Update README.md
Browse files
README.md
CHANGED
|
@@ -1,10 +1,48 @@
|
|
| 1 |
---
|
| 2 |
-
title:
|
| 3 |
-
emoji:
|
| 4 |
colorFrom: gray
|
| 5 |
colorTo: blue
|
| 6 |
sdk: static
|
| 7 |
pinned: false
|
| 8 |
---
|
| 9 |
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
title: Veltraxor AI
|
| 3 |
+
emoji: β‘
|
| 4 |
colorFrom: gray
|
| 5 |
colorTo: blue
|
| 6 |
sdk: static
|
| 7 |
pinned: false
|
| 8 |
---
|
| 9 |
|
| 10 |
+
# Veltraxor AI
|
| 11 |
+
|
| 12 |
+
**Veltraxor AI** is a private foundation model stack led by **Libo Wang**, focused on advanced reasoning, cognitive architectures, and multimodal integration.
|
| 13 |
+
Our research and deployment stack emphasizes **Dynamic Chain-of-Thought (D-CoT)** and **Graph-of-Causal Evolution (GoCE)** β two original frameworks introduced in 2025.
|
| 14 |
+
|
| 15 |
+
---
|
| 16 |
+
|
| 17 |
+
## π¬ Research Focus
|
| 18 |
+
- **Reasoning Architectures**: Dynamic CoT, GoCE, controllable multi-hop reasoning.
|
| 19 |
+
- **Fine-tuning Methods**: LoRA, QLoRA, adaptive adapter variants.
|
| 20 |
+
- **Multimodal Integration**: Text, documents, and images via unified extraction pipelines.
|
| 21 |
+
- **Evolutionary Control**: Causal-graph-based self-evolution for long-context stability.
|
| 22 |
+
- **Super Coding**: Repository-grounded code synthesis with minimal diffs.
|
| 23 |
+
|
| 24 |
+
---
|
| 25 |
+
|
| 26 |
+
## π¦ Core Model
|
| 27 |
+
- **Veltraxor-1**: Advanced multimodal reasoning model stack.
|
| 28 |
+
- Public repo (weights/config): [Veltraxor/Veltraxor_1](https://huggingface.co/Veltraxor/Veltraxor_1)
|
| 29 |
+
- Private repo (fine-tuning & deployment assets): [Veltraxor/veltraxor-1](https://huggingface.co/Veltraxor/veltraxor-1)
|
| 30 |
+
|
| 31 |
+
---
|
| 32 |
+
|
| 33 |
+
## π References
|
| 34 |
+
- [Dynamic Chain-of-Thought (D-CoT) β arXiv:2502.10428](https://doi.org/10.48550/arXiv.2502.10428)
|
| 35 |
+
- [Graph-of-Causal Evolution (GoCE) β arXiv:2506.07501](https://doi.org/10.48550/arXiv.2506.07501)
|
| 36 |
+
- [Veltraxor Prototypes (GitHub)](https://github.com/brucewang123456789/GeniusTrail)
|
| 37 |
+
|
| 38 |
+
---
|
| 39 |
+
|
| 40 |
+
## π Acknowledgements
|
| 41 |
+
Veltraxor-1 builds on **DeepSeek R1** open-source weights released under the MIT License.
|
| 42 |
+
We acknowledge the DeepSeek-AI team and the broader open-source ecosystem (PyTorch, Hugging Face, SentenceTransformers, pgVector, etc.) for enabling this work.
|
| 43 |
+
|
| 44 |
+
---
|
| 45 |
+
|
| 46 |
+
## π¬ Contact
|
| 47 |
+
For collaborations, partnerships, or inquiries:
|
| 48 |
+
**free.equality.anyone@gmail.com**
|