Tag
Demonstrates running two 80B Qwen models simultaneously on a MacBook Pro using BBQ-FP4 quantization, claiming functionally lossless performance and speed.
GetCompress is a tool for lossless media compression without context switching.
Blog post exploring lossless recompression of GIF images using exhaustive search over LZW encoding, similar to Zopfli for PNG, to achieve smaller file sizes.
Explains two memory patterns for AI agents: GBrain (queryable company wiki) and Lossless (full conversation recording), helping agents retain and retrieve facts across and within conversations.
Orthrus is a dual-architecture framework that combines autoregressive LLM fidelity with diffusion model speed, delivering up to 7.8x speedup on Qwen3 models while guaranteeing identical output distribution.