Hub Python Library ################## 安装 ==== :: pip install 'huggingface_hub["cli"]' pip install --upgrade huggingface_hub[inference] // async http (or use pip install aiohttp) 基本操作 ======== Login ----- cli:: huggingface-cli login # or using an environment variable huggingface-cli login --token $HUGGINGFACE_TOKEN python:: from huggingface_hub import login login() Create a repository ------------------- create a private repo:: from huggingface_hub import HfApi api = HfApi() api.create_repo(repo_id="super-cool-model") create a private repo:: from huggingface_hub import HfApi api = HfApi() api.create_repo(repo_id="super-cool-model", private=True) Upload files ------------ :: from huggingface_hub import HfApi api = HfApi() api.upload_file( path_or_fileobj="/home/lysandre/dummy-test/README.md", path_in_repo="README.md", repo_id="lysandre/test-model", ) 环境变量 ======== 预训练模型本地缓存地址:: 1. default: HUGGINGFACE_HUB_CACHE or TRANSFORMERS_CACHE. 2. HF_HOME 3. XDG_CACHE_HOME + /huggingface 默认: ~/.cache/huggingface/hub 通用参数:: HF_ENDPOINT: https://huggingface.co HF_HOME: ~/.cache/huggingface HUGGINGFACE_HUB_CACHE: ~/.cache/huggingface/hub HUGGINGFACE_ASSETS_CACHE: ~/.cache/huggingface/assets HUGGING_FACE_HUB_TOKEN: $HF_HOME/token HF_HUB_OFFLINE:: 如果设置，则在尝试获取文件时不会进行任何HTTP调用 Offline mode 离线模式 ===================== 启用离线模式:: 设置环境变量 TRANSFORMERS_OFFLINE=1 设置环境变量 HF_DATASETS_OFFLINE=1: Add Datasets to your offline training workflow 示例:: python examples/pytorch/translation/run_translation.py --model_name_or_path t5-small \ --dataset_name wmt16 --dataset_config ro-en ... 在脱机实例中运行此相同程序： HF_DATASETS_OFFLINE=1 TRANSFORMERS_OFFLINE=1 \ python examples/pytorch/translation/run_translation.py --model_name_or_path t5-small \ --dataset_name wmt16 --dataset_config ro-en ... 获取模型和分词器以离线使用 -------------------------- 1. 在 `Model Hub `_ 直接下载文件 2. Use the `PreTrainedModel.from_pretrained()` and `PreTrainedModel.save_pretrained()` workflow:: a. Download your files ahead of time with `PreTrainedModel.from_pretrained()`: from transformers import AutoTokenizer, AutoModelForSeq2SeqLM tokenizer = AutoTokenizer.from_pretrained("bigscience/T0_3B") model = AutoModelForSeq2SeqLM.from_pretrained("bigscience/T0_3B") b. Save your files to a specified directory with `PreTrainedModel.save_pretrained()`: tokenizer.save_pretrained("./your/path/bigscience_t0") model.save_pretrained("./your/path/bigscience_t0") c. Now when you’re offline, reload your files with `PreTrainedModel.from_pretrained()` from the specified directory: tokenizer = AutoTokenizer.from_pretrained("./your/path/bigscience_t0") model = AutoModel.from_pretrained("./your/path/bigscience_t0") 3. Programmatically download files with the huggingface_hub library:: a. Install the huggingface_hub library in your virtual environment: python -m pip install huggingface_hub b. Use the `hf_hub_download()` function to download a file to a specific path. from huggingface_hub import hf_hub_download hf_hub_download(repo_id="bigscience/T0_3B", filename="config.json", cache_dir="./your/path/bigscience_t0") c. Once your file is downloaded and locally cached, specify it’s local path to load and use it: from transformers import AutoConfig config = AutoConfig.from_pretrained("./your/path/bigscience_t0/config.json") Download files from the Hub =========================== Download and cache a single file:: >>> from huggingface_hub import hf_hub_download >>> hf_hub_download(repo_id="lysandre/arxiv-nlp", filename="config.json") '/root/.cache/huggingface/hub/models--lysandre--arxiv-nlp/snapshots/894a...fade/config.json' # Download from a dataset hf_hub_download(repo_id="google/fleurs", filename="fleurs.py", repo_type="dataset") # Download from the `v1.0` tag hf_hub_download(repo_id="lysandre/arxiv-nlp", filename="config.json", revision="v1.0") Download and cache an entire repository:: >>> from huggingface_hub import snapshot_download >>> snapshot_download(repo_id="lysandre/arxiv-nlp") '/home/lysandre/.cache/huggingface/hub/models--lysandre--arxiv-nlp/snapshots/894a...fade' Download files to a local folder:: 在使用hf_hub_download() and snapshot_download()函数时可以借用如下参数把文件存储到指定目录: cache_dir: filename: local_dir: local_dir_use_symlinks=True Example: filename="data/train.csv" local_dir="path/to/folder" => the returned filepath will be "path/to/folder/data/train.csv". Upload files to the Hub ======================= Upload a file:: from