Pinned Loading
-
hyperion
hyperion PublicCompositional reasoning at 100% SCAN, 99.98% PCFG, 99.75% COGS gen, 100% 1D-ARC — zero gradient descent, ~1000 lines of Python
Python
-
lemmas
lemmas PublicFour reliability primitives for any LLM API: Chain-of-Verification, self-consistency, prompt-drift detection, hedged execution.
Python
-
SOFAR
SOFAR PublicDrop-in adapter that helps any HuggingFace transformer stop forgetting things buried in long context. Bio-inspired SVD-on-residual-stream channel mapping + LoRA beam-steering attention.
Python
-
aiproof
aiproof PublicESLint for AI prompts — a static analyzer for LLM system prompts, user templates, Jinja/Mustache, YAML configs, and SDK call sites.
Makefile
-
If the problem persists, check the GitHub status page or contact support.