Overview

DeBERTa v3 base mnli fever anli is a specialized NLI (Natural Language Inference) model optimized for zero-shot text classification. By leveraging the enhanced mask language modeling of DeBERTa v3 and fine-tuning on a combination of MNLI, FEVER, and ANLI datasets, it excels at determining whether a premise entails or contradicts a given hypothesis. For developers, this means you can categorize text into custom labels on the fly without needing a labeled training set for every new category. It is a highly efficient drop-in replacement for standard BERT-based classifiers when accuracy in semantic relationship detection is critical, making it ideal for content moderation, automated fact-checking, and dynamic tagging systems.

Highlights

High-accuracy zero-shot classification without task-specific training
Fine-tuned on MNLI, FEVER, and ANLI for robust inference
Efficient DeBERTa v3 architecture for superior semantic understanding
Ideal for dynamic labeling and automated fact-verification pipelines

Usage

Install

# Install Hugging Face transformers
pip install transformers torch

SDK Usage

# Load model with transformers
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained("MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli")
tokenizer = AutoTokenizer.from_pretrained("MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli")

Hugging Face Download

We recommend downloading the model via the Hugging Face CLI or Hub SDK.

Guidance：Before downloading, install huggingface_hub with:

Guidance

pip install -U huggingface_hub

CLI Download

Download the full repository

huggingface-cli download MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli

Download a single file to a local folder (e.g. config.json into ./dir)

huggingface-cli download MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli config.json --local-dir ./dir

See the official docs for more CLI options

SDK Download

# 模型下载
from huggingface_hub import snapshot_download
model_dir = snapshot_download('MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli')

Git Download

Make sure git-lfs is installed first

Git Download

git lfs install
git clone https://huggingface.co/MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli

To skip LFS large-file downloads, use:

Skip LFS

GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli

Model 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 torch

Load the model and run inference

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained('MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli')
tokenizer = AutoTokenizer.from_pretrained('MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli')

Full Documentation

来源: HuggingFace

---
language:

en

license: mit
tags:

text-classification

zero-shot-classification

datasets:

multi_nli

facebook/anli

fever

metrics:

accuracy

pipeline_tag: zero-shot-classification
model-index:

