• Stars
    star
    1,912
  • Rank 23,354 (Top 0.5 %)
  • Language
    Jupyter Notebook
  • License
    Apache License 2.0
  • Created almost 2 years ago
  • Updated about 1 month ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

Efficient few-shot learning with Sentence Transformers

🤗 Models & Datasets | 📖 Blog | 📃 Paper

SetFit - Efficient Few-shot Learning with Sentence Transformers

SetFit is an efficient and prompt-free framework for few-shot fine-tuning of Sentence Transformers. It achieves high accuracy with little labeled data - for instance, with only 8 labeled examples per class on the Customer Reviews sentiment dataset, SetFit is competitive with fine-tuning RoBERTa Large on the full training set of 3k examples 🤯!

Compared to other few-shot learning methods, SetFit has several unique features:

  • 🗣 No prompts or verbalisers: Current techniques for few-shot fine-tuning require handcrafted prompts or verbalisers to convert examples into a format that's suitable for the underlying language model. SetFit dispenses with prompts altogether by generating rich embeddings directly from text examples.
  • 🏎 Fast to train: SetFit doesn't require large-scale models like T0 or GPT-3 to achieve high accuracy. As a result, it is typically an order of magnitude (or more) faster to train and run inference with.
  • 🌎 Multilingual support: SetFit can be used with any Sentence Transformer on the Hub, which means you can classify text in multiple languages by simply fine-tuning a multilingual checkpoint.

Installation

Download and install setfit by running:

python -m pip install setfit

If you want the bleeding-edge version, install from source by running:

python -m pip install git+https://github.com/huggingface/setfit.git

Usage

The examples below provide a quick overview on the various features supported in setfit. For more examples, check out the notebooks folder.

Training a SetFit model

setfit is integrated with the Hugging Face Hub and provides two main classes:

  • SetFitModel: a wrapper that combines a pretrained body from sentence_transformers and a classification head from either scikit-learn or SetFitHead (a differentiable head built upon PyTorch with similar APIs to sentence_transformers).
  • SetFitTrainer: a helper class that wraps the fine-tuning process of SetFit.

Here is an end-to-end example using a classification head from scikit-learn:

from datasets import load_dataset
from sentence_transformers.losses import CosineSimilarityLoss

from setfit import SetFitModel, SetFitTrainer, sample_dataset


# Load a dataset from the Hugging Face Hub
dataset = load_dataset("sst2")

# Simulate the few-shot regime by sampling 8 examples per class
train_dataset = sample_dataset(dataset["train"], label_column="label", num_samples=8)
eval_dataset = dataset["validation"]

# Load a SetFit model from Hub
model = SetFitModel.from_pretrained("sentence-transformers/paraphrase-mpnet-base-v2")

# Create trainer
trainer = SetFitTrainer(
    model=model,
    train_dataset=train_dataset,
    eval_dataset=eval_dataset,
    loss_class=CosineSimilarityLoss,
    metric="accuracy",
    batch_size=16,
    num_iterations=20, # The number of text pairs to generate for contrastive learning
    num_epochs=1, # The number of epochs to use for contrastive learning
    column_mapping={"sentence": "text", "label": "label"} # Map dataset columns to text/label expected by trainer
)

# Train and evaluate
trainer.train()
metrics = trainer.evaluate()

# Push model to the Hub
trainer.push_to_hub("my-awesome-setfit-model")

# Download from Hub and run inference
model = SetFitModel.from_pretrained("lewtun/my-awesome-setfit-model")
# Run inference
preds = model(["i loved the spiderman movie!", "pineapple on pizza is the worst 🤮"])

Here is an end-to-end example using SetFitHead:

from datasets import load_dataset
from sentence_transformers.losses import CosineSimilarityLoss

from setfit import SetFitModel, SetFitTrainer, sample_dataset


# Load a dataset from the Hugging Face Hub
dataset = load_dataset("sst2")

# Simulate the few-shot regime by sampling 8 examples per class
train_dataset = sample_dataset(dataset["train"], label_column="label", num_samples=8)
eval_dataset = dataset["validation"]

