Update README.md
Browse files
README.md
CHANGED
|
@@ -6,6 +6,102 @@ colorTo: purple
|
|
| 6 |
sdk: static
|
| 7 |
pinned: false
|
| 8 |
thumbnail: >-
|
| 9 |
-
https://cdn-uploads.huggingface.co/production/uploads/6631ebb1060ab1f6667d1f36/
|
| 10 |
---
|
| 11 |
-
|
| 12 |
+
|
| 13 |
+
# Ars AI
|
| 14 |
+
|
| 15 |
+
> *Explore. Analyze. Understand. Advance.*
|
| 16 |
+
|
| 17 |
+
---
|
| 18 |
+
|
| 19 |
+
## About
|
| 20 |
+
|
| 21 |
+
**Ars AI** is a one-person (for now), independent AI research initiative exploring the practical boundaries of open-source generative models.
|
| 22 |
+
|
| 23 |
+
This organization exists primarily as my personal research laboratory—a place to experiment, break things, rebuild them, and occasionally discover something genuinely useful along the way.
|
| 24 |
+
|
| 25 |
+
Most projects begin with curiosity rather than a roadmap. If an idea sounds interesting enough, it's probably getting trained.
|
| 26 |
+
|
| 27 |
+
---
|
| 28 |
+
|
| 29 |
+
## Research Areas
|
| 30 |
+
|
| 31 |
+
- Stable Diffusion
|
| 32 |
+
- Model Merging
|
| 33 |
+
- Cyclical Training
|
| 34 |
+
- Fine-Tuning
|
| 35 |
+
- Prompt Analysis
|
| 36 |
+
- Experimental Workflows
|
| 37 |
+
|
| 38 |
+
---
|
| 39 |
+
|
| 40 |
+
## Philosophy
|
| 41 |
+
|
| 42 |
+
> **Knowledge is defined not by what is known, but by what remains absent.**
|
| 43 |
+
|
| 44 |
+
Every failed experiment teaches something.
|
| 45 |
+
Every successful experiment raises even more questions.
|
| 46 |
+
|
| 47 |
+
---
|
| 48 |
+
|
| 49 |
+
## Current Projects
|
| 50 |
+
|
| 51 |
+
### Aeterna Opus
|
| 52 |
+
|
| 53 |
+
Experimental research into model merging methodologies, cyclical training strategies, and next-generation Stable Diffusion architectures.
|
| 54 |
+
|
| 55 |
+
### Ars Divina
|
| 56 |
+
|
| 57 |
+
An ongoing family of Stable Diffusion models focused on illustration quality, stylistic versatility, and iterative refinement.
|
| 58 |
+
|
| 59 |
+
### Indominus Rex XL
|
| 60 |
+
|
| 61 |
+
Another ongoing family of Stable Diffusion models focused on illustration quality, anamotical-correctness, and 'RBG' goodies.
|
| 62 |
+
|
| 63 |
+
---
|
| 64 |
+
|
| 65 |
+
## Methodology
|
| 66 |
+
|
| 67 |
+
Ars AI favors experimentation over assumptions.
|
| 68 |
+
|
| 69 |
+
If a hypothesis looks ridiculous but technically possible...
|
| 70 |
+
|
| 71 |
+
...it's probably worth testing.
|
| 72 |
+
|
| 73 |
+
---
|
| 74 |
+
|
| 75 |
+
## Open Source
|
| 76 |
+
|
| 77 |
+
Everything developed under Ars AI is intended to contribute back to the open-source AI community whenever practical.
|
| 78 |
+
|
| 79 |
+
Research notes, model releases, workflows, and unexpected discoveries are shared whenever they're useful to others.
|
| 80 |
+
|
| 81 |
+
---
|
| 82 |
+
|
| 83 |
+
## FAQ
|
| 84 |
+
|
| 85 |
+
**Q: How many researchers are in Ars AI?**
|
| 86 |
+
|
| 87 |
+
A: One, for now.
|
| 88 |
+
|
| 89 |
+
**Q: Who reviews your ideas?**
|
| 90 |
+
|
| 91 |
+
A: Usually the GPU. Sometimes reality.
|
| 92 |
+
|
| 93 |
+
**Q: What's the mission?**
|
| 94 |
+
|
| 95 |
+
A: Officially:
|
| 96 |
+
|
| 97 |
+
> Advance open-source AI through empirical experimentation and transparent research.
|
| 98 |
+
|
| 99 |
+
Unofficially:
|
| 100 |
+
|
| 101 |
+
> *Doing model merges and cyclical training for the shits and giggles.*
|
| 102 |
+
|
| 103 |
+
Surprisingly, that has produced some decent models.
|
| 104 |
+
|
| 105 |
+
---
|
| 106 |
+
|
| 107 |
+
*"Beyond the Visible."*
|