mesabo/agri-chat-multilingual

AgriChat Multilingual - Agricultural Assistant

A multilingual chatbot fine-tuned for agricultural assistance, specifically designed for farmers in West Africa and beyond.

Model Description

• Base Model: Qwen/Qwen2-0.5B-Instruct
• Fine-tuning Method: LoRA (Low-Rank Adaptation)
• Languages: English, French, Spanish, Portuguese, Swahili, Japanese, Arabic
• Domain: Agricultural crop diseases, farming practices, pest management
• License: Apache 2.0

Supported Languages

Language	Code	Coverage
English	en	Full
French	fr	Full
Spanish	es	Full
Portuguese	pt	Full
Swahili	sw	Full
Japanese	ja	Full
Arabic	ar	Full

Use Cases

• Crop Disease Identification: Ask about symptoms and treatments for plant diseases
• Farming Advice: Get guidance on agricultural practices
• Pest Management: Learn about controlling pests affecting crops
• Multilingual Support: Communicate in 7 different languages

Quick Start

from transformers import AutoModelForCausalLM, AutoTokenizer

# Load model
model = AutoModelForCausalLM.from_pretrained("mesabo/agri-chat-multilingual")
tokenizer = AutoTokenizer.from_pretrained("mesabo/agri-chat-multilingual")

# Chat example
messages = [
    {"role": "user", "content": "How do I identify cassava mosaic disease?"}
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt")

outputs = model.generate(**inputs, max_new_tokens=256, temperature=0.7)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)

Training Details

• Training Data: 38 curated Q&A examples across 7 languages
• Epochs: 3
• LoRA Parameters: 2.16M trainable (0.44% of total)
• Training Loss: 2.54
• Hardware: NVIDIA RTX 3090 (25.3 GB)
• Training Time: ~17 seconds

Covered Topics

Crop Diseases

• Cassava mosaic disease
• Maize leaf blight
• Tomato bacterial wilt
• Cashew anthracnose
• Rice blast disease

Farming Practices

• Organic pest control
• Soil health management
• Crop rotation benefits
• Water conservation

Limitations

• Fine-tuned on limited agricultural domain data
• Best suited for common crop diseases in West Africa
• May not cover specialized or rare conditions
• Responses should be verified with local agricultural experts

Intended Use

This model is designed for:

• Agricultural extension workers
• Small-scale farmers
• Agricultural education platforms
• Farming assistance applications

Citation

@misc{agri-chat-multilingual,
  author = {mesabo},
  title = {AgriChat Multilingual - Agricultural Assistant},
  year = {2024},
  publisher = {Hugging Face},
  url = {https://huggingface.co/mesabo/agri-chat-multilingual}
}

Related Models

• mesabo/agri-plant-disease-resnet50 - Plant disease image classification (95%+ accuracy)

Contact

For questions or issues, please open a discussion on the model page.