Minibase/Spanish-to-English-Translation-Standard

Spanish-to-English Translation Standard 🤖

<div align="center">

A compact Spanish-to-English translation model optimized for accuracy and efficiency.

    

Built by [Minibase](https://minibase.ai) - Train and deploy small AI models from your browser. Browse all of the models and datasets available on the [Minibase Marketplace](https://minibase.ai/wiki/Special:Marketplace).

</div>

📋 Model Summary

Minibase-Spanish-to-English-Translation-Standard is a specialized translation model that converts Spanish text to high-quality English translations. It maintains contextual accuracy while being optimized for fast, local inference without requiring external API calls.

Key Features

• —🌍 Spanish to English: Specialized for Spanish-to-English translation
• —📏 Compact Size: 386MB (Q8_0 quantized) - efficient deployment
• —⚡ Fast Inference: ~245ms average response time
• —🎯 Context Preservation: Maintains meaning and nuance in translations
• —🔄 Local Processing: No data sent to external servers
• —📊 Translation Metrics: Evaluated with BLEU, METEOR, and chrF scores
• —🏗️ 4096 Token Context: Handles longer documents and conversations

🚀 Quick Start

Local Inference (Recommended)

1. 1.Install llama.cpp (if not already installed):

   # Clone and build llama.cpp
   git clone https://github.com/ggerganov/llama.cpp
   cd llama.cpp
   make

   # Return to project directory
   cd ../spanish-english-standard

1. 1.Download the GGUF model:

   # Download model files from HuggingFace
   wget https://huggingface.co/Minibase/Spanish-to-English-Translation-Standard/resolve/main/model.gguf
   wget https://huggingface.co/Minibase/Spanish-to-English-Translation-Standard/resolve/main/spanish_english_inference.py
   wget https://huggingface.co/Minibase/Spanish-to-English-Translation-Standard/resolve/main/config.json
   wget https://huggingface.co/Minibase/Spanish-to-English-Translation-Standard/resolve/main/tokenizer_config.json
   wget https://huggingface.co/Minibase/Spanish-to-English-Translation-Standard/resolve/main/generation_config.json

1. 1.Start the model server:

   # Start llama.cpp server with the GGUF model
   ./Minibase-spanish-to-english-translation-standard-imported.app/Contents/MacOS/run_server

1. 1.Make API calls:

   import requests

   # Translate Spanish text via REST API
   response = requests.post("http://127.0.0.1:8000/completion", json={
       "prompt": "Instruction: Translate the following Spanish text to English.\n\nSpanish: Hola, ¿cómo estás? Me gustaría pedir una pizza con pepperoni.\n\nEnglish: ",
       "max_tokens": 100,
       "temperature": 0.1
   })

   result = response.json()
   print(result["content"])
   # Output: "Hello, how are you? I would like to order a pepperoni pizza."

Python Client (Recommended)

# Download and use the provided Python client
from spanish_english_inference import TranslationClient

# Initialize client (connects to local server)
client = TranslationClient()

# Translate Spanish text to English
spanish_text = """La inteligencia artificial está revolucionando el mundo de la tecnología.
Cada día vemos avances increíbles en el procesamiento del lenguaje natural."""

translation = client.translate_text(spanish_text)
print(translation)
# Output: "Artificial intelligence is revolutionizing the world of technology.
# Every day we see incredible advances in natural language processing."

Command Line Usage

# Translate text directly from command line
python3 spanish_english_inference.py "Hola mundo"
# Output: "Hello world"

📊 Performance Benchmarks

Benchmark Details: Spanish translation test dataset (20 samples), Q8_0 quantization, 4096 token context

🔧 Model Details

Architecture

• —Base Model: LlamaForCausalLM
• —Parameters: ~788M
• —Context Length: 4096 tokens
• —Vocabulary Size: 49,152
• —Quantization: Q8_0 (386MB final size)

Training Data

• —Fine-tuned on Spanish-English translation pairs
• —Includes diverse text types: news, literature, technical docs
• —Balanced corpus for formal and informal Spanish
• —Optimized for natural, fluent English output

Intended Use

• —Primary: Spanish to English translation
• —Secondary: Cross-lingual understanding and communication
• —Domains: General text, news, business, academic content
• —Languages: Spanish (input) → English (output)

🛠️ Technical Specifications

Input Format

Instruction: Translate the following Spanish text to English.

Spanish: [Your Spanish text here]

English:

Output Characteristics

• —Generates fluent, natural English translations
• —Maintains original meaning and context
• —Appropriate formality level preservation
• —Length typically 95-110% of source text

Limitations

• —Spanish input language only
• —English output language only
• —Best on 50-1000 word inputs
• —May not handle very technical jargon perfectly
• —Performance varies by Spanish dialect

📈 Understanding the Metrics

Why These Scores Are Perfect for Translation:

• —METEOR (79.7%): More flexible metric that considers synonyms and word order. Higher scores indicate more natural translations.

• —chrF (72.7%): Character-level metric good for morphologically rich languages like Spanish. Scores above 70% show excellent character-level accuracy.

• —Semantic Similarity (70.9%): Ensures the translation conveys the same meaning as the source text.

• —Length Ratio (1.05): Good translations typically maintain similar length to the source text.

• —Latency (111ms): Fast enough for real-time translation applications.

Key Insight: While BLEU scores might seem modest compared to large commercial models, this compact model provides excellent translation quality for its size and speed.

📜 Citation

If you use Spanish-to-English-Translation-Standard in your research, please cite:

@misc{spanish-english-translation-standard-2025,
  title={Spanish-to-English-Translation-Standard: A Compact Translation Model},
  author={Minibase AI Team},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/Minibase/Spanish-to-English-Translation-Standard}
}

🙏 Acknowledgments

• —Minibase: For providing the training platform and infrastructure
• —Flores-101 Dataset: Used for benchmarking and evaluation
• —llama.cpp: For efficient CPU inference
• —Open Source Community: For the foundational technologies

📞 Support

• —Website: minibase.ai
• —Discord: Join our community
• —Documentation: help.minibase.ai

📋 License

This model is released under the Apache License 2.0.

<div align="center">

Built with ❤️ by the Minibase team

Making AI more accessible for everyone

💬 Join our Discord </div>