Downloads · 30 days
0
TITAN-10-Intelligence
ShiroOnigami23/TITAN-10-Intelligence
TITAN-10-Intelligence is a machine learning model from ShiroOnigami23. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for titan-10. The card lists the license as mit.
This is a from-scratch implementation of a High-Performance Deep Learning Engine. It bypasses standard frameworks (PyTorch/TensorFlow) to use a custom Virtual Memory Manager (VMM) and a Tape-based Autograd system.
Access
Public
Updated Jan 11, 2026
Repo size
210 KB
Likes
4
Public
Repo makeup
Click a slice to open those files.
.pkl123 KB · 51%
At a glance
- Library
- titan-10
- License
- mit
- Access
- Public
- Created
- Jan 5, 2026
- Updated
- Jan 11, 2026
- SHA
- 8c16555f
Model card
From the Hugging Face model README
ShiroOnigami23: TITAN-10 Intelligence Engine
This is a from-scratch implementation of a High-Performance Deep Learning Engine. It bypasses standard frameworks (PyTorch/TensorFlow) to use a custom Virtual Memory Manager (VMM) and a Tape-based Autograd system.
Key Features:
- Zero-Copy Autograd: Minimal memory overhead via pre-allocated VMM buffers.
- Flash-Attention-2: Tiled attention implementation for $O(N)$ memory efficiency.
- RoPE: Rotary Positional Embeddings for infinite sequence context.
- LoRA Adaptation: Intelligence contained in a ~120KB Low-Rank weight matrix.
Architecture
Designed to surpass the CMU 11-785 (Deep Learning) curriculum within a 10-day execution window.
Final Audit
- Effective Rank: 32 (Verified via SVD)
- Spectral Energy (First 4 Dims): 16.79%
- Precision: FP16 Safetensors
- Library
- titan-10
- License
- mit
- Languages
- en
- Created
- Jan 5, 2026
- Updated
- Jan 11, 2026