Downloads · 30 days
0
geolip-flow-predictions
AbstractPhil/geolip-flow-predictions
geolip-flow-predictions is a machine learning model from AbstractPhil. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
First barrage tests and compilation tests are finished. The orbital is yielding a true representation of omega.
Access
Public
Updated Apr 2, 2026
Repo size
—
Likes
0
Public
Repo makeup
Click a slice to open those files.
.py37.6 KB · 80%
At a glance
- License
- apache-2.0
- Access
- Public
- Created
- Apr 2, 2026
- Updated
- Apr 2, 2026
- SHA
- 7c76663b
Model card
From the Hugging Face model README
First barrage tests and compilation tests are finished. The orbital is yielding a true representation of omega.
The speed and accuracy is both representationally superior and compiles safe to full-graph.
geolip_core.linalg: available
geolip.linalg backend:
CUDA: yes
Triton: 3.6.0
FL eigh: enabled
Triton SVD: enabled
GPU: NVIDIA RTX PRO 6000 Blackwell Server Edition
========================================================================
Flow Ensemble — Expanded Test Suite
========================================================================
device=cuda geolip_core.linalg=True
GPU: NVIDIA RTX PRO 6000 Blackwell Server Edition
========================================================================
1. SMOKE TEST
========================================================================
Flow Params Shape Time Conf Res norm
────────────────────── ──────── ────────────── ────────── ──────── ──────────
QuaternionFlow 70,725 (16, 64, 128) 267us 0.489 1.355
QuaternionLiteFlow 87,109 (16, 64, 128) 194us 0.471 1.282
VelocityFlow 70,210 (16, 64, 128) 175us 0.495 0.058
MagnitudeFlow 73,293 (16, 64, 128) 315us 0.526 0.601
OrbitalFlow 25,445 (16, 64, 128) 378us 0.524 1.650
AlignmentFlow 8,922 (16, 64, 128) 347us 0.485 2.905
========================================================================
2. LINALG INTEGRATION
========================================================================
Testing eigh dispatch in MagnitudeFlow and OrbitalFlow...
magnitude finite=True conf=0.469
orbital finite=True conf=0.462
Gram eigenspectrum: shape=(16, 12) range=[0.0002, 7.5887]
Eigenvector orth err: 6.86e+01
========================================================================
3. MULTI-SCALE
========================================================================
OrbitalFlow across scales:
Config B n k d Time OK
────────── ──── ───── ───── ───── ────────── ────
tiny 4 16 8 64 328us OK
small 16 64 32 128 377us OK
medium 32 128 64 256 379us OK
large 64 256 128 256 430us OK
wide 8 512 256 512 426us OK
========================================================================
4. ENSEMBLE FUSION
========================================================================
weighted: time=2.02ms norm=0.805 diversity=0.905
quaternion conf=0.489±0.002 res=1.355
quat_lite conf=0.471±0.002 res=1.285
velocity conf=0.495±0.003 res=0.058
magnitude conf=0.526±0.004 res=0.608
orbital conf=0.524±0.002 res=1.647
alignment conf=0.485±0.014 res=3.110
gated: time=2.02ms norm=0.663 diversity=0.905
quaternion conf=0.489±0.002 res=1.355
quat_lite conf=0.471±0.002 res=1.285
velocity conf=0.495±0.003 res=0.058
magnitude conf=0.526±0.004 res=0.608
orbital conf=0.524±0.002 res=1.647
alignment conf=0.485±0.014 res=3.110
residual: time=2.03ms norm=0.805 diversity=0.905
quaternion conf=0.489±0.002 res=1.355
quat_lite conf=0.471±0.002 res=1.285
velocity conf=0.495±0.003 res=0.058
magnitude conf=0.526±0.004 res=0.608
orbital conf=0.524±0.002 res=1.647
alignment conf=0.485±0.014 res=3.110
========================================================================
5. GRADIENT HEALTH
========================================================================
Flow Loss Grad norm Status
────────────────── ────────── ──────────── ────────
quaternion mse 8.90e-04 OK
quat_lite mse 6.98e-04 OK
velocity mse 3.14e-04 OK
magnitude mse 8.74e-04 OK
orbital mse 1.53e-03 OK
alignment mse 1.55e-02 OK
quaternion cosine 9.37e-02 OK
quat_lite cosine 7.39e-02 OK
velocity cosine 2.66e-02 OK
magnitude cosine 8.45e-02 OK
orbital cosine 1.91e-01 OK
alignment cosine 1.94e+00 OK
quaternion norm 8.57e-02 OK
quat_lite norm 6.37e-02 OK
velocity norm 7.90e-03 OK
magnitude norm 6.39e-02 OK
orbital norm 1.45e-01 OK
alignment norm 2.95e+00 OK
========================================================================
6. ABLATION (100 training steps, rotation target)
========================================================================
Configuration MSE Params
─────────────────────────────────── ────────── ──────────
QuaternionFlow 0.0011 280,709
QuaternionLiteFlow 0.0005 346,245
VelocityFlow 0.0061 279,682
MagnitudeFlow 0.0081 285,837
OrbitalFlow 0.0031 91,813
AlignmentFlow 0.0081 26,010
Quat + Orbital 0.0009 372,524
Velocity + Magnitude 0.0062 565,521
Orbital + Alignment 0.0026 117,825
Velocity + Orbital 0.0021 371,497
Full (weighted) 0.0007 1,310,302
Full (residual) 0.0007 1,310,302
========================================================================
7. COMPILE COMPATIBILITY
========================================================================
Flow fullgraph Raw Compiled
────────────────────── ──────────── ────────── ────────────
QuaternionFlow OK 274us 200us
QuaternionLiteFlow OK 199us 181us
VelocityFlow OK 181us 168us
MagnitudeFlow OK 328us 351us
OrbitalFlow OK 376us 347us
AlignmentFlow OK 348us 362us
========================================================================
8. MEMORY (B=32, n=128, k=64, d=256)
========================================================================
Flow Peak MB
────────────────────── ──────────
QuaternionFlow 36.2
QuaternionLiteFlow 28.3
VelocityFlow 48.0
MagnitudeFlow 24.2
OrbitalFlow 10.9
AlignmentFlow 15.0
Full ensemble 178.7
========================================================================
Done.
========================================================================
- License
- apache-2.0
- Created
- Apr 2, 2026
- Updated
- Apr 2, 2026