|
|
--- |
|
|
license: mit |
|
|
datasets: |
|
|
- antonypamo/savantorganized |
|
|
language: |
|
|
- en |
|
|
--- |
|
|
# 🧬 ProSavantEngine Φ9.3 — Icosahedral Resonance Language Model |
|
|
|
|
|
**Author:** [Antony Padilla Morales](https://huggingface.co/antonypamo |
|
|
|
|
|
## 🧠 Overview |
|
|
|
|
|
`ProSavantEngine Φ9.3` extends the *Resonance of Reality Framework (RRF)* by coupling **language semantics** and **icosahedral geometry** through node-conditioned tokens `[NODE_1]`–`[NODE_12]`. |
|
|
Each text sample during training was enriched with its geometric node context, allowing the model to align meaning with spatial-frequency symmetry. |
|
|
|
|
|
This version fine-tunes from **Φ9.2-Lite** on the full RRF corpus `corpus_unificado_total.jsonl`, augmented with `icosahedron_nodes.json`. |
|
|
|
|
|
--- |
|
|
|
|
|
## 🚀 Quick Start |
|
|
|
|
|
Install dependencies: |
|
|
|
|
|
```bash |
|
|
pip install torch transformers datasets scipy plotly gradio |
|
|
|
|
|
|
|
|
Run inference: |
|
|
|
|
|
from transformers import AutoTokenizer, AutoModelForMaskedLM |
|
|
|
|
|
tok = AutoTokenizer.from_pretrained("antonypamo/ProSavantEngine_Phi9_3") |
|
|
model = AutoModelForMaskedLM.from_pretrained("antonypamo/ProSavantEngine_Phi9_3") |
|
|
|
|
|
text = "Quantum resonance aligns with [NODE_5]" |
|
|
inputs = tok(text, return_tensors="pt") |
|
|
outputs = model(**inputs, labels=inputs["input_ids"]) |
|
|
print(outputs.loss) |
|
|
|
|
|
🧩 Fine-Tuning from the Hub |
|
|
|
|
|
You can continue training directly from the Hub: |
|
|
|
|
|
from transformers import Trainer, TrainingArguments |
|
|
|
|
|
args = TrainingArguments.from_pretrained("antonypamo/ProSavantEngine_Phi9_3") |
|
|
trainer = Trainer.from_pretrained( |
|
|
"antonypamo/ProSavantEngine_Phi9_3", |
|
|
args=args, |
|
|
train_dataset=my_dataset, |
|
|
eval_dataset=my_eval |
|
|
) |
|
|
trainer.train() |
|
|
|
|
|
📦 Requirements |
|
|
torch |
|
|
transformers |
|
|
datasets |
|
|
scipy |
|
|
plotly |
|
|
gradio |
|
|
|
|
|
📚 Dataset |
|
|
|
|
|
The model was trained on the unified corpus |
|
|
antonypamo/savantorganized |
|
|
|
|
|
and linked with icosahedron_nodes.json providing the 12-node geometric structure of the icosahedral lattice. |
|
|
|
|
|
🔮 Applications |
|
|
|
|
|
Resonant rewriting and coherence scoring |
|
|
|
|
|
Prompt optimization and semantic filtration |
|
|
|
|
|
Geometric–linguistic embeddings for RRF AI models |
|
|
|
|
|
Integration into AGORA / SavantEngine resonance simulations |
|
|
|
|
|
Cognitive field modeling and symbolic AI research |
|
|
|
|
|
🧭 Related Resources |
|
|
|
|
|
antonypamo/ProSavantEngine_Phi9_2_Lite |
|
|
— prior iteration |
|
|
|
|
|
antonypamo/savantorganized |
|
|
— training corpus |
|
|
|
|
|
ProSavantEngine Resonance Space |
|
|
— live interactive demo |
|
|
|
|
|
📜 Citation |
|
|
@software{padilla2025prosavantengine, |
|
|
author = {Padilla Morales, Antony}, |
|
|
title = {ProSavantEngine Φ9.3 — Icosahedral Resonant Language Model}, |
|
|
year = {2025}, |
|
|
url = {https://huggingface.co/antonypamo/ProSavantEngine_Phi9_3} |
|
|
} |
|
|
|
|
|
⚙️ Developer Notes |
|
|
|
|
|
Add your dataset card or a link to any .jsonl corpus used. |
|
|
|
|
|
Include training_args.json for reproducibility. |
|
|
|
|
|
The model supports multi-node resonance learning via [NODE_X] tokens. |
|
|
|
|
|
Compatible with both CPU and GPU environments. |