huggingface_hub import HfApi api = HfApi() api.upload_file( path_or_fileobj="/path/to/local/folder/README.md", path_in_repo="README.md", repo_id="username/test-dataset", repo_type="dataset", ) Upload a folder:: from huggingface_hub import HfApi api = HfApi() api.upload_folder( folder_path="/path/to/local/space", repo_id="username/my-cool-space", repo_type="space", ) Non-blocking upload(使用 run_as_future 参数):: >> from huggingface_hub import HfApi >> api = HfApi() >> future = api.upload_folder( # Upload in the background (non-blocking action) .. repo_id="username/my-model", .. folder_path="checkpoints-001", .. run_as_future=True, .. ) >> future Future(...) >> future.done() False >> future.result() # Wait for the upload to complete (blocking action) ... Upload a folder by chunks:: upload_folder( folder_path="local/checkpoints", repo_id="username/my-dataset", repo_type="dataset", multi_commits=True, multi_commits_verbose=True, ) HfFileSystem ============ * Interact with the Hub through the Filesystem API Usage ----- :: from huggingface_hub import HfFileSystem fs = HfFileSystem() # List all files in a directory fs.ls("datasets/my-username/my-dataset-repo/data", detail=False) # List all ".csv" files in a repo fs.glob("datasets/my-username/my-dataset-repo/**.csv") # Read a remote file with fs.open("datasets/my-username/my-dataset-repo/data/train.csv", "r") as f: train_data = f.readlines() # Read the content of a remote file as a string train_data = fs.read_text("datasets/my-username/my-dataset-repo/data/train.csv", revision="dev") # Write a remote file with fs.open("datasets/my-username/my-dataset-repo/data/validation.csv", "w") as f: f.write("text,label") f.write("Fantastic movie!,good") Integrations ------------ 文件格式:: hf://[][@]/ Reading/writing a Pandas DataFrame from/to a Hub repository:: import pandas as pd # Read a remote CSV file into a dataframe df = pd.read_csv("hf://datasets/my-username/my-dataset-repo/train.csv") # Write a dataframe to a remote CSV file df.to_csv("hf://datasets/my-username/my-dataset-repo/test.csv") Querying (remote) Hub files with DuckDB:: from huggingface_hub import HfFileSystem import duckdb fs = HfFileSystem() duckdb.register_filesystem(fs) # Query a remote file and get the result back as a dataframe fs_query_file = "hf://datasets/my-username/my-dataset-repo/data_dir/data.parquet" df = duckdb.query(f"SELECT * FROM '{fs_query_file}' LIMIT 10").df() Authentication -------------- :: from huggingface_hub import HfFileSystem fs = HfFileSystem(token=token) Repository ========== Repo creation and deletion -------------------------- create_repo():: >> from huggingface_hub import create_repo >> create_repo("lysandre/test-model") 'https://huggingface.co/lysandre/test-model' >> create_repo("lysandre/test-dataset", repo_type="dataset") >> create_repo("lysandre/test-private", private=True) delete_repo():: delete_repo(repo_id="lysandre/my-corrupted-dataset", repo_type="dataset") Clone a repository (only for Spaces):: >> from huggingface_hub import duplicate_space >> duplicate_space("multimodalart/dreambooth-training", private=False) RepoUrl('https://huggingface.co/spaces/nateraw/dreambooth-training',...) Other OPs --------- :: from huggingface_hub import create_branch, create_tag from huggingface_hub import list_repo_refs from huggingface_hub import update_repo_visibility from huggingface_hub import move_repo from huggingface_hub import whoami from huggingface_hub import InferenceClient Manage a local copy of your repository -------------------------------------- Use a local repository:: from huggingface_hub import Repository repo = Repository(local_dir="//") Clone:: from huggingface_hub import Repository repo = Repository(local_dir="w2v2", clone_from="facebook/wav2vec2-large-960h-lv60") repo = Repository(local_dir="huggingface-hub", clone_from="https://huggingface.co/facebook/wav2vec2-large-960h-lv60") Inference ========= * Run Inference on servers * InferenceClient(python): https://huggingface.co/docs/huggingface_hub/v0.16.3/en/package_reference/inference_client * Inference Endpoints: https://huggingface.co/docs/inference-endpoints/index * Inference API: https://huggingface.co/docs/api-inference/index * JS client: https://huggingface.co/docs/huggingface.js/inference/README There are several services you can connect to:: Inference API: a service that allows you to run accelerated inference on Hugging Face’s infrastructure for free. This service is a fast way to get started, test different models, and prototype AI products. Inference Endpoints: a product to easily deploy models to production. Inference is run by Hugging Face in a dedicated, fully managed infrastructure on a cloud provider of your choice. 