itigges22/ATLAS
ATLAS Geometric Lens: Training Data & Models What is ATLAS? ATLAS (Adaptive Test-time Learning and Autonomous Specialization) is a fully self-hosted coding assistant that runs on a single consumer GPU. It wraps a frozen open-weight model in an inference scaffold (constraint-driven generation, energy-based verification, self-verified repair) instead of fine-tuning or routing to a hosted API. The headline result is 74.6% pass@1-v(k=3) on LiveCodeBench v5 using a… See the full description on the dataset page: https://huggingface.co/datasets/itigges22/ATLAS.
| Author / organization | itigges22 |
|---|---|
| License (as declared) | agpl-3.0 |
| Downloads | 695 (as reported by Hugging Face) |
| Likes | 2 |
| Created | 5 April 2026 |
| Last updated at source | 2 July 2026 |
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 16 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.