# Load a SetFit model from Hub
model = SetFitModel.from_pretrained(
    "sentence-transformers/paraphrase-mpnet-base-v2",
    use_differentiable_head=True,
    head_params={"out_features": num_classes},
)

# Create trainer
trainer = SetFitTrainer(
    model=model,
    train_dataset=train_dataset,
    eval_dataset=eval_dataset,
    loss_class=CosineSimilarityLoss,
    metric="accuracy",
    batch_size=16,
    num_iterations=20, # The number of text pairs to generate for contrastive learning
    num_epochs=1, # The number of epochs to use for contrastive learning
    column_mapping={"sentence": "text", "label": "label"} # Map dataset columns to text/label expected by trainer
)

# Train and evaluate
trainer.freeze() # Freeze the head
trainer.train() # Train only the body

# Unfreeze the head and freeze the body -> head-only training
trainer.unfreeze(keep_body_frozen=True)
# or
# Unfreeze the head and unfreeze the body -> end-to-end training
trainer.unfreeze(keep_body_frozen=False)

trainer.train(
    num_epochs=25, # The number of epochs to train the head or the whole model (body and head)
    batch_size=16,
    body_learning_rate=1e-5, # The body's learning rate
    learning_rate=1e-2, # The head's learning rate
    l2_weight=0.0, # Weight decay on **both** the body and head. If `None`, will use 0.01.
)
metrics = trainer.evaluate()

# Push model to the Hub
trainer.push_to_hub("my-awesome-setfit-model")

# Download from Hub and run inference
model = SetFitModel.from_pretrained("lewtun/my-awesome-setfit-model")
# Run inference
preds = model(["i loved the spiderman movie!", "pineapple on pizza is the worst 🤮"])

Based on our experiments, SetFitHead can achieve similar performance as using a scikit-learn head. We use AdamW as the optimizer and scale down learning rates by 0.5 every 5 epochs. For more details about the experiments, please check out here. We recommend using a large learning rate (e.g. 1e-2) for SetFitHead and a small learning rate (e.g. 1e-5) for the body in your first attempt.

Training on multilabel datasets

To train SetFit models on multilabel datasets, specify the multi_target_strategy argument when loading the pretrained model:

Example using a classification head from scikit-learn:

from setfit import SetFitModel

model = SetFitModel.from_pretrained(
    model_id,
    multi_target_strategy="one-vs-rest",
)

This will initialise a multilabel classification head from sklearn - the following options are available for multi_target_strategy:

  • one-vs-rest: uses a OneVsRestClassifier head.
  • multi-output: uses a MultiOutputClassifier head.
  • classifier-chain: uses a ClassifierChain head.

From here, you can instantiate a SetFitTrainer using the same example above, and train it as usual.

Example using the differentiable SetFitHead:

from setfit import SetFitModel

model = SetFitModel.from_pretrained(
    model_id,
    multi_target_strategy="one-vs-rest"
    use_differentiable_head=True,
    head_params={"out_features": num_classes},
)

Note: If you use the differentiable SetFitHead classifier head, it will automatically use BCEWithLogitsLoss for training. The prediction involves a sigmoid after which probabilities are rounded to 1 or 0. Furthermore, the "one-vs-rest" and "multi-output" multi-target strategies are equivalent for the differentiable SetFitHead.

Zero-shot text classification

SetFit can also be applied to scenarios where no labels are available. To do so, create a synthetic dataset of training examples:

from datasets import Dataset
from setfit import get_templated_dataset

candidate_labels = ["negative", "positive"]
train_dataset = get_templated_dataset(candidate_labels=candidate_labels, sample_size=8)

This will create examples of the form "This sentence is {}", where the {} is filled in with one of the candidate labels. From here you can train a SetFit model as usual:

from setfit import SetFitModel, SetFitTrainer

model = SetFitModel.from_pretrained("sentence-transformers/paraphrase-mpnet-base-v2")
trainer = SetFitTrainer(
    model=model,
    train_dataset=train_dataset
)
trainer.train()

We find this approach typically outperforms the zero-shot pipeline in 🤗 Transformers (based on MNLI with Bart), while being 5x faster to generate predictions with.

