Zamani-xgb-model

Zamani Archetype Regressor

📖 Overview

The Zamani Archetype Regressor is part of the Soulprint framework, designed to measure foresight and long-term vision in text.
It predicts a continuous Zamani foresight score (0.0–1.0) using XGBoost regression on top of SentenceTransformer embeddings.

• Low foresight (0.0–0.3): short-sighted, reactive, impulsive decisions.
• Mid foresight (0.3–0.6): near-term planning, seasonal or semester-level foresight.
• High foresight (0.6–1.0): strategic patience, legacy thinking, multi-generational vision.

---

📊 Dataset

• Size: 900 rows
• Balanced: 300 Low / 300 Mid / 300 High foresight examples
• Output range: 0.11 → 0.96
• Sentence variation:
  • Cycle of 1-sentence, 2–3 sentences, 4–5 sentences, then repeat
  • Mix of first-person and third-person perspectives
  • Varied structures (After…, Though…, Because…, inversions, compounds)

---

⚙️ Model Details

• Architecture: XGBoost Regressor
• Embeddings: all-mpnet-base-v2 (768-dim)
• Training size: 900 balanced rows
• File format: Zamani_xgb_model.json

---

📈 Performance

• MSE: 0.0117
• R²: 0.833

This indicates the model explains 83% of the variance in foresight scores — a strong fit given the dataset size and complexity.

---

🚀 Usage

import xgboost as xgb
from sentence_transformers import SentenceTransformer
from huggingface_hub import hf_hub_download

# -----------------------------
# 1. Download model from Hugging Face Hub
# -----------------------------
REPO_ID = "mjpsm/Zamani-xgb-model"
FILENAME = "Zamani_xgb_model.json"

model_path = hf_hub_download(repo_id=REPO_ID, filename=FILENAME)

# -----------------------------
# 2. Load model + embedder
# -----------------------------
model = xgb.XGBRegressor()
model.load_model(model_path)

embedder = SentenceTransformer("all-mpnet-base-v2")

# -----------------------------
# 3. Example prediction
# -----------------------------
text = "She planted seeds for trees her grandchildren would one day sit under."
embedding = embedder.encode([text])
score = model.predict(embedding)[0]

print("Predicted Zamani Score:", round(float(score), 3))