OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Chinese Xlnet

Chinese Xlnet is a collection of pretrained XLNet models for Chinese natural-language processing, providing base and mid-sized checkpoints for developers and researchers to use as starting weights or compare with other Chinese NLP models.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Chinese Xlnet — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when a Chinese NLP project calls for pretrained XLNet weights and you want to select between the listed base and mid checkpoints, or consult the repository’s reported baselines for reading comprehension and sentiment classification.

Who it is for

  • Researchers exploring pretrained models for Chinese-language NLP.
  • Developers prototyping Chinese NLP systems or fine-tuning pretrained weights for domain-specific text tasks.
  • Teams comparing XLNet checkpoints on Chinese reading-comprehension or sentiment-classification tasks.

Documented uses

  • Load pretrained Chinese XLNet weights as a starting point for an NLP prototype.
  • Fine-tune a checkpoint for a domain-specific Chinese text task.
  • Compare XLNet-base and XLNet-mid with other pretrained Chinese models on a task dataset.
Before you choose

Requirements & limits

  • The available sizes differ substantially: XLNet-base has 12 layers and 117 million parameters; XLNet-mid has 24 layers and 209 million parameters.
  • The README gives different download routes by framework: Hugging Face for PyTorch and Baidu Netdisk for TensorFlow.
  • The documented quick-loading method uses Transformers AutoTokenizer and AutoModel with the model names hfl/chinese-xlnet-base and hfl/chinese-xlnet-mid.
  • The README identifies the XLNet-mid pretraining corpus as Chinese Wikipedia plus general data totaling 5.4 billion words.
  • Reported benchmark results cover selected datasets and tasks, so they do not establish expected results on other data.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Load a Chinese XLNet checkpoint with Transformers

Load one of the repository’s named Hugging Face checkpoints using the documented Transformers interface.

Before you start

  • Python environment with the Transformers library; the README’s quick-loading instructions refer to Transformers 2.2.2.
  • Choose the checkpoint name hfl/chinese-xlnet-base or hfl/chinese-xlnet-mid.

Steps

  1. Set MODEL_NAME to hfl/chinese-xlnet-base or hfl/chinese-xlnet-mid.

  2. Use AutoTokenizer.from_pretrained(MODEL_NAME) to load the tokenizer.

  3. Use AutoModel.from_pretrained(MODEL_NAME) to load the model.

Expected outcome

The tokenizer and pretrained model are loaded through the documented AutoTokenizer and AutoModel interface.

Guide limitations

  • The README’s quick-loading example documents loading, not task-specific fine-tuning or inference steps.
  • The README cites Transformers 2.2.2; it does not establish compatibility with other versions.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
ymcui/Chinese-XLNet · Official homepage title · retained in its source language
How we review sourcesSuggest a correction