Running hyperparameter search

SetFitTrainer provides a hyperparameter_search() method that you can use to find good hyperparameters for your data. To use this feature, first install the optuna backend:

python -m pip install setfit[optuna]

To use this method, you need to define two functions:

  • model_init(): A function that instantiates the model to be used. If provided, each call to train() will start from a new instance of the model as given by this function.
  • hp_space(): A function that defines the hyperparameter search space.

Here is an example of a model_init() function that we'll use to scan over the hyperparameters associated with the classification head in SetFitModel:

from setfit import SetFitModel

def model_init(params):
    params = params or {}
    max_iter = params.get("max_iter", 100)
    solver = params.get("solver", "liblinear")
    params = {
        "head_params": {
            "max_iter": max_iter,
            "solver": solver,
        }
    }
    return SetFitModel.from_pretrained("sentence-transformers/paraphrase-albert-small-v2", **params)

Similarly, to scan over hyperparameters associated with the SetFit training process, we can define a hp_space() function as follows:

def hp_space(trial):  # Training parameters
    return {
        "learning_rate": trial.suggest_float("learning_rate", 1e-6, 1e-4, log=True),
        "num_epochs": trial.suggest_int("num_epochs", 1, 5),
        "batch_size": trial.suggest_categorical("batch_size", [4, 8, 16, 32, 64]),
        "seed": trial.suggest_int("seed", 1, 40),
        "num_iterations": trial.suggest_categorical("num_iterations", [5, 10, 20]),
        "max_iter": trial.suggest_int("max_iter", 50, 300),
        "solver": trial.suggest_categorical("solver", ["newton-cg", "lbfgs", "liblinear"]),
    }

Note: In practice, we found num_iterations to be the most important hyperparameter for the contrastive learning process.

The next step is to instantiate a SetFitTrainer and call hyperparameter_search():

from datasets import Dataset
from setfit import SetFitTrainer

dataset = Dataset.from_dict(
            {"text_new": ["a", "b", "c"], "label_new": [0, 1, 2], "extra_column": ["d", "e", "f"]}
        )

trainer = SetFitTrainer(
    train_dataset=dataset,
    eval_dataset=dataset,
    model_init=model_init,
    column_mapping={"text_new": "text", "label_new": "label"},
)
best_run = trainer.hyperparameter_search(direction="maximize", hp_space=hp_space, n_trials=20)

Finally, you can apply the hyperparameters you found to the trainer, and lock in the optimal model, before training for a final time.

trainer.apply_hyperparameters(best_run.hyperparameters, final_model=True)
trainer.train()

Compressing a SetFit model with knowledge distillation

If you have access to unlabeled data, you can use knowledge distillation to compress a trained SetFit model into a smaller version. The result is a model that can run inference much faster, with little to no drop in accuracy. Here's an end-to-end example (see our paper for more details):

from datasets import load_dataset
from sentence_transformers.losses import CosineSimilarityLoss

from setfit import SetFitModel, SetFitTrainer, DistillationSetFitTrainer, sample_dataset

# Load a dataset from the Hugging Face Hub
dataset = load_dataset("ag_news")

# Create a sample few-shot dataset to train the teacher model
train_dataset_teacher = sample_dataset(dataset["train"], label_column="label", num_samples=16)
# Create a dataset of unlabeled examples to train the student
train_dataset_student = dataset["train"].shuffle(seed=0).select(range(500))
# Dataset for evaluation
eval_dataset = dataset["test"]

# Load teacher model
teacher_model = SetFitModel.from_pretrained(
    "sentence-transformers/paraphrase-mpnet-base-v2"
)

# Create trainer for teacher model
teacher_trainer = SetFitTrainer(
    model=teacher_model,
    train_dataset=train_dataset_teacher,
    eval_dataset=eval_dataset,
    loss_class=CosineSimilarityLoss,
)

# Train teacher model
teacher_trainer.train()

# Load small student model
student_model = SetFitModel.from_pretrained("paraphrase-MiniLM-L3-v2")

# Create trainer for knowledge distillation
student_trainer = DistillationSetFitTrainer(
    teacher_model=teacher_model,
    train_dataset=train_dataset_student,
    student_model=student_model,
    eval_dataset=eval_dataset,
    loss_class=CosineSimilarityLoss,
    metric="accuracy",
    batch_size=16,
    num_iterations=20,
    num_epochs=1,
)

# Train student with knowledge distillation
student_trainer.train()

Reproducing the results from the paper

We provide scripts to reproduce the results for SetFit and various baselines presented in Table 2 of our paper. Check out the setup and training instructions in the scripts/ directory.

