Upload README.md with huggingface_hub

README.md

@@ -6,6 +6,7 @@ tags:
 - safetensors
 - vision-transformer
 - warm-restarts
+- geometric-coalescence
 library_name: pytorch
 datasets:
 - cifar10
@@ -16,80 +17,29 @@ metrics:
 
 # vit-beans-v3
 
-**Geometric Deep Learning with Cantor Multihead Fusion + AdamW Warm Restarts**
+**Geometric Deep Learning with Cantor Multihead Fusion + Shatter-Reconstruct Training**
 
-This repository contains multiple training runs using Cantor fusion architecture with pentachoron structures, geometric routing, and **CosineAnnealingWarmRestarts** for automatic exploration cycles.
+This repository contains training runs using Cantor fusion architecture with:
+- Pentachoron (5-simplex) structures for geometric routing
+- CosineAnnealingWarmRestarts for exploration cycles
+- GeometricCoalescenceLoss for shatter-reconstruct training
 
-## Training Strategy: AdamW + Warm Restarts
+### 🚀 LR Boost + Geometric Coalescence
+This run uses **restart_lr_mult = 1.15x** with **GeometricCoalescenceLoss**:
+- LR boosts create aggressive exploration cycles
+- Coalescence loss provides geometric scaffolding during weight thrashing
+- Adaptive weighting: 0.1 → 0.8 during LR spikes
+- Model reconstructs from geometric first principles when patterns shatter
 
-This model uses **AdamW with Cosine Annealing Warm Restarts** (SGDR):
-- **Drop phase**: LR decays from 0.0003 → 1e-07 over 12 epochs
-- **Restart phase**: LR jumps back to 0.0003 to explore new regions
-- **Cycle multiplier**: Each cycle is 1.75x longer than previous
-- **Benefits**: Automatic exploration + exploitation, finds better minima, robust training
-
-### 🚀 LR Boost at Restarts (NEW!)
-This run uses **restart_lr_mult = 1.2x**:
-- Normal restart: 3e-4 → 1e-7 → restart at 3e-4
-- **Boosted restart**: 3e-4 → 1e-7 → restart at 3.60e-04 (1.2x!)
-- Creates **wider exploration curves** to escape solidified local minima
-- Each restart provides progressively stronger exploration boost
-
-
-### Restart Schedule
-```
-Epochs 0-12:   LR: 0.0003 → 1e-07 (first cycle)
-Epoch 12:      LR: RESTART to 0.00035999999999999997 🔄
-Epochs 12-33.0: LR: 0.00035999999999999997 → 1e-07 (longer cycle)
-...
-```
 
 ## Current Run
 
-**Latest**: `cifar100_weighted_ADAMW_WarmRestart_boost1.2x_20251123_160217`
+**Latest**: `cifar100_learned_ADAMW_WarmRestart_boost1.15x_coal0.5_20251124_010657`
 - **Dataset**: CIFAR100
-- **Fusion Mode**: weighted
-- **Optimizer**: AdamW (adaptive moments)
-- **Scheduler**: CosineAnnealingWarmRestarts
-- **Restart LR Mult**: 1.2x
-- **Architecture**: 12 blocks, 9 heads
-- **Simplex**: 8-simplex (9 vertices)
-
-## Architecture
-
-The Cantor Fusion architecture uses:
-- **Geometric Routing**: Pentachoron (5-simplex) structures for token routing
-- **Cantor Multihead Fusion**: Multiple fusion heads with geometric attention
-- **Beatrix Consciousness Routing**: Optional consciousness-aware token fusion
-- **SafeTensors Format**: All model weights use SafeTensors (not pickle)
-
-## Usage
-```python
-from huggingface_hub import hf_hub_download
-from safetensors.torch import load_file
-
-model_path = hf_hub_download(
-    repo_id="AbstractPhil/vit-beans-v3",
-    filename="runs/YOUR_RUN_NAME/checkpoints/best_model.safetensors"
-)
-
-state_dict = load_file(model_path)
-model.load_state_dict(state_dict)
-```
-
-## Citation
-```bibtex
-@misc{vit_beans_v3,
-  author = {AbstractPhil},
-  title = {vit-beans-v3: Geometric Deep Learning with Warm Restarts},
-  year = {2025},
-  publisher = {HuggingFace},
-  url = {https://huggingface.co/AbstractPhil/vit-beans-v3}
-}
-```
+- **Fusion Mode**: learned
+- **Coalescence**: λ=0.5 ✓
+- **LR Boost**: 1.15x 🚀
 
 ---
 
 **Repository maintained by**: [@AbstractPhil](https://huggingface.co/AbstractPhil)
-
-**Latest update**: 2025-11-23 16:02:20