multilingual MiniLMv2 L12 mnli xnli
Overview
Highlights
- Zero-shot classification across multiple languages without retraining
- Low-latency inference suitable for CPU-based deployments
- Compact architecture reduces memory overhead in production
- Ideal for dynamic labeling and real-time intent detection
Usage
# Install Hugging Face transformers
pip install transformers torch# Load model with transformers
from transformers import AutoModel, AutoTokenizer
model = AutoModel.from_pretrained("MoritzLaurer/multilingual-MiniLMv2-L12-mnli-xnli")
tokenizer = AutoTokenizer.from_pretrained("MoritzLaurer/multilingual-MiniLMv2-L12-mnli-xnli")Hugging Face Download
We recommend downloading the model via the Hugging Face CLI or Hub SDK.
Guidance：Before downloading, install huggingface_hub with:
pip install -U huggingface_hubCLI Download
Download the full repository
huggingface-cli download MoritzLaurer/multilingual-MiniLMv2-L12-mnli-xnliDownload a single file to a local folder (e.g. config.json into ./dir)
huggingface-cli download MoritzLaurer/multilingual-MiniLMv2-L12-mnli-xnli config.json --local-dir ./dirSee the official docs for more CLI options
SDK Download
# 模型下载
from huggingface_hub import snapshot_download
model_dir = snapshot_download('MoritzLaurer/multilingual-MiniLMv2-L12-mnli-xnli')Git Download
Make sure git-lfs is installed first
git lfs install
git clone https://huggingface.co/MoritzLaurer/multilingual-MiniLMv2-L12-mnli-xnliTo skip LFS large-file downloads, use:
GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/MoritzLaurer/multilingual-MiniLMv2-L12-mnli-xnliModel files are hosted on the Hugging Face Hub — download directly via HF CLI / SDK / Git, not through this site.
PyTorch / Transformers Usage
Install Transformers
pip install -U transformers torchLoad the model and run inference
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained('MoritzLaurer/multilingual-MiniLMv2-L12-mnli-xnli')
tokenizer = AutoTokenizer.from_pretrained('MoritzLaurer/multilingual-MiniLMv2-L12-mnli-xnli')Full Documentation
---
language:
- multilingual
- en
- ar
- bg
- de
- el
- es
- fr
- hi
- ru
- sw
- th
- tr
- ur
- vi
- zh
license: mit
tags:
- zero-shot-classification
- text-classification
- nli
- pytorch
metrics:
- accuracy
datasets:
- multi_nli
- xnli
pipeline_tag: zero-shot-classification
widget:
- text: "Angela Merkel ist eine Politikerin in Deutschland und Vorsitzende der CDU"
candidate_labels: "politics, economy, entertainment, environment"
---
---
Multilingual MiniLMv2-L12-mnli-xnli
Model description
This multilingual model can perform natural language inference (NLI) on 100+ languages and is therefore also
suitable for multilingual zero-shot classification. The underlying multilingual-MiniLM-L12 model was created
by Microsoft and was distilled from XLM-RoBERTa-large (see details in the original paper
and newer information in this repo).
The model was then fine-tuned on the XNLI dataset, which contains hypothesis-premise pairs from 15 languages,
as well as the English MNLI dataset.
The main advantage of distilled models is that they are smaller (faster inference, lower memory requirements) than their teachers (XLM-RoBERTa-large).
The disadvantage is that they lose some of the performance of their larger teachers.
For highest inference speed, I recommend using the 6-layer model
(the model on this page has 12 layers and is slower). For higher performance I recommend
mDeBERTa-v3-base-mnli-xnli (as of 14.02.2023).
How to use the model
#### Simple zero-shot classification pipelinefrom transformers import pipeline
classifier = pipeline("zero-shot-classification", model="MoritzLaurer/multilingual-MiniLMv2-L12-mnli-xnli")
sequence_to_classify = "Angela Merkel ist eine Politikerin in Deutschland und Vorsitzende der CDU"
candidate_labels = ["politics", "economy", "entertainment", "environment"]
output = classifier(sequence_to_classify, candidate_labels, multi_label=False)
print(output)
#### NLI use-case
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
model_name = "MoritzLaurer/multilingual-MiniLMv2-L12-mnli-xnli"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name)
premise = "Angela Merkel ist eine Politikerin in Deutschland und Vorsitzende der CDU"
hypothesis = "Emmanuel Macron is the President of France"
input = tokenizer(premise, hypothesis, truncation=True, return_tensors="pt")
output = model(input["input_ids"].to(device)) # device = "cuda:0" or "cpu"
prediction = torch.softmax(output["logits"][0], -1).tolist()
label_names = ["entailment", "neutral", "contradiction"]
prediction = {name: round(float(pred) * 100, 1) for pred, name in zip(prediction, label_names)}
print(prediction)
Training data
This model was trained on the XNLI development dataset and the MNLI train dataset. The XNLI development set consists of 2490 professionally translated texts from English to 14 other languages (37350 texts in total) (see this paper). Note that the XNLI contains a training set of 15 machine translated versions of the MNLI dataset for 15 languages, but due to quality issues with these machine translations, this model was only trained on the professional translations from the XNLI development set and the original English MNLI training set (392 702 texts). Not using machine translated texts can avoid overfitting the model to the 15 languages; avoids catastrophic forgetting of the other languages it was pre-trained on; and significantly reduces training costs.Training procedure
The model was trained using the Hugging Face trainer with the following hyperparameters. The exact underlying model is mMiniLMv2-L12-H384-distilled-from-XLMR-Large.training_args = TrainingArguments(
num_train_epochs=3, # total number of training epochs
learning_rate=4e-05,
per_device_train_batch_size=64, # batch size per device during training
per_device_eval_batch_size=120, # batch size for evaluation
warmup_ratio=0.06, # number of warmup steps for learning rate scheduler
weight_decay=0.01, # strength of weight decay
)Eval results
The model was evaluated on the XNLI test set on 15 languages (5010 texts per language, 75150 in total). Note that multilingual NLI models are capable of classifying NLI texts without receiving NLI training data in the specific language (cross-lingual transfer). This means that the model is also able of doing NLI on the other languages it was training on, but performance is most likely lower than for those languages available in XNLI.The average XNLI performance of multilingual-MiniLM-L12 reported in the paper is 0.711 (see table 11).
This reimplementation has an average performance of 0.75.
This increase in performance is probably thanks to the addition of MNLI in the training data and this model was distilled from
XLM-RoBERTa-large instead of -base (multilingual-MiniLM-L12-v2).
|Datasets|avg_xnli|ar|bg|de|el|en|es|fr|hi|ru|sw|th|tr|ur|vi|zh|
| :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: |
|Accuracy|0.75|0.73|0.78|0.762|0.754|0.821|0.779|0.775|0.724|0.76|0.689|0.738|0.732|0.7|0.762|0.751|
|Speed text/sec (A100 GPU, eval_batch=120)|4535.0|4629.0|4417.0|4500.0|3938.0|4959.0|4634.0|4152.0|4190.0|4368.0|4630.0|4698.0|4929.0|4291.0|4420.0|5275.0|
|Datasets|mnli_m|mnli_mm|
| :---: | :---: | :---: |
|Accuracy|0.818|0.831|
|Speed text/sec (A100 GPU, eval_batch=120)|2912.0|2902.0|