hugging face tutorial-Chinese translation-pipeline-based reasoning

基于pipeline的推理

Pipeline ()使得使用 Model Hub It becomes very simple to infer any model in various tasks,比如文本生成、Image segmentation and audio classification.Even if you have no experience with a particular pattern or understand the code that supports the model,You can still use them with pipeline一起使用() ！This tutorial will teach you:

• 使用 pipeline ()进行推理.
• Use a specific tokenizer or model.
• 使用pipeline()Perform audio and visual tasks.

查看pipeline()文档,to get supported taska 的完整列表.

pipeline的使用

Although each task has an associated onepipeline(),But use contains all specific taskspipeline的通用pipeline()Abstraction is simpler.Pipeline ()Automatically loads a default model and tokenizer,Ability to infer your tasks.

1. 首先创建一个 pipeline ()and specify an inference task:

from transformers import pipeline
generator = pipeline(task="text-generation")

2. Pass the input text topipeline():

generator("Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone")

if you have more than one input,Pass your input as a list:

generator(
...     [
... "Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone",
... "Nine for Mortal Men, doomed to die, One for the Dark Lord on his dark throne",
...     ]
... )

Any other parameters of the task can also be includedpipeline()中.The text generation task has one with multiple parameters generate ()method to control the output.例如,If you want to generate multiple outputs,请设置 num_return_sequences 参数:

generator(
... "Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone",
...     num_return_sequences=2,
... )

3. Select a model and tokenizer

Pipeline ()接受来自 modelhub any model of.Model Hub There are some marks on it,They allow you to filter the models you wish to use for the task.Once you have selected a suitable model,加载相应的 autoodelfor 和[‘ AutoTokenizer’]类.例如,Loaded for a causal language modeling task automatodelforcausallm 类:

from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("distilgpt2")
model = AutoModelForCausalLM.from_pretrained("distilgpt2")

Create one for your task pipeline () ,and specify your loaded model and tokenizer:

from transformers import pipeline
generator = pipeline(task="text-generation", model=model, tokenizer=tokenizer)

Pass the input text to pipeline ()to generate some text:

generator("Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone")
[{
    'generated_text': 'Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone, Seven for the Dragon-lords (for them to rule in a world ruled by their rulers, and all who live within the realm'}]

音频pipeline

流水线()The flexibility means it can also be extended to audio tasks.

例如,Let's start with John·f·肯尼迪(johnf.Kennedy)著名的“我们选择登月”A short clip of a speech to categorize emotions.在 Model Hub Find an audio classification model for emotion recognition on ,并将其加载到pipeline中() :

from transformers import pipeline
audio_classifier = pipeline(
...     task="audio-classification", model="ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition"... )

Pass the audio file to pipeline() :

audio_classifier("jfk_moon_speech.wav")
[{
    'label': 'calm', 'score': 0.13856211304664612},
 {
    'label': 'disgust', 'score': 0.13148026168346405},
 {
    'label': 'happy', 'score': 0.12635163962841034},
 {
    'label': 'angry', 'score': 0.12439591437578201},
 {
    'label': 'fearful', 'score': 0.12404385954141617}]

愿景pipeline

最后,使用pipeline()Performing vision tasks is virtually the same.

Specify your vision task and pass the image to the classifier.An image image can be a link or a local path to the image.例如,Which feline is shown below？

from transformers import pipeline
vision_classifier = pipeline(task="image-classification")
vision_classifier(
...     images="

"... )
[{
    'label': 'lynx, catamount', 'score': 0.4403027892112732},
 {
    'label': 'cougar, puma, catamount, mountain lion, painter, panther, Felis concolor',
  'score': 0.03433405980467796},
 {
    'label': 'snow leopard, ounce, Panthera uncia',
  'score': 0.032148055732250214},
 {
    'label': 'Egyptian cat', 'score': 0.02353910356760025},
 {
    'label': 'tiger cat', 'score': 0.023034192621707916}]

This article is a translation study of English content,仅学习使用

原文链接