Developer installation

To run the code in this project, first create a Python virtual environment using e.g. Conda:

conda create -n setfit python=3.9 && conda activate setfit

Then install the base requirements with:

python -m pip install -e '.[dev]'

This will install datasets and packages like black and isort that we use to ensure consistent code formatting.

Formatting your code

We use black and isort to ensure consistent code formatting. After following the installation steps, you can check your code locally by running:

make style && make quality

Project structure

├── LICENSE
├── Makefile        <- Makefile with commands like `make style` or `make tests`
├── README.md       <- The top-level README for developers using this project.
├── notebooks       <- Jupyter notebooks.
├── final_results   <- Model predictions from the paper
├── scripts         <- Scripts for training and inference
├── setup.cfg       <- Configuration file to define package metadata
├── setup.py        <- Make this project pip installable with `pip install -e`
├── src             <- Source code for SetFit
└── tests           <- Unit tests

Related work

Citation

  doi = {10.48550/ARXIV.2209.11055},
  url = {https://arxiv.org/abs/2209.11055},
  author = {Tunstall, Lewis and Reimers, Nils and Jo, Unso Eun Seo and Bates, Luke and Korat, Daniel and Wasserblat, Moshe and Pereg, Oren},
  keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
  title = {Efficient Few-Shot Learning Without Prompts},
  publisher = {arXiv},
  year = {2022},
  copyright = {Creative Commons Attribution 4.0 International}}

More Repositories

1

transformers

🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
Python
125,320
star
2

pytorch-image-models

PyTorch image models, scripts, pretrained weights -- ResNet, ResNeXT, EfficientNet, NFNet, Vision Transformer (ViT), MobileNet-V3/V2, RegNet, DPN, CSPNet, Swin Transformer, MaxViT, CoAtNet, ConvNeXt, and more
Python
28,073
star
3

diffusers

🤗 Diffusers: State-of-the-art diffusion models for image and audio generation in PyTorch and FLAX.
Python
22,776
star
4

datasets

🤗 The largest hub of ready-to-use datasets for ML models with fast, easy-to-use and efficient data manipulation tools
Python
17,530
star
5

peft

🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning.
Python
14,007
star
6

candle

Minimalist ML framework for Rust
Rust
12,686
star
7

tokenizers

💥 Fast State-of-the-Art Tokenizers optimized for Research and Production
Rust
8,286
star
8

trl

Train transformer language models with reinforcement learning.
Python
8,181
star
9

text-generation-inference

Large Language Model Text Generation Inference
Python
7,240
star
10

accelerate

🚀 A simple way to launch, train, and use PyTorch models on almost any device and distributed configuration, automatic mixed precision (including fp8), and easy-to-configure FSDP and DeepSpeed support
Python
7,008
star
11

chat-ui

Open source codebase powering the HuggingChat app
TypeScript
5,586
star
12

deep-rl-class

This repo contains the syllabus of the Hugging Face Deep Reinforcement Learning Course.
MDX
3,541
star
13

alignment-handbook

Robust recipes to align language models with human and AI preferences
Python
3,485
star
14

autotrain-advanced

🤗 AutoTrain Advanced
Python
3,283
star
15

diffusion-models-class

Materials for the Hugging Face Diffusion Models Course
Jupyter Notebook
3,126
star
16

notebooks

Notebooks using the Hugging Face libraries 🤗
Jupyter Notebook
3,114
star
17

distil-whisper

Distilled variant of Whisper for speech recognition. 6x faster, 50% smaller, within 1% word error rate.
Python
2,964
star
18

neuralcoref

✨Fast Coreference Resolution in spaCy with Neural Networks
C
2,806
star
19

knockknock

🚪✊Knock Knock: Get notified when your training ends with only two additional lines of code
Python
2,682
star
20

