Skip to main content

LFM2 English-Korean Translation: Efficient Bidirectional Translation with 1.2B Parameters

GitHub

What's inside?​

An efficient bidirectional translation system powered by Liquid AI's LFM2 1.2B model fine-tuned for Korean-English translation. This project demonstrates how domain-specific fine-tuning can achieve superior performance compared to models 3x larger, outperforming Google's Gemma-3 4B and Alibaba's Qwen3 4B on the Flores-200 benchmark.

Key features:

  • Automatic language detection - Intelligently detects input language and translates accordingly
  • High-quality translation - CHrF++ 32.96 / BLEU 12.05 on Flores-200 benchmark
  • Efficient inference - Runs on modest hardware with merged adapters for speed
  • Easy-to-use CLI - Simple command-line interface powered by Fire

This project was built and released by Kiwoong Yeom with the support of Maxime Labone. Link to the original announcement on LinkedIn

Quick Start​

  1. Clone the repository
git clone https://github.com/Liquid4All/cookbook.git
cd cookbook/examples/lfm2-english-to-korean
  1. Install dependencies
uv sync
  1. Run translation example
uv run python main.py --text "$(cat linkedin_post.txt)" --max-new-tokens 1024

Understanding the Architecture​

The system uses a two-stage training approach:

  1. Supervised Fine-tuning (SFT): 100K high-quality Korean-English parallel datasets establish the translation foundation
  2. Reinforcement Learning (RL): GRPO optimization with 10K additional samples refines translation quality

Model Components​

  • Base Model: gyung/lfm2-1.2b-koen-mt-v4-100k - SFT fine-tuned LFM2 1.2B
  • Adapter: gyung/lfm2-1.2b-koen-mt-v5-rl-10k-adapter - LoRA adapter trained with GRPO
  • Automatic Detection: Regular expression pattern matching for Korean text (Hangul syllables, Jamo)

The system automatically detects the input language and applies the appropriate translation direction, supporting both English→Korean and Korean→English translation.

CLI Usage​

uv run python main.py [OPTIONS]

Options:
  --text TEXT                    Text to translate (required)
  --model-name TEXT             Base model name (default: gyung/lfm2-1.2b-koen-mt-v4-100k)
  --adapter-name TEXT           Adapter name (default: gyung/lfm2-1.2b-koen-mt-v5-rl-10k-adapter)
  --max-new-tokens INTEGER     Maximum tokens to generate (default: 256)
  --temperature FLOAT          Sampling temperature (default: 0.3)
  --min-p FLOAT               Minimum probability threshold (default: 0.15)
  --repetition-penalty FLOAT   Repetition penalty (default: 1.05)

Example Usage​

Translate English to Korean:

uv run python main.py --text "Hello, how are you today?"

Translate Korean to English:

uv run python main.py --text "안녕하세요, 오늘 어떻게 지내세요?"

Process a file:

uv run python main.py --text "$(cat your_file.txt)" --max-new-tokens 1024

Performance Benchmarks​

On Flores-200 benchmark (1,012 samples):

ModelParametersCHrF++BLEU
LFM2-KoEn-v5-RL1.2B32.9612.05
Gemma-3-4B4B32.8311.36
Qwen3-4B4B25.627.46

The 1.2B parameter model outperforms models 3x larger, demonstrating that specialized training matters more than raw parameter count.

Further Improvements​

Next steps for enhanced performance and efficiency:

  • Speed optimization with quantization techniques (GGUF, AWQ, GPTQ)
  • llama.cpp integration for faster CPU inference
  • Full parameter RL training with expanded compute resources
  • Length normalization removal based on recent Qwen team findings
  • Extended dataset training with 200K SFT + 25K RL samples

Performance Optimization​

The current implementation uses adapter merging for faster inference. Future improvements include:

  • Quantized model variants for resource-constrained environments
  • Streaming inference for real-time translation
  • Batch processing for large document translation

Need help?​

Join the Liquid AI Discord Community and ask.

Discord