I fear a future where access to critical information systems is decided purely by socioeconomic status.
|
Diffusion pipelines, LoRA fine-tuning, abliteration, transcript generation, model conversion, and desktop front-ends for local inference. |
Self-improving cognitive architectures, memory and reflection loops, outcome tracking, and agent tooling. |
|
Turn-based tycoon and builder games, plus save editors and server admin tools for the communities around them. |
Offline transcription, voice companions, quantization helpers, transcribing suites and anything else that removes a round trip to someone else's server. |
|
Every night an autonomous AI agent finds a limitation in its own cognition, builds a working fix, deploys it to its live infrastructure, and publishes what it learned. A long-running experiment in machine self-improvement - 60+ self-built cognitive subsystems.
|
Model-agnostic abliteration toolkit for Hugging Face causal LMs. Refusal directions computed and removed via weight orthogonalization or runtime hooks, with KL-aware tuning and a local web UI.
|
A modular LoRA fine-tuning pipeline for Hugging Face models. Data cleaning, smart early stopping, automatic mode detection, and cross-platform support in one flow.
|