Capabilities

• Accurate diagnosis of crop diseases from text descriptions or leaf symptoms
• Practical treatment recommendations (organic + conventional options)
• Region-aware, climate-resilient advice on yield, irrigation, soil health, and prevention
• Confidence scores, risk flags, and safety warnings
• Clear, actionable, farmer-friendly language

Performance

• 76% win rate against the base model (Mistral-8x7B-Instruct)
• Trained on ~21k instruction-completion pairs (PlantVillage + adapted data)

Quick Start

python
Copy code
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "brimbim/agriculture-crop-disease-model-v1"

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name, device_map="auto")

prompt = "You are an expert agronomist helping smallholder farmers. A tomato plant has yellow spots with white centers on the leaves. Diagnose and recommend treatment."

inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=400, temperature=0.7, do_sample=True)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Training Details

• Base Model: mistralai/Mixtral-8x7B-Instruct-v0.1
• Method: LoRA (via AutoScientist)
• Dataset: 15k+ agriculture-specific examples + general-purpose diversity
• License: Apache 2.0