name: MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli

results:
- task:
type: natural-language-inference
name: Natural Language Inference
dataset:
name: anli
type: anli
config: plain_text
split: test_r3
metrics:
- type: accuracy
value: 0.495
name: Accuracy
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiYWViYjQ5YTZlYjU4NjQyN2NhOTVhNjFjNGQyMmFiNmQyZjRkOTdhNzJmNjc3NGU4MmY0MjYyMzY5MjZhYzE0YiIsInZlcnNpb24iOjF9.S8pIQ7gEGokd_wKXMi6Bc3B2DThIP3cvVkTFErZ-2JxXTSCy1TBuulY3dzGfaiP7kTHbL52OuBhG_-wb7Ue9DQ
- type: precision
value: 0.4984740618243923
name: Precision Macro
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiOTllZDU3NmVmYjk4ZmYzNjAwNzExMGZjNDMzOWRkZjRjMTRhNzhlZmI0ZmNlM2E0Mzk4OWE5NTM5MTYyYWU5NCIsInZlcnNpb24iOjF9.WHz_TUJgPVn-rU-9vBCDdmSMOuWzADwr09rJY6ktqRM46zytbyWs7Vcm7jqDrTkfU-rp0_7IyoNv_xEsKhJbBA
- type: precision
value: 0.495
name: Precision Micro
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiZjllODE3ZjUxZDhiMTI0MzZmYjY5OTUwYWI2OTc4ZjJhNTVjMjY2ODdkMmJlZjQ5YWQ1Mjk2ZThmYjJlM2RlYSIsInZlcnNpb24iOjF9.a9V06-O7l9S0Bv4vj0aard8128SAP61DZdXl_3XqdmNgt_C6KAoDBVueF2M2kF_kT6lRfEz6YW0ACIfJNXDYAA
- type: precision
value: 0.4984357572868885
name: Precision Weighted
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiNjhiMzYzY2JiMmYwN2YxYzEwZTQ3NGI1NzFmMzliNjJkMDE2YzI5Njg1ZjEzMGIxODdiMDNmYmI4Y2Y2MmJkMiIsInZlcnNpb24iOjF9.xvZZaUMogw9MJjb3ls6h5liDlTqHMmNgqk6KbyDqQWfCcD255brCU3Xo6nECwaChS4te0dQu_iWGBqR_o2kYAA
- type: recall
value: 0.49461028192371476
name: Recall Macro
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiZDVjYTEzOTI0ZjVhOTk3ZTkzZmZhNTk5ODcxMWJhYWU4ZTRjYWVhNzcwOWY5YmI2NGFlYWE4NjM5MDY5NTExOSIsInZlcnNpb24iOjF9.xgHCB2rbCQBzHzUokw4u8JyOdhtF4yvPv1t8t7YiEkaAuM5MAPsVuCZ1VtlLapHS_IWetlocizsVl6akjh3cAQ
- type: recall
value: 0.495
name: Recall Micro
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiYTEyYmM0ZDQ0M2RiMDNhNjIxNzQ4OWZiNTBiOTAwZDFkNjNmYjBhNjA4NmQ0NjFkNmNiZTljNDkxNDg3NzIyYSIsInZlcnNpb24iOjF9.3FJPwNtwgFNvMjVxVAayaVXXR1sWlr0sqAYmXzmMzMxl7IJh6RS77dGPwFaqD3jamLVBiqPn9wsfz5lFK5yTAA
- type: recall
value: 0.495
name: Recall Weighted
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiNmY1MjZlZTQ4OTg5YzdlYmFhZDMzMmNlNjNkYmIyZGI4M2NjZjQ1ZDVkNmZkMTUxNjI3M2UwZmI1MDM1NDYwOSIsInZlcnNpb24iOjF9.cnbM6xjTLRa9z0wEDGd_Q4lTXVLRKIQ6_YLGLjf-t7Nto4lzxAeWF-RrwA0Mq9OPITlJq2Jk1Eg_0Utb13d9Dg
- type: f1
value: 0.4942810999491704
name: F1 Macro
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiN2U3NGM1MDM4YTM4NzQxMGM4ZTIyZDM2YTQ1MGNlZWM1MzEzM2MxN2ZmZmRmYTM0OWJmZGJjYjM5OWEzMmZjNSIsInZlcnNpb24iOjF9.vMtge1F-tmMn9D3aVUuwcNEXjqpNgEyHAl9f5UDSoTYcOgTwi2vi5yRGRCl8y6Fx7BtgaCwMyoZVNbP5-GRtCA
- type: f1
value: 0.495
name: F1 Micro
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiNjBjMTQ5MmQ5OGE5OWJjZGMyNzg4N2RmNDUzMzQ5Zjc4ZTc4N2JlMTk0MTc2M2RjZTgzOTNlYWQzODAwNDI0NCIsInZlcnNpb24iOjF9.yxXG0CNWW8__xJC14BjbTY9QkXD75x6uCIXR51oKDemkP0b_xGyd-A2wPIuwNJN1EYkQevPY0bhVpRWBKyO9Bg
- type: f1
value: 0.4944671868893595
name: F1 Weighted
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiMzczNjQzY2FmMmY4NTAwYjNkYjJlN2I2NjI2Yjc0ZmQ3NjZiN2U5YWEwYjk4OTUyOTMzZTYyZjYzOTMzZGU2YiIsInZlcnNpb24iOjF9.mLOnst2ScPX7ZQwaUF12W2nv7-w9lX9-BxHl3-0T0gkSWnmtBSwYcL5faTX0_I5q33Fjz5tfkjpCJuxP5JYIBQ
- type: loss
value: 1.8788293600082397
name: loss