表格:currently supported tasks:: +------------+----------------------------------+----+----------------------------------+ | Domain | Task | Sup| Method | +============+==================================+====+==================================+ | Audio | Audio Classification | ✅ | audio_classification() | +------------+----------------------------------+----+----------------------------------+ | | Automatic Speech Recognition | ✅ | automatic_speech_recognition() | +------------+----------------------------------+----+----------------------------------+ | | Text-to-Speech | ✅ | text_to_speech() | +------------+----------------------------------+----+----------------------------------+ | CV | Image Classification | ✅ | image_classification() | +------------+----------------------------------+----+----------------------------------+ | | Image Segmentation | ✅ | image_segmentation() | +------------+----------------------------------+----+----------------------------------+ | | Image-to-Image | ✅ | image_to_image() | +------------+----------------------------------+----+----------------------------------+ | | Image-to-Text | ✅ | image_to_text() | +------------+----------------------------------+----+----------------------------------+ | | Object Detection | | | +------------+----------------------------------+----+----------------------------------+ | | Text-to-Image | ✅ | text_to_image() | +------------+----------------------------------+----+----------------------------------+ | | Zero-Shot-Image-Classification | ✅ | zero_shot_image_classification() | +------------+----------------------------------+----+----------------------------------+ | Multimodal | Documentation Question Answering | | | +------------+----------------------------------+----+----------------------------------+ | | Visual Question Answering | | | +------------+----------------------------------+----+----------------------------------+ | NLP | Conversational | ✅ | conversational() | +------------+----------------------------------+----+----------------------------------+ | | Feature Extraction | ✅ | feature_extraction() | +------------+----------------------------------+----+----------------------------------+ | | Fill Mask | | | +------------+----------------------------------+----+----------------------------------+ | | Question Answering | | | +------------+----------------------------------+----+----------------------------------+ | | Sentence Similarity | ✅ | sentence_similarity() | +------------+----------------------------------+----+----------------------------------+ | | Summarization | ✅ | summarization() | +------------+----------------------------------+----+----------------------------------+ | | Table Question Answering | | | +------------+----------------------------------+----+----------------------------------+ | | Text Classification | | | +------------+----------------------------------+----+----------------------------------+ | | Text Generation | ✅ | text_generation() | +------------+----------------------------------+----+----------------------------------+ | | Token Classification | | | +------------+----------------------------------+----+----------------------------------+ | | Translation | | | +------------+----------------------------------+----+----------------------------------+ | | Zero Shot Classification | | | +------------+----------------------------------+----+----------------------------------+ | Tabular | Tabular Classification | | | +------------+----------------------------------+----+----------------------------------+ | | Tabular Regression | | | +------------+----------------------------------+----+----------------------------------+ Inference Client ---------------- 生成client命令:: from huggingface_hub import InferenceClient # default token client = InferenceClient( # Authentication client = InferenceClient(token='hf_.....') # Using a specific model client = InferenceClient(model="prompthero/openjourney-v4") # Using a specific URL client = InferenceClient(model="https://uu149rez6gw9ehej.eu-west-1.aws.endpoints.huggingface.cloud/deepfloyd-if") 使用示例:: from huggingface_hub import InferenceClient client = InferenceClient() image = client.image_to_image("cat.jpg", prompt="turn the cat into a tiger") image.save("tiger.jpg") .. note:: 上面只是其中一个示例，`这儿 `_ 有各种使用示例 Custom requests ^^^^^^^^^^^^^^^ :: from huggingface_hub import InferenceClient client = InferenceClient() response = client.post( json={"inputs": "An astronaut riding a horse on the moon."