Skip to main content

👑 Agba Engine: Open-Source Heritage Search Engine

Agba Engine is a cross-modal Vision-Language Model (VLM) architecture designed for zero-shot heritage artifact retrieval. By mapping raw visual feature representations and authentic cultural metadata into a shared 256-dimensional Interlingua space, Agba Engine bridges high-dimensional spatial geometry and globally underrepresented heritage regalia.


🌟 Key Architectural Features

  • Dual Projection Translators: Fine-tuned linear projection heads mapped on top of OpenCLIP (ViT-B-32).
  • InfoNCE Contrastive Calibration: Trained via AdamW optimizer, driving alignment loss from 1.9409 down to 0.0003.
  • Zero-Shot Cultural Search: Query tangible artifacts using free-form natural language strings.
  • Cognitive Spatial Discrimination: Evaluates visual clarity, geometry, and occlusion over simple label memorization.

📊 Performance & Zero-Shot Validation

Training Loss Convergence (Epochs 1 ──► 50)

  • Initial InfoNCE Baseline: 1.9409
  • Final Converged Loss: 0.0003
  • Optimizer: AdamW (lr = 5e-4, weight_decay = 1e-2)

Zero-Shot Retrieval Benchmark Results

Search Query Top Match Anchor Indigenous Nomenclature Match Confidence
"Royal Wrist Beads" wrist_frame_01.jpg Ìlẹ̀kẹ̀ Ọwọ́ 67.74% 🎯
"Royal Leg Beads" leg_frame_31.jpg Ìlẹ̀kẹ̀ Ẹsẹ̀ (Macro-Left) 61.39% 🎯

🛡️ License

Distributed under the MIT License. See LICENSE for more details.

Metadata

Release files for agba-engine 5.0.9

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for agba-engine 5.0.9
File Size Uploaded Provenance
agba_engine-5.0.9.tar.gz 3.5 kB Without provenance Details

Built distribution (wheel)

Table of built distributions (wheels) for agba-engine 5.0.9
File Interpreter ABI Platform
agba_engine-5.0.9-py3-none-any.whl Python 3 none any Details

Total release size: 7.3 kB

Release files / agba_engine-5.0.9.tar.gz

Download URL agba_engine-5.0.9.tar.gz
Size 3.5 kB
Tags Source
SHA-256 checksum
How to use checksums
2e114a9cb748758eb6b7389e2ed5b410bb5d4e368e331e742096e369659df8cd
BLAKE2b-256 checksum
How to use checksums
181969238733796b293b02fda2afda24fcf8d070ca2c712597a10c6288ab01bf
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.15

Release files / agba_engine-5.0.9-py3-none-any.whl

Download URL agba_engine-5.0.9-py3-none-any.whl
Size 3.8 kB
Tags Python 3
SHA-256 checksum
How to use checksums
075cc6832d3e52b08857c0bcc5c1c745f1db4dbf2168a0ec96ec77d29a1a5c46
BLAKE2b-256 checksum
How to use checksums
a1723e5c2a67443939190d3e03865721fa749b54511ef1d5a6d6f110a376730f
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.15

Security for agba-engine 5.0.9

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page