo39WUnziHGp7zWNO0YaqkXeXAoIxvyhw42w=";

  };



  sourceRoot = "${src.name}/openllm-core";





  postPatch = ''

    substituteInPlace pyproject.toml \

      --replace-fail "hatch-vcs==" "hatch-vcs>=" \

      --replace-fail "hatchling==" "hatchling>=" \

      --replace-fail "hatch-fancy-pypi-readme==" "hatch-fancy-pypi-readme>="

  '';



  pythonRelaxDeps = [ "cattrs" ];



  build-system = [

    hatch-fancy-pypi-readme

    hatch-vcs

    hatchling

  ];



  dependencies = [

    attrs

    cattrs

    pydantic

    # not listed in pyproject.toml, but required at runtime

    click-option-group

    deepmerge

    inflection

    mypy-extensions

    orjson

    typing-extensions

  ];



  optional-dependencies = {

    vllm = [

      # vllm

    ];

    bentoml = [ bentoml ];

    fine-tune = [

      accelerate

      bitsandbytes

      datasets

      peft

      transformers

      # trl

    ] ++ transformers.optional-dependencies.torch ++ transformers.optional-dependencies.tokenizers;

    full =

      with optional-dependencies;

      (

        vllm

        # use absolute path to disambiguate with derivbation argument

        ++ optional-dependencies.bentoml

        ++ fine-tune

      );

  };



  # there is no tests

  doCheck = false;



  pythonImportsCheck = [ "openllm_core" ];



  meta = with lib; {

    description = "Core components for OpenLLM";

    homepage = "https://github.com/bentoml/OpenLLM/tree/main/openllm-core";

    changelog = "https://github.com/bentoml/OpenLLM/blob/${src.rev}/CHANGELOG.md";

    license = licenses.asl20;

    maintainers = with maintainers; [ natsukium ];

  };

}

pkgs/development/python-modules/openllm/default.nix

deleted100644 → 0
+0 −200
Original line number Diff line number Diff line 
{

  lib,

  buildPythonPackage,

  hatch-fancy-pypi-readme,

  hatch-vcs,

  hatchling,

  pytestCheckHook,

  pythonOlder,

  accelerate,

  bentoml,

  bitsandbytes,

  build,

  click,

  ctranslate2,

  datasets,

  docker,

  einops,

  ghapi,

  huggingface-hub,

  hypothesis,

  ipython,

  jupyter,

  jupytext,

  nbformat,

  notebook,

  openai,

  openllm-client,

  openllm-core,

  optimum,

  peft,

  pytest-mock,

  pytest-randomly,

  pytest-rerunfailures,

  pytest-xdist,

  safetensors,

  scipy,

  sentencepiece,

  soundfile,

  syrupy,

  tabulate,

  tiktoken,

  transformers,

  triton,

  xformers,

}:



buildPythonPackage rec {

  inherit (openllm-core) src version;

  pname = "openllm";

  pyproject = true;



  disabled = pythonOlder "3.8";



  sourceRoot = "${src.name}/openllm-python";





  pythonRemoveDeps = [

    # remove cuda-python as it has an unfree license

    "cuda-python"

  ];



  build-system = [

    hatch-fancy-pypi-readme

    hatch-vcs

    hatchling

  ];



  dependencies =

    [

      accelerate

      bentoml

      bitsandbytes

      build

      click

      einops

      ghapi

      openllm-client

      openllm-core

      optimum

      safetensors

      scipy

      sentencepiece

      transformers

    ]

    ++ bentoml.optional-dependencies.io

    ++ tabulate.optional-dependencies.widechars

    ++ transformers.optional-dependencies.tokenizers

    ++ transformers.optional-dependencies.torch;



  optional-dependencies = {

    agents = [

      # diffusers

      soundfile

      transformers

    ] ++ transformers.optional-dependencies.agents;

    awq = [

      # autoawq

    ];

    baichuan = [

      # cpm-kernels

    ];

    chatglm = [

      # cpm-kernels

    ];

    ctranslate = [ ctranslate2 ];

    falcon = [ xformers ];

    fine-tune = [

      datasets

      huggingface-hub

      peft

      # trl

    ];

    ggml = [

      # ctransformers

    ];

    gptq = [

      # auto-gptq

    ]; # ++ autogptq.optional-dependencies.triton;

    grpc = [ bentoml ] ++ bentoml.optional-dependencies.grpc;

    mpt = [ triton ];

    openai = [

      openai

      tiktoken

    ] ++ openai.optional-dependencies.datalib;

    playground = [

      ipython

      jupyter

      jupytext

      nbformat

      notebook

    ];

    starcoder = [ bitsandbytes ];

    vllm = [

      # vllm

    ];

    full =

      with optional-dependencies;

      (

        agents

        ++ awq

        ++ baichuan

        ++ chatglm

        ++ ctranslate

        ++ falcon

        ++ fine-tune

        ++ ggml

        ++ gptq

        ++ mpt

        # disambiguate between derivation input and passthru field

        ++ optional-dependencies.openai

        ++ playground

        ++ starcoder

        ++ vllm

      );

    all = optional-dependencies.full;

  };



  nativeCheckInputs = [

    docker

    hypothesis

    pytest-mock

    pytest-randomly

    pytest-rerunfailures

    pytest-xdist

    pytestCheckHook

    syrupy

  ];



  preCheck = ''

    export HOME=$TMPDIR

    # skip GPUs test on CI

    export GITHUB_ACTIONS=1

    # disable hypothesis' deadline

    export CI=1

  '';



  disabledTestPaths = [

    # require network access

    "tests/models"

  ];



  disabledTests = [

    # incompatible with recent TypedDict

    # https://github.com/bentoml/OpenLLM/blob/f3fd32d596253ae34c68e2e9655f19f40e05f666/openllm-python/tests/configuration_test.py#L18-L21

    "test_missing_default"

  ];



  pythonImportsCheck = [ "openllm" ];



  meta = with lib; {

    description = "Operating LLMs in production";

    homepage = "https://github.com/bentoml/OpenLLM/tree/main/openllm-python";

    changelog = "https://github.com/bentoml/OpenLLM/blob/${src.rev}/CHANGELOG.md";

    license = licenses.asl20;

    maintainers = with maintainers; [

      happysalada

      natsukium

    ];

  };

}