swift-coreml-diffusers

Swift app demonstrating Core ML Stable Diffusion
Swift
2,377
star
21

safetensors

Simple, safe way to store and distribute tensors
Python
2,347
star
22

optimum

🚀 Accelerate training and inference of 🤗 Transformers and 🤗 Diffusers with easy to use hardware optimization tools
Python
2,086
star
23

awesome-papers

Papers & presentation materials from Hugging Face's internal science day
1,996
star
24

blog

Public repo for HF blog posts
Jupyter Notebook
1,962
star
25

text-embeddings-inference

A blazing fast inference solution for text embeddings models
Rust
1,845
star
26

course

The Hugging Face course on Transformers
MDX
1,832
star
27

evaluate

🤗 Evaluate: A library for easily evaluating machine learning models and datasets.
Python
1,825
star
28

transfer-learning-conv-ai

🦄 State-of-the-Art Conversational AI with Transfer Learning
Python
1,654
star
29

swift-coreml-transformers

Swift Core ML 3 implementations of GPT-2, DistilGPT-2, BERT, and DistilBERT for Question answering. Other Transformers coming soon!
Swift
1,543
star
30

pytorch-openai-transformer-lm

🐥A PyTorch implementation of OpenAI's finetuned transformer language model with a script to import the weights pre-trained by OpenAI
Python
1,464
star
31

cookbook

Open-source AI cookbook
Jupyter Notebook
1,357
star
32

huggingface_hub

All the open source things related to the Hugging Face Hub.
Python
1,311
star
33

Mongoku

🔥The Web-scale GUI for MongoDB
TypeScript
1,289
star
34

huggingface.js

Utilities to use the Hugging Face Hub API
TypeScript
1,193
star
35

hmtl

🌊HMTL: Hierarchical Multi-Task Learning - A State-of-the-Art neural network model for several NLP tasks based on PyTorch and AllenNLP
Python
1,185
star
36

gsplat.js

JavaScript Gaussian Splatting library.
TypeScript
1,114
star
37

llm-vscode

LLM powered development for VSCode
TypeScript
1,060
star
38

datatrove

Freeing data processing from scripting madness by providing a set of platform-agnostic customizable pipeline processing blocks.
Python
1,033
star
39

pytorch-pretrained-BigGAN

🦋A PyTorch implementation of BigGAN with pretrained weights and conversion scripts.
Python
986
star
40

torchMoji

😇A pyTorch implementation of the DeepMoji model: state-of-the-art deep learning model for analyzing sentiment, emotion, sarcasm etc
Python
880
star
41

naacl_transfer_learning_tutorial

Repository of code for the tutorial on Transfer Learning in NLP held at NAACL 2019 in Minneapolis, MN, USA
Python
718
star
42

awesome-huggingface

🤗 A list of wonderful open-source projects & applications integrated with Hugging Face libraries.
698
star
43

optimum-nvidia

Python
680
star
44

nanotron

Minimalistic large language model 3D-parallelism training
Python
661
star
45

dataset-viewer

Lightweight web API for visualizing and exploring any dataset - computer vision, speech, text, and tabular - stored on the Hugging Face Hub
Python
614
star
46

transformers-bloom-inference

Fast Inference Solutions for BLOOM
Python
546
star
47

exporters

Export Hugging Face models to Core ML and TensorFlow Lite
Python
540
star
48

pytorch_block_sparse

Fast Block Sparse Matrices for Pytorch
C++
523
star
49

llm.nvim

LLM powered development for Neovim
Lua
507
star
50

swift-transformers

Swift Package to implement a transformers-like API in Swift
Swift
482
star
51

node-question-answering

Fast and production-ready question answering in Node.js
TypeScript
459
star
52

large_language_model_training_playbook

An open collection of implementation tips, tricks and resources for training large language models
Python
431
star
53

llm-ls

LSP server leveraging LLMs for code completion (and more?)
Rust
416
star
54

llm_training_handbook

An open collection of methodologies to help with successful training of large language models.
Python
385
star
55

