Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -1,3 +1,51 @@
|
|
| 1 |
-
|
| 2 |
-
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
cat > README.md << 'EOF'
|
| 2 |
+
---
|
| 3 |
+
title: Boiling Intelligence
|
| 4 |
+
emoji: 🔬
|
| 5 |
+
colorFrom: blue
|
| 6 |
+
colorTo: indigo
|
| 7 |
+
sdk: static
|
| 8 |
+
pinned: false
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# Boiling Intelligence
|
| 12 |
+
|
| 13 |
+
A minimal prototype of an autonomous AI physicist for phase-change
|
| 14 |
+
mass-transfer discovery.
|
| 15 |
+
|
| 16 |
+
## v0.4
|
| 17 |
+
|
| 18 |
+
The current synthetic benchmark demonstrates a closed scientific loop:
|
| 19 |
+
|
| 20 |
+
1. Maintain competing mass-transfer closures.
|
| 21 |
+
2. Design discriminating tests.
|
| 22 |
+
3. Query a hidden synthetic physical world.
|
| 23 |
+
4. Update evidence.
|
| 24 |
+
5. Detect model-class failure.
|
| 25 |
+
6. Infer a minimal constitutive correction.
|
| 26 |
+
7. Compile the correction into a new executable model.
|
| 27 |
+
8. Re-enter the revised model into the falsification loop.
|
| 28 |
+
|
| 29 |
+
Example discovered closure:
|
| 30 |
+
|
| 31 |
+
M3 = 0.8 ΔT + 0.00196515 ΔT²
|
| 32 |
+
|
| 33 |
+
The hidden synthetic law is used only to validate the autonomous
|
| 34 |
+
discovery architecture. This release does not claim discovery of a
|
| 35 |
+
real boiling constitutive law.
|
| 36 |
+
|
| 37 |
+
## Run
|
| 38 |
+
|
| 39 |
+
```bash
|
| 40 |
+
pip install -r requirements.txt
|
| 41 |
+
ollama serve
|
| 42 |
+
ollama pull gpt-oss:20b
|
| 43 |
+
python main_v04.py
|
| 44 |
+
|
| 45 |
+
## Roadmap
|
| 46 |
+
|
| 47 |
+
Agentic scientific reasoning
|
| 48 |
+
Conservative multiscale recurrent neural operators
|
| 49 |
+
Basilisk phase-change simulations
|
| 50 |
+
Experimental ground-truth integration
|
| 51 |
+
Closed-loop physical model discovery
|