From 980126b83a056db380e62cf53ddcb40aeac53de9 Mon Sep 17 00:00:00 2001 From: Yzc216 <101054010+Yzc216@users.noreply.github.com> Date: Thu, 24 Jul 2025 17:42:09 +0800 Subject: [PATCH] [Feature] multi source download (#3005) * multi-source download * multi-source download * huggingface download revision * requirement * style * add revision arg * test * pre-commit * Change default download * change requirements.txt * modify English Documentation * documentation --- docs/supported_models.md | 7 ++++--- docs/zh/supported_models.md | 11 ++++++----- fastdeploy/envs.py | 2 +- fastdeploy/utils.py | 36 +++++++++++++++++++++--------------- requirements.txt | 1 - 5 files changed, 32 insertions(+), 25 deletions(-) diff --git a/docs/supported_models.md b/docs/supported_models.md index 0fac24f7e..c6bb969ae 100644 --- a/docs/supported_models.md +++ b/docs/supported_models.md @@ -1,14 +1,15 @@ # Supported Models -FastDeploy currently supports the following models, which can be downloaded via three methods: +FastDeploy currently supports the following models, which can be downloaded automatically during FastDeploy deployment.Specify the ``model`` parameter as the model name in the table below to automatically download model weights (all supports resumable downloads). The following three download sources are supported: -- 1. During FastDeploy deployment, specify the ``model`` parameter as the model name in the table below to automatically download model weights from AIStudio (supports resumable downloads) +- 1. Search for corresponding Paddle-version ERNIE models on [AIStudio/PaddlePaddle](https://aistudio.baidu.com/modelsoverview), e.g., `ERNIE-4.5-0.3B-Paddle` - 2. Download Paddle-version ERNIE models from [HuggingFace/baidu/models](https://huggingface.co/baidu/models), e.g., `baidu/ERNIE-4.5-0.3B-Paddle` - 3. Search for corresponding Paddle-version ERNIE models on [ModelScope/PaddlePaddle](https://www.modelscope.cn/models?name=PaddlePaddle&page=1&tabKey=task), e.g., `ERNIE-4.5-0.3B-Paddle` -For the first method (auto-download), the default download path is ``~/`` (user home directory). Users can modify this path by setting the ``FD_MODEL_CACHE`` environment variable, e.g.: +When using automatic download, the default download source is AIStudio. Users can modify the default download source by setting the ``FD_MODEL_SOURCE`` environment variable, which can be set to “AISTUDIO”, ‘MODELSCOPE’ or “HUGGINGFACE”. The default download path is ``~/`` (i.e., the user's home directory). Users can modify the default download path by setting the ``FD_MODEL_CACHE`` environment variable, e.g.: ```bash +export FD_MODEL_SOURCE=AISTUDIO # "AISTUDIO", "MODELSCOPE" or "HUGGINGFACE" export FD_MODEL_CACHE=/ssd1/download_models ``` diff --git a/docs/zh/supported_models.md b/docs/zh/supported_models.md index 562640de9..f7b95541f 100644 --- a/docs/zh/supported_models.md +++ b/docs/zh/supported_models.md @@ -1,14 +1,15 @@ # 支持模型列表 -FastDeploy目前支持模型列表如下，以下模型提供如下3种下载方式， +FastDeploy目前支持模型列表如下，在FastDeploy部署时，指定 ``model``参数为如下表格中的模型名，即可自动下载模型权重（均支持断点续传），支持如下3种下载源， -- 1. 在FastDeploy部署时，指定 ``model``参数为如下表格中的模型名，即可自动从AIStudio下载模型权重（支持断点续传） -- 2. [HuggingFace/baidu/models](https://huggingface.co/baidu/models) 下载Paddle后缀ERNIE模型，如baidu/ERNIE-4.5-0.3B-Paddle -- 3. [ModelScope/PaddlePaddle](https://www.modelscope.cn/models?name=PaddlePaddle&page=1&tabKey=task) 搜索相应Paddle后缀ERNIE模型，如ERNIE-4.5-0.3B-Paddle +- 1. [AIStudio/PaddlePaddle](https://aistudio.baidu.com/modelsoverview) 搜索相应Paddle后缀ERNIE模型，如ERNIE-4.5-0.3B-Paddle +- 2. [ModelScope/PaddlePaddle](https://www.modelscope.cn/models?name=PaddlePaddle&page=1&tabKey=task) 搜索相应Paddle后缀ERNIE模型，如ERNIE-4.5-0.3B-Paddle +- 3. [HuggingFace/baidu/models](https://huggingface.co/baidu/models) 下载Paddle后缀ERNIE模型，如baidu/ERNIE-4.5-0.3B-Paddle -其中第一种方式自动下载时，默认下载路径为 ``~/``(即用户主目录)，用户可以通过配置环境变量 ``FD_MODEL_CACHE``修改默认下载的路径，例如 +使用自动下载时，默认从AIStudio下载，用户可以通过配置环境变量 ``FD_MODEL_SOURCE``修改默认下载来源，可取值"AISTUDIO"，"MODELSCOPE"或"HUGGINGFACE"；默认下载路径为 ``~/``(即用户主目录)，用户可以通过配置环境变量 ``FD_MODEL_CACHE``修改默认下载的路径，例如 ``` +export FD_MODEL_SOURCE=AISTUDIO # "AISTUDIO", "MODELSCOPE" or "HUGGINGFACE" export FD_MODEL_CACHE=/ssd1/download_models ``` diff --git a/fastdeploy/envs.py b/fastdeploy/envs.py index 964e2f078..901ef3f5a 100644 --- a/fastdeploy/envs.py +++ b/fastdeploy/envs.py @@ -31,7 +31,7 @@ environment_variables: dict[str, Callable[[], Any]] = { # Number of days to keep fastdeploy logs. "FD_LOG_BACKUP_COUNT": lambda: os.getenv("FD_LOG_BACKUP_COUNT", "7"), # Model download source, can set "AISTUDIO", "MODELSCOPE" or "HUGGINGFACE". - "FD_MODEL_SOURCE": lambda: os.getenv("FD_MODEL_SOURCE", "MODELSCOPE"), + "FD_MODEL_SOURCE": lambda: os.getenv("FD_MODEL_SOURCE", "AISTUDIO"), # Model download cache directory. "FD_MODEL_CACHE": lambda: os.getenv("FD_MODEL_CACHE", None), # Maximum number of stop sequences. diff --git a/fastdeploy/utils.py b/fastdeploy/utils.py index f9cfe9d40..0ceba665a 100644 --- a/fastdeploy/utils.py +++ b/fastdeploy/utils.py @@ -32,8 +32,6 @@ from typing import Literal, TypeVar, Union import requests import yaml from aistudio_sdk.snapshot_download import snapshot_download as aistudio_download -from huggingface_hub._snapshot_download import snapshot_download as huggingface_download -from modelscope.hub.snapshot_download import snapshot_download as modelscope_download from tqdm import tqdm from typing_extensions import TypeIs, assert_never @@ -496,25 +494,14 @@ def none_or_str(value): def retrive_model_from_server(model_name_or_path, revision="master"): """ - Download pretrained model from MODELSCOPE, AIStudio or HUGGINGFACE automatically + Download pretrained model from AIStudio, MODELSCOPE or HUGGINGFACE automatically """ if os.path.exists(model_name_or_path): return model_name_or_path model_source = envs.FD_MODEL_SOURCE local_path = envs.FD_MODEL_CACHE repo_id = model_name_or_path - if model_source == "MODELSCOPE": - try: - if repo_id.lower().strip().startswith("baidu"): - repo_id = "PaddlePaddle" + repo_id.strip()[5:] - if local_path is None: - local_path = f'{os.getenv("HOME")}' - local_path = f"{local_path}/{repo_id}/{revision}" - modelscope_download(repo_id=repo_id, revision=revision, local_dir=local_path) - model_name_or_path = local_path - except Exception: - raise Exception(f"The setting model_name_or_path:{model_name_or_path} is not exist.") - elif model_source == "AISTUDIO": + if model_source == "AISTUDIO": try: if repo_id.lower().strip().startswith("baidu"): repo_id = "PaddlePaddle" + repo_id.strip()[5:] @@ -525,8 +512,27 @@ def retrive_model_from_server(model_name_or_path, revision="master"): model_name_or_path = local_path except Exception: raise Exception(f"The setting model_name_or_path:{model_name_or_path} is not exist.") + elif model_source == "MODELSCOPE": + try: + from modelscope.hub.snapshot_download import ( + snapshot_download as modelscope_download, + ) + + if repo_id.lower().strip().startswith("baidu"): + repo_id = "PaddlePaddle" + repo_id.strip()[5:] + if local_path is None: + local_path = f'{os.getenv("HOME")}' + local_path = f"{local_path}/{repo_id}/{revision}" + modelscope_download(repo_id=repo_id, revision=revision, local_dir=local_path) + model_name_or_path = local_path + except Exception: + raise Exception(f"The setting model_name_or_path:{model_name_or_path} is not exist.") elif model_source == "HUGGINGFACE": try: + from huggingface_hub._snapshot_download import ( + snapshot_download as huggingface_download, + ) + if revision == "master": revision = "main" repo_id = model_name_or_path diff --git a/requirements.txt b/requirements.txt index 4717f532a..f9166c8c2 100644 --- a/requirements.txt +++ b/requirements.txt @@ -30,7 +30,6 @@ use-triton-in-paddle crcmod fastsafetensors==0.1.14 msgpack -modelscope opentelemetry-api>=1.24.0 opentelemetry-sdk>=1.24.0 opentelemetry-instrumentation-redis