swift-chat

Mac app to demonstrate swift-transformers
Swift
375
star
56

tflite-android-transformers

DistilBERT / GPT-2 for on-device inference thanks to TensorFlow Lite with Android demo apps
Java
368
star
57

community-events

Place where folks can contribute to 🤗 community events
Jupyter Notebook
368
star
58

nn_pruning

Prune a model while finetuning or training.
Jupyter Notebook
360
star
59

text-clustering

Easily embed, cluster and semantically label text datasets
Python
335
star
60

speechbox

Python
328
star
61

100-times-faster-nlp

🚀100 Times Faster Natural Language Processing in Python - iPython notebook
HTML
325
star
62

education-toolkit

Educational materials for universities
Jupyter Notebook
307
star
63

controlnet_aux

Python
306
star
64

optimum-intel

🤗 Optimum Intel: Accelerate inference with Intel optimization tools
Jupyter Notebook
295
star
65

datablations

Scaling Data-Constrained Language Models
Jupyter Notebook
293
star
66

unity-api

C#
284
star
67

open-muse

Open reproduction of MUSE for fast text2image generation.
Python
284
star
68

audio-transformers-course

The Hugging Face Course on Transformers for Audio
MDX
247
star
69

hub-docs

Docs of the Hugging Face Hub
221
star
70

lighteval

LightEval is a lightweight LLM evaluation suite that Hugging Face has been using internally with the recently released LLM data processing library datatrove and LLM training library nanotron.
Python
208
star
71

quanto

A pytorch Quantization Toolkit
Python
201
star
72

simulate

🎢 Creating and sharing simulation environments for embodied and synthetic data research
Python
185
star
73

ratchet

A cross-platform browser ML framework.
Rust
184
star
74

optimum-benchmark

A unified multi-backend utility for benchmarking Transformers, Timm, Diffusers and Sentence-Transformers with full support of Optimum's hardware optimizations & quantization schemes.
Python
183
star
75

hf_transfer

Rust
181
star
76

olm-datasets

Pipeline for pulling and processing online language model pretraining data from the web
Python
169
star
77

instruction-tuned-sd

Code for instruction-tuning Stable Diffusion.
Python
167
star
78

optimum-neuron

Easy, fast and very cheap training and inference on AWS Trainium and Inferentia chips.
Jupyter Notebook
163
star
79

llm-swarm

Manage scalable open LLM inference endpoints in Slurm clusters
Python
156
star
80

OBELICS

Code used for the creation of OBELICS, an open, massive and curated collection of interleaved image-text web documents, containing 141M documents, 115B text tokens and 353M images.
Python
147
star
81

workshops

Materials for workshops on the Hugging Face ecosystem
Jupyter Notebook
146
star
82

cosmopedia

Python
138
star
83

api-inference-community

Python
131
star
84

diffusion-fast

Faster generation with text-to-image diffusion models.
Python
127
star
85

diarizers

Python
106
star
86

optimum-habana

Easy and lightning fast training of 🤗 Transformers on Habana Gaudi processor (HPU)
Python
106
star
87

sharp-transformers

A Unity plugin for using Transformers models in Unity.
C#
104
star
88

competitions

Python
101
star
89

hf-hub

Rust client for the huggingface hub aiming for minimal subset of features over `huggingface-hub` python package
Rust
93
star
90

olm-training

Repo for training MLMs, CLMs, or T5-type models on the OLM pretraining data, but it should work with any hugging face text dataset.
Python
87
star
91

fuego

[WIP] A 🔥 interface for running code in the cloud
Python
84
star
92

tune

Python
83
star
93

datasets-viewer

Viewer for the 🤗 datasets library.
Python
82
star
94

optimum-graphcore

Blazing fast training of 🤗 Transformers on Graphcore IPUs
Python
78
star
95

frp

FRP Fork
Go
73
star
96

paper-style-guide

71
star
97

block_movement_pruning

Block Sparse movement pruning
Python
70
star
98

amused

Python
68
star
99

doc-builder

The package used to build the documentation of our Hugging Face repos
Python
67
star
100

data-measurements-tool

Developing tools to automatically analyze datasets
Python
67
star