}, model="stabilityai/stable-diffusion-2-1" ) response.content # raw bytes Async Inference Client ---------------------- 安装:: pip install --upgrade huggingface_hub[inference] # or # pip install aiohttp 生成async client命令:: from huggingface_hub import AsyncInferenceClient client = AsyncInferenceClient(token='hf_.....') InferenceAPI ------------ 示例:: from huggingface_hub.inference_api import InferenceApi # Mask-fill example inference = InferenceApi("bert-base-uncased") inference(inputs="The goal of life is [MASK].") # Question Answering example inference = InferenceApi("deepset/roberta-base-squad2") inputs = { "question": "What's my name?", "context": "My name is Clara and I live in Berkeley.", } inference(inputs) # Zero-shot example inference = InferenceApi("typeform/distilbert-base-uncased-mnli") inputs = "Hi, I recently bought a device from your company but it is not working as advertised and I would like to get reimbursed!" params = {"candidate_labels": ["refund", "legal", "faq"]} inference(inputs, params) # Overriding configured task inference = InferenceApi("bert-base-uncased", task="feature-extraction") # Text-to-image # Using a specific model client = InferenceClient(model="prompthero/openjourney-v4") # Return as raw response to parse the output yourself inference = InferenceApi("mio/amadeus") response = inference("hello world", raw_response=True) response.headers response.content # raw bytes from server Cache ===== Understand caching ------------------ * The Hugging Face Hub cache-system is designed to be the central cache shared across libraries that depend on the Hub. The caching system is designed as follows:: ├─ ├─ ├─ 示例: ├─ models--julien-c--EsperBERTo-small ├─ models--lysandrejik--arxiv-nlp ├─ models--bert-base-cased ├─ datasets--glue ├─ datasets--huggingface--DataMeasurementsFiles ├─ spaces--dalle-mini--dalle-mini * is customizable with the ``cache_dir`` argument on all methods, or by specifying either ``HF_HOME`` or ``HUGGINGFACE_HUB_CACHE`` environment variable. Caching assets -------------- * 除了从 Hub 缓存文件外，下游库通常需要缓存与 HF 相关但未直接被 ``huggingface_hub`` 处理的其他文件 （例如：从 GitHub 下载的文件、预处理的数据、日志,...） * 为了缓存这些文件，调用 assets 使用 huggingface_hub 工具的 cached_assets_path() 方法来管理缓存:: from huggingface_hub import cached_assets_path assets_path = cached_assets_path( library_name="datasets", namespace="SQuAD", subfolder="download" ) # Do anything you like in your assets folder ! something_path = assets_path / "something.json" Scan your cache --------------- Scan cache from the terminal ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ➜ 示例:: $ huggingface-cli scan-cache REPO ID REPO TYPE SIZE ON DISK LAST_MODIFIED REFS LOCAL PATH ---------------- --------- ------------ ------------- ------------------- ------------- glue dataset 116.3K 4 days ago 2.4.0, main, 1.17.0 ~/.cache/huggingface/hub/datasets--glue google/fleurs dataset 64.9M 1 week ago refs/pr/1, main ~/.cache/huggingface/hub/datasets--google--fleurs Jean-Baptiste/ner model 441.0M 16 hours ago main ~/.cache/huggingface/hub/models--Jean-Baptiste--ner bert-base-cased model 1.9G 2 years ago ~/.cache/huggingface/hub/models--bert-base-cased t5-base model 10.1K 3 months ago main ~/.cache/huggingface/hub/models--t5-base t5-small model 970.7M 3 days ago refs/pr/1, main ~/.cache/huggingface/hub/models--t5-small Done in 0.0s. Scanned 6 repo(s) for a total of 3.4G. Got 1 warning(s) while scanning. Use -vvv to print details. Example:: ➜ huggingface-cli scan-cache -v ➜ eval "huggingface-cli scan-cache -v" | grep "t5-small" Scan cache from Python ^^^^^^^^^^^^^^^^^^^^^^ * 具体参见: https://huggingface.co/docs/huggingface_hub/v0.16.3/en/package_reference/cache#huggingface_hub.scan_cache_dir * 与命令 ``huggingface-cli scan-cache`` 对应的python源码:: from huggingface_hub import scan_cache_dir hf_cache_info = scan_cache_dir() Clean your cache ---------------- Clean cache from the terminal ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Using