Popular repositories Loading

1. llama.cpp-turboq-mtp llama.cpp-turboq-mtp Public

   Qwen4-Exp (Qwen3.8-Flash-Next) SWA + MTP + PLE-on-iswa — bounded deep-context decode w/ long-range recall, TBQ4 4.25 bpv KV, DSpark + RotorQuant. Main (master) = Qwen4-Exp build; the Qwen35 SWA hyb…

   C++ 96 14

2. Xolotl Xolotl Public

   Xolotl (formerly Strata-DSV4.1): DeepSeek-V4.1-Flash, all 552B, on one gaming PC (RTX 4090 + 90 GB RAM) at ~27 tok/s, plus GLM-5.3-Flash, Qwen3.8-Flash-Next and Qwen3.8-27B. A heavily extended fork…

   C++ 93 12

3. GLM-4.7-Flash-Rapport GLM-4.7-Flash-Rapport Public

   Fast GLM-4.7-Flash wrapper for Claude Code with Google Search MCP and Vision Proxy

   JavaScript 10 5

4. llama.cpp-TurboQuant-DSV4 llama.cpp-TurboQuant-DSV4 Public

   llama.cpp fork: TurboQuant KV (TBQ3/TBQ4) + RotorQuant + MTP + DSV4 native TBQ4 KV via dequant-at-read (1M context on 96GB, zero swap)

   C++ 2

5. ComfyUI-H3O ComfyUI-H3O Public

   MiniMax H3 node pack by Indra's Mirror — H3 Prompt Enhancer + H3 Aspect Ratio Detector

   Python 1

6. Strata-GLM Strata-GLM Public

   GLM-5.3-Flash (320B MoE) on one gaming PC: 25 tok/s decode, ~400 tok/s prefill, 512K context on a 24 GB GPU + 90 GB RAM. Maya-S-v2 + runtime abliteration LoRA, OpenAI/Anthropic API. A Strata engine…

   C++ 1