mjpsm / Jali-xgb-model

Jali Regression Model

Model Overview

The Jali Regression Model predicts a continuous score between 0.0 and 1.0 that reflects the degree of Jali expressiveness in a text input. The Jali archetype represents the expressive voice, charisma, and rhythm in communication, rooted in the tradition of griots and modern spoken word artists. High scores indicate powerful, charismatic, and rhythmic expression, while low scores indicate weak, hesitant, or muted expression.

• Model Type: Regression (XGBoost)
• Embedding Model: SentenceTransformer all-mpnet-base-v2
• Dataset Size: 368 rows (balanced across 0.0–1.0)
• Output Range: 0.0 → 1.0
• Intended Use: To measure expressiveness and narrative impact in text, and to support culturally grounded AI projects under the Soulprint framework.

Performance Metrics

Evaluated on held-out test data:

• MSE: 0.0086
• RMSE: 0.0928
• R² Score: 0.896

Interpretation:

• Predictions are on average within ±0.1 of the true label.
• The model explains nearly 90 percent of the variance in the dataset.
• This performance indicates reliable and consistent regression results.

Dataset Details

The dataset was created with 368 labeled rows , designed to capture a spectrum of Jali expressiveness.

• Inputs: Natural language reflections, scenarios, and short narratives varying in length (1–5 sentences).
• Labels: Continuous scores from 0.0–1.0 indicating expressiveness level.
• Balance: Equal coverage across weak (0.0–0.3), moderate (0.4–0.6), and strong (0.7–1.0) expressiveness.
• Perspective Variety: Includes first-person, observer, audience, and group perspectives to avoid overfitting to “I” statements.

Limitations

• The dataset is synthetic and may not capture the full complexity of real-world expressive styles.
• Cultural nuance beyond the Jali archetype may not be represented.
• The model is optimized for short passages (1–5 sentences) and may not generalize well to longer documents.

Intended Uses

• Scoring expressiveness in reflective or narrative writing.
• Supporting Soulprint-aligned projects that explore archetypal strengths.
• Research into Afrocentric AI models that map cultural archetypes into measurable traits.

Not intended for:

• Judgment of individuals in real-world contexts.
• Clinical, legal, or high-stakes decision making.

How to Use

import joblib
from sentence_transformers import SentenceTransformer
from huggingface_hub import hf_hub_download

# -----------------------------
# 1. Download model from Hugging Face Hub
# -----------------------------
REPO_ID = "mjpsm/Jali-xgb-model"  # replace with your repo if different
FILENAME = "Jali_xgb_model.pkl"

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

# -----------------------------
# 2. Load model + embedder
# -----------------------------
model = joblib.load(model_path)
embedder = SentenceTransformer("all-mpnet-base-v2")

# -----------------------------
# 3. Example prediction
# -----------------------------
text = "The poet delivered each line with rhythm that moved the audience deeply."
embedding = embedder.encode([text])
score = model.predict(embedding)[0]

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

Soulprint context

The Jali archetype is Expressive. It embodies the role of messenger, MC, and spoken word artist who transforms chaos into clarity and uplifts morale with charisma and rhythm. Symbolic inspirations include Maya Angelou, Gil Scott-Heron, and spoken word cyphers.