the TUI (Terminal User Interface):: // 安装 pip install huggingface_hub["cli"] // 使用 huggingface-cli delete-cache huggingface-cli delete-cache --dir ~/.cache/huggingface/hub Clean cache from Python ^^^^^^^^^^^^^^^^^^^^^^^ :: from huggingface_hub import scan_cache_dir delete_strategy = scan_cache_dir().delete_revisions( "81fd1d6e7847c99f5862c9fb81387956d99ec7aa" "e2983b237dccf3ab4937c97fa717319a9ca1a96d", "6c0e6080953db56375760c0471a8c5f2929baf11", ) print("Will free " + delete_strategy.expected_freed_size_str) delete_strategy.execute() Model Card ========== Load a Model Card from the Hub ------------------------------ :: from huggingface_hub import ModelCard card = ModelCard.load('nateraw/vit-base-beans') => card.data: Returns a ModelCardData instance with the model card’s metadata. Call .to_dict() on this instance to get the representation as a dictionary. card.text: Returns the text of the card, excluding the metadata header. card.content: Returns the text content of the card, including the metadata header. => card.data.to_dict()['language'] card.data.to_dict()['license'] ... Create Model Cards ------------------ 很多种，最简单的方式:: content = """ --- language: en license: mit --- # My Model Card """ card = ModelCard(content) Share Model Cards ----------------- create a new repo called ‘hf-hub-modelcards-pr-test’:: from huggingface_hub import whoami, create_repo user = whoami()['name'] repo_id = f'{user}/hf-hub-modelcards-pr-test' url = create_repo(repo_id, exist_ok=True) create a card from the default template:: card_data = ModelCardData(language='en', license='mit', library_name='keras') card = ModelCard.from_template( card_data, model_id='my-cool-model', model_description="this model does this and that", developers="Nate Raw", repo="https://github.com/huggingface/huggingface_hub", ) push that up to the hub:: >> card.push_to_hub(repo_id) or push a card as a pull request >> card.push_to_hub(repo_id, create_pr=True) Include Evaluation Results ^^^^^^^^^^^^^^^^^^^^^^^^^^ :: card_data = ModelCardData( language='en', license='mit', model_name='my-cool-model', eval_results = EvalResult( task_type='image-classification', dataset_type='beans', dataset_name='Beans', metric_type='accuracy', metric_value=0.7 ) ) card = ModelCard.from_template(card_data) print(card.data) ``card.data`` should look like this:: language: en license: mit model-index: - name: my-cool-model results: - task: type: image-classification dataset: name: Beans type: beans metrics: - type: accuracy value: 0.7 Integrate Library ================= There are four main ways to integrate a library with the Hub:: 1. Push to Hub: implement a method to upload a model to the Hub. This includes the model weights, model card and necessary data(for example, training logs). This method is often called `push_to_hub`(). 2. Download from Hub: implement a method to load a model from the Hub. The method should download the model configuration/weights and load the model. This method is often called `from_pretrained` or `load_from_hub`(). 3. Inference API: use our servers to run inference on models supported by your library for free. 4. Widgets: display a widget on the landing page of your models on the Hub. It allows users to quickly try a model from the browser. A flexible approach: helpers ---------------------------- * The first approach to integrate a library to the Hub is to actually implement the ``push_to_hub`` and ``from_pretrained`` methods by yourself. from_pretrained:: def from_pretrained(model_id: str) -> MyModelClass: # Download model from Hub cached_model = hf_hub_download( repo_id=repo_id, filename="model.pkl", library_name="fastai", library_version=get_fastai_version(), ) # Load model return load_model(cached_model) push_to_hub:: def push_to_hub(model: MyModelClass, repo_name: str) -> None: api = HfApi() # Create repo if not existing yet and get the associated repo_id repo_id = api.create_repo(repo_name, exist_ok=True) # Save all files in a temporary directory and push them in a single commit with TemporaryDirectory() as tmpdir: tmpdir = Path(tmpdir) # Save weights save_model(model, tmpdir / "model.safetensors") # Generate model card card = generate_model_card(model) (tmpdir / "README.md").write_text(card) # Save logs # Save figures # Save evaluation metrics # ... # Push to hub return api.upload_folder(repo_id=repo_id, folder_path=tmpdir) Limitations ^^^^^^^^^^^ when loading files from the Hub, it is common to offer parameters like:: token: to download from a private repo revision: to download from a specific branch cache_dir: to cache files in a specific directory ... when pushing models, similar parameters are supported:: commit_message: custom commit message private: create a private repo if missing create_pr: create a PR instead of pushing to main ... .. note:: All of these parameters can be added to the implementations we saw above and passed to the huggingface_hub methods. However, if a parameter changes or a new feature is added, you will need to update your package. Supporting those parameters also means more documentation to maintain on your side. A more complex approach: class inheritance ------------------------------------------ * A Mixin is a class that is meant to extend an existing class with a set of specific features using multiple inheritance. * Example: huggingface_hub provides its own mixin, the ModelHubMixin. * The ``ModelHubMixin`` class implements 3 public methods (``push_to_hub``, ``save_pretrained`` and ``from_pretrained``). 步骤:: 1. Make your Model class inherit from ModelHubMixin. 2. Implement the private methods: _save_pretrained() _from_pretrained() A concrete example: PyTorch ^^^^^^^^^^^^^^^^^^^^^^^^^^^ How to use it """"""""""""" 示例:: import torch import torch.nn as nn from huggingface_hub import PyTorchModelHubMixin # 1. Define your Pytorch model exactly the same way you are used to class MyModel(nn.Module, PyTorchModelHubMixin): # multiple inheritance def __init__(self): super().__init__() self.param = nn.Parameter(torch.rand(3, 4)) self.linear = nn.Linear(4, 5) def forward(self, x): return self.linear(x + self.param) model = MyModel() # 2. (optional) Save model to local directory model.save_pretrained("path/to/my-awesome-model") # 3. Push model weights to the Hub model.push_to_hub("my-awesome-model") # 4. Initialize model from the Hub model = MyModel.from_pretrained("username/my-awesome-model") Implementation """""""""""""" 1. First, inherit your class from ModelHubMixin:: from huggingface_hub import ModelHubMixin class PyTorchModelHubMixin(ModelHubMixin): (...) 2. Implement the ``_save_pretrained`` method:: from huggingface_hub import ModelCard, ModelCardData class PyTorchModelHubMixin(ModelHubMixin): (...) def _save_pretrained(self, save_directory: Path): """Generate Model Card and save weights from a Pytorch model to a local directory.""" model_card = ModelCard.from_template( card_data=ModelCardData( license='mit', library_name="pytorch", ... ), model_summary=..., model_type=..., ... ) (save_directory / "README.md").write_text(str(model)) torch.save(obj=self.module.state_dict(), f=save_directory / "pytorch_model.bin") 3. Implement the ``_from_pretrained`` method:: class PyTorchModelHubMixin(ModelHubMixin): (...) @classmethod # Must be a classmethod! def _from_pretrained( cls, *, model_id: str, revision: str, cache_dir: str, force_download: bool, proxies: Optional[Dict], resume_download: bool, local_files_only: bool, token: Union[str, bool, None], map_location: str = "cpu", # additional argument strict: bool = False, # additional argument **model_kwargs, ): """Load Pytorch pretrained weights and return the loaded model.""" if os.path.isdir(model_id): # Can either be a local directory print("Loading weights from local directory") model_file = os.path.join(model_id, "pytorch_model.bin") else: # Or a model on the Hub model_file = hf_hub_download( # Download from the hub, passing same input args repo_id=model_id, filename="pytorch_model.bin", revision=revision, cache_dir=cache_dir, force_download=force_download, proxies=proxies, resume_download=resume_download, token=token, local_files_only=local_files_only, ) # Load model and return - custom logic depending on your framework model = cls(**model_kwargs) state_dict = torch.load(model_file, map_location=torch.device(map_location)) model.load_state_dict(state_dict, strict=strict) model.eval() return model Quick comparison ---------------- 1. 用户体验:: - helpers:需要加载并推送模型,代码较复杂 - ModelHubMixin:直接继承模型类,用from_pretrained和push_to_hub方法,更简单直观 2. 灵活性:: - helpers:非常灵活,可以完全自定义实现 - ModelHubMixin:不太灵活,需要框架有模型类 3. 维护:: - helpers:需要自行维护配置、新特性等,解决用户报告的issue - ModelHubMixin:由huggingface_hub包处理大部分交互,维护更少 4. 文档/类型注解:: - helpers:需要手动编写文档和注解 - ModelHubMixin:部分由huggingface_hub自动生成