Skip to content

EXAONE-3.5-7.8B-Instruct

A bilingual English and Korean instruction model

Visit official site

What EXAONE-3.5-7.8B-Instruct is

The model is described as one of a series of instruction-tuned bilingual generative models from LG AI Research that support long-context processing.

What you can do with it

  • Generate text in English and Korean
  • Process inputs up to 32K tokens
  • Apply a chat template to input messages

Getting started

  • Install transformers v4.43 or newer
  • Load the model and tokenizer from Hugging Face
  • Build input IDs with apply_chat_template

Pros & cons

✓ What we like

  • Long context up to 32K is supported
  • The material describes competitive general-domain results

! What to watch out for

  • Loading needs transformers v4.43 or newer
  • A chat template is applied before generation

FAQ

Which languages are named?

The material names English and Korean as the bilingual pair.

How is it loaded?

The steps name AutoModelForCausalLM and AutoTokenizer from Hugging Face.

Last reviewed: 2026-11-15

More AI coding tools tools

View all →

How we review