Practical overview
When it fits
Use it when a Python project needs one of the Korean NLP tasks and model families listed in the repository, and its documented PyTorch and Transformers requirements fit the intended environment.
Who it is for
- Python developers building Korean-language NLP workflows with Transformers.
- Teams choosing among the repository's listed KoELECTRA base, small and task-specific model variants.
Documented uses
- Classify Korean text with NSMC-finetuned models through a sentiment-analysis pipeline.
- Extract named entities from Korean text with Naver-NER models and the project's customized NerPipeline.
- Answer questions against Korean context text with listed KorQuad models.
Requirements & limits
The repository is public and specifies the Apache-2.0 license.
- The README specifies torch>=1.4.0 and transformers==3.0.2; account for this pinned Transformers version when planning an environment.
- Available examples and model pairings are task-specific: NSMC for sentiment analysis, Naver-NER for entity recognition, and KorQuad for question answering.
- The custom NerPipeline returns piece-level results; the README presents its modifications as addressing wordpiece handling and provides options for special tokens and O labels.
- The repository lists both base and smaller models for NSMC and Naver-NER, and base-v2 and small-v2 options for KorQuad.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Set up a Korean sentiment-analysis pipeline
Load a listed NSMC model and create a Transformers sentiment-analysis pipeline.
Before you start
- Python environment with torch>=1.4.0 and transformers==3.0.2.
- Access to the listed Hugging Face NSMC model identifier.
Steps
Load ElectraTokenizer and ElectraForSequenceClassification from a listed NSMC model identifier.
Pass the tokenizer and model to transformers.pipeline with the task set to sentiment-analysis.
Provide Korean text to the resulting pipeline, as shown in the README example.
Expected outcome
A pipeline configured to return sentiment labels and scores for supplied text, as illustrated by the README.
Guide limitations
- The guide follows the README's example and does not verify execution in a particular environment.
- This recipe covers NSMC sentiment classification, not the NER or KorQuad tasks.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 17 October 2026
- Original website title
- monologg/KoELECTRA-Pipeline · Official homepage title · retained in its source language