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiMzRlOTYwYjU1Y2Y4ZGM0NDBjYTE2MmEzNWIwN2NiMWVkOWZlNzA2ZmQ3YjZjNzI4MjQwYWZhODIwMzU3ODAyZiIsInZlcnNpb24iOjF9._Xs9bl48MSavvp5eyamrP2iNlFWv35QZCrmWjJXLkUdIBx0ElCjEdxBb3dxPGnUxdpDzGMmOoKCPI44ZPXrtDw
- task:
type: natural-language-inference
name: Natural Language Inference
dataset:
name: anli
type: anli
config: plain_text
split: test_r1
metrics:
- type: accuracy
value: 0.712
name: Accuracy
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiYWYxMGY0ZWU0YTEyY2I3NmQwZmQ3YmFmNzQxNGU5OGNjN2ViN2I0ZjdkYWUzM2RmYzkzMDg3ZjVmNGYwNGZkZCIsInZlcnNpb24iOjF9.snWBusAeo1rrQqWk--vTxb-CBcFqM298YCtwTQGBZiFegKGSTSKzj-SM6HMNsmoQWmMuv7UfYPqYlnzEthOSAg
- type: precision
value: 0.7134839439315348
name: Precision Macro
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiNjMxMjg1Y2QwNzMwM2ZkNGM3ZTJhOGJmY2FkNGI1ZTFhOGQ3ODViNTJmZTYwMWJkZDYyYWRjMzFmZDI1NTM5YSIsInZlcnNpb24iOjF9.ZJnY6zYOBn-YEtN7uKzQ-VKXPwlIO1zq19Yuo37vBJNSs1dGDd8f1jgfdZuA19e_wA3Nc5nQKe9VXRwPHPgwAQ
- type: precision
value: 0.712
name: Precision Micro
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiZWM4YWQyODBlYTIwMWQxZDA1NmY1M2M2ODgwNDJiY2RhMDVhYTlkMDUzZTJkMThkYzRmNDg2YTdjMjczNGUwOCIsInZlcnNpb24iOjF9.SogsKHdbdlEs05IBYwXvlnaC_esg-DXAPc2KPRyHaVC5ItVHbxa63NpybSpao4baOoMlLG9aRe7TjG4gtB2dAQ
- type: precision
value: 0.7134676028447461
name: Precision Weighted
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiODdjMzFkM2IwNWZiM2I4ZWViMmQ4NWM5MDY5ZWQxZjc1MGRmNjhmNzJhYWFmOWEwMjg3ZjhiZWM3YjlhOTIxNSIsInZlcnNpb24iOjF9._0JNIbiqLuDZrp_vrCljBe28xexZJPmigLyhkcO8AtH2VcNxWshwCpZuRF4bqvpMvnApJeuGMf3vXjCj0MC1Bw
- type: recall
value: 0.7119814425203647
name: Recall Macro
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiYjU4MWEyMzkyYzg1ZTIxMTc0M2NhMTgzOGEyZmY5OTg3M2Q1ZmMwNmU3ZmU1ZjA1MDk0OGZkMzM5NDVlZjBlNSIsInZlcnNpb24iOjF9.sZ3GTcmGGthpTLL7_Zovq8aBmE3Dp_PZi5v8ZI9yG9N6B_GjWvBuPC8ENXK1NwmwiHLsSvtKTG5JmAum-su0Dg
- type: recall
value: 0.712
name: Recall Micro
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiZDg3NGViZTlmMWM2ZDNhMzIzZGZkYWZhODQxNzg2MjNiNjQ0Zjg0NjQ1OWZkY2I5ODdiY2Y3Y2JjNzRmYjJkMiIsInZlcnNpb24iOjF9.bCZUzJamsozKWehnNph6E5coww5zZTrJdbWevWrSyfT0PyXc_wkZ-NKdyBAoqprBz3_8L3i5hPM6Qsy56b4BDA
- type: recall
value: 0.712
name: Recall Weighted
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiMDk1MDJiOGUzZThlZjJjMzY4NjMzODFiZjUzZmIwMjIxY2UwNzBiN2IxMWEwMGJjZTkxODA0YzUxZDE3ODRhOCIsInZlcnNpb24iOjF9.z0dqvB3aBVYt3xRIb_M4svWebfQc0QaDFVFzHnlA5QGEHkHOW3OecGhHE4EzBqTDI3DASWZTGMjrMDDt0uOMBw
- type: f1
value: 0.7119226991285647
name: F1 Macro
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiM2U0YjMwNzhmOTEyNDZhODU3MTU0YTM4MmQ0NzEzNWI1YjY0ZWQ3MWRiMTdiNTUzNWRkZThjMWE4M2NkZmI0MiIsInZlcnNpb24iOjF9.hhj1BXkuWi9wXrCjT9NwqaPETtOoYNiyqYsJEw-ufA8A4hVThKA6ZBtma1Q_M65-DZFfPEBDBNASLZ7EPSbmDw
- type: f1
value: 0.712
name: F1 Micro
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiODk0Y2EyMzc5M2ZlNWFlNDg2Zjc1OTQxNGY3YjA5YjUxYTYzZjRlZmU4ODYxNjA3ZjkxNGUzYjBmNmMxMzY5YiIsInZlcnNpb24iOjF9.DvKk-3hNh2LhN2ug5e0FgUntL3Ozdfl06Kz7jvmB-deOJH6INi2a2ZySXoEePoo8t2nR6ENFYu9QjMA2ojnpCA
- type: f1
value: 0.7119242267218338
name: F1 Weighted
verified: true
verifyToken: eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiN2MxOWFlMmI2NGRiMjkwN2Q5MWZhNDFlYzQxNWNmNzQ3OWYxZThmNDU2OWU1MTE5OGY2MWRlYWUyNDM3OTkzZCIsInZlcnNpb24iOjF9.QrTD1gE8_wRok9u59W-Mx0cX89K-h2Ad6qa8J5rmP8lc_rkG0ft2n5_GqH1CBZBJwMFYv9