Griot-xgb-model

Griot Regression Model Card

• Model name: mjpsm/Griot_xgb_model
• Model type: Regression model (XGBoost)
• Archetype: Griot — Reflective, sacred archivist of memory
• Author: Mazamesso “JP” Meba (Coding in Color / MyVillage Project)
• License: MIT
• Version: v1.0

This model predicts a Griot score (0–1) for a given input text, reflecting how closely the text embodies the Griot archetype traits: wise, introspective, articulate, attentive, empathic, grounded.

• Primary use: Ancestral-AI aligned applications where reflective storytelling or wisdom-sharing needs to be measured.
• Applications:
  • Analyzing transcripts to detect reflective/archival tones
  • Powering Soulprint-aligned chatbots or recommendation systems
  • Scoring personal reflections or intergenerational narratives

Not intended for: general sentiment analysis, factual correctness evaluation, or unrelated text classification.

• Training dataset: griot_balanced_training_data_strict_unique.jsonl

• Size: ~1,200 examples (balanced, strict uniqueness applied)

• Format: JSONL with fields:
  {"input": "During a family gathering, I reminded everyone of our shared history.", "output": 0.82}

• Sources: Culturally-informed prompts and Afrocentric wisdom narratives created under the Soulprint system.

• Base embeddings: sentence-transformers/all-mpnet-base-v2

• Regressor: XGBoost (regression objective)

• Saved file: griot_xgb_regression_model_updated_parameters.pkl

Example Usage

import joblib
from sentence_transformers import SentenceTransformer
from huggingface_hub import hf_hub_download

# -----------------------------
# 1. Download model from Hugging Face Hub
# -----------------------------
REPO_ID = "mjpsm/Griot_xgb_model"
FILENAME = "griot_xgb_regression_model_updated_parameters.pkl"

model_path = hf_hub_download(repo_id=REPO_ID, filename=FILENAME)
model = joblib.load(model_path)

# -----------------------------
# 2. Load embedder
# -----------------------------
embedder = SentenceTransformer("all-mpnet-base-v2")

# -----------------------------
# 3. Example prediction
# -----------------------------
text = "During a heated family argument, I stayed calm and reminded everyone of our values."
embedding = embedder.encode([text])
score = model.predict([embedding])[0]

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