CapraMind-V1-small

🐑 CapraMind-V1-small

A Specialized Large Language Model for Dairy Goat Farming Management

CapraMind-V1-small is a lightweight, domain-specific language model fine-tuned on high-quality dairy goat husbandry knowledge. It provides accurate, safe, and practical guidance on nutrition, reproduction, disease prevention, milking, and daily management—tailored for farmers, veterinarians, and agricultural extension workers.

---

📌 Overview

• Base Model: Qwen3-1.7B-Base
• Finetuning Method: LoRA (Low-Rank Adaptation)
• Training Data: 2,755+ expert-verified QA pairs from veterinary manuals, FAO guidelines, and Chinese agricultural standards
• Parameters: ~1.7B (LoRA rank=8, alpha=16)
• Languages: Simplified Chinese (primary), with technical terms in English (e.g., "glucose calcium", "mastitis")
• License: Non-commercial use only (see License)

---

✨ Key Features

✅ Expert-Level Knowledge
Covers core topics:

• Estrus detection & breeding management
• Lactation nutrition & feed formulation
• Common diseases (e.g., mastitis, postpartum hypocalcemia)
• Vaccination & deworming schedules
• Milking hygiene & udder care

✅ Safety-First Design

• Refuses to recommend human medications or unverified treatments
• Cites standard protocols (e.g., “According to the Dairy Goat Management Handbook, 2022…”)
• Falls back to: “Please consult a licensed veterinarian” for high-risk queries

✅ Lightweight & Deployable

• Runs on consumer GPUs (e.g., RTX 3090/4090 with 24GB VRAM)
• Supports GGUF quantization for CPU/edge deployment

---

🚀 Quick Start

1. Install Dependencies

pip install transformers accelerate torch sentencepiece

2. Load the Model (Hugging Face Style)

from transformers import AutoTokenizer, AutoModelForCausalLM

model_name = "CapraMind-V1-small"

tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    trust_remote_code=True,
    device_map="auto",
    torch_dtype="auto"
)
model.eval()

3. Inference Example

prompt = "奶山羊产后瘫痪的紧急处理措施是什么？"
messages = [
    {"role": "system", "content": "你是一名资深奶山羊兽医，请基于专业规范回答。"},
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)

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

💡 Note: Use the Qwen chat template (tokenizer.apply_chat_template) for best results.

---

📊 Performance

Evaluated by 2 certified livestock veterinarians on real-world farmer questions.

---

⚠️ Limitations

• Not a substitute for veterinary care: For emergencies, always contact a professional.
• Coverage bias: Strong in common diseases & nutrition; weaker in rare genetic disorders.
• Language: Optimized for Chinese-speaking users; English queries may yield suboptimal results.

---

🔒 License

This model is released under the MIT License.

• ✅ You may freely use, modify, and distribute this model for both non-commercial and commercial purposes.
• ✅ You are required to include the original copyright notice and license text in any redistribution.

⚠️ Important Note:
This model is built upon Qwen3-1.7B-Base, which is licensed under Alibaba’s Qwen License.
The MIT license applies only to the LoRA adapter weights and training code in this repository.
You must comply with Qwen3’s license terms when using or redistributing the base model.

In practice:

• You may deploy CapraMind-V1-small in your app or service.
• But you must not redistribute Qwen3-1.7B-Base unless you comply with Alibaba’s terms (e.g., attribution, non-resale of base weights).

For commercial deployment, we recommend using only the LoRA adapter alongside a legally obtained Qwen3 base model.

---

🙏 Acknowledgements

• Training data curated from:
  • Dairy Goat Production Handbook (China Agricultural Press, 2022)
  • FAO Guidelines on Small Ruminant Management
  • National Dairy Goat Association (NDGA) protocols
• Built with LLaMA-Factory and Qwen