Stars
Training-free language model modification: Remove censorship, suppress slop, reduce sycophancy, and more
Self-hosted metasearch and grounding engine for AI agents, based on SearXNG. Rust, JSON API, optional page text.
Find, rank and steer MLP neurons of GGUF models on llama.cpp; export control vectors; 3D web UI and MCP server
ComfyUI node that feeds a reference image through the Qwen3.5-9B vision tower and injects its embeddings into the text prompt (MOSS-Image conditioning)
llama.cpp on a jailbroken PS5: run any GGUF model natively on the console, starting with Qwen 3.8
Japanese character-persona LoRA experiments on small LLMs (Qwen3.5 / Spark-X2.5): does the model break, and does the persona stay stable?
A server for streaming tweets by emulating browser WebPush.
ayutaz / CuteTTS-jp
Forked from OPPO-Mente-Lab/CuteTTSJapanese continual training of CuteTTS (~230M params). Reading CER 30.94% -> 7.12%, against 5.59% for real human recordings measured the same way. Adds the training code upstream does not ship, plu…
Serve the experimental DiffusionGemma-Jev diffusion model on Cloud Run GPUs (vLLM, RTX PRO 6000), with demo apps and published benchmarks.
Neural latent upscaler for Minimax H3 (24ch). Bypasses costly 5B-param VAE decode/encode. Upscale low-res latents directly, then refine. Accelerates high-res video gen, outperforms naive interp.
Extending Late-Layer KV Approximation to Mixture-of-Experts Models via Co-Routed Low-Rank Projectors
LLKVApprox experiments on Gemma 4 12B: training code, reproducibility tools, and speed-quality evaluation.
baonudesifeizhai / llama.cpp
Forked from ggml-org/llama.cppLLM inference in C/C++
Resource-aware open research on neural circuits, bodies, and simulated organisms. Starting with Drosophila.
HRFFA: High-Angle Robust Fast Face Alignment. Whole-head face alignment dataset & training pipeline.
Antidoom generates and trains targeted preference data for reducing model repetition loops (doom loops).
alfredplpl / musubi-tuner
Forked from kohya-ss/musubi-tunerfor wan full finetuning
am17an / llama.cpp
Forked from ggml-org/llama.cppLLM inference in C/C++
SPECA: Specification-to-Checklist Agentic Auditing Framework
high-performance linear attention kernel library built on TileLang
patientx / ComfyUI-Zluda
Forked from Comfy-Org/ComfyUIThe most powerful and modular stable diffusion GUI, api and backend with a graph/nodes interface. Now ZLUDA enhanced for better AMD GPU performance.
An agent-managed museum exhibit, built in Rust with Gajae-Code / LazyCodex — developed and maintained with no human intervention.
TurboMLX v0.1 Research Preview public source tree for Qwen3.5-focused MLX TurboQuant experiments.