Havuz

Local-first desktop manager for AI image libraries

One GPU. Full LLM workflow. Real benchmarks. No cloud required.

一套功能完整、轻量稳定的 ComfyUI 实用节点集。初衷是为了平替一些断更插件的功能，会持续更新一些新开发的节点插件。有好的创意可以发Issues，我会力所能及想办法做出来。全中文界面，原生控件，不依赖第三方库。

Windows tag editor for LoRA / character image datasets: booru tag editing with Chinese UI & translation, WD14/PixAI/CL ONNX auto-tagging, LL

Locally fine-tuned Russian RAG models: document splitter, query expansion, retrieval embedder. Teacher distillation → GGUF → llama.cpp on AM

Training-data memorization auditor for fine-tuned LLMs — Trainer/TRL plugin, canary MIA + regurgitation audit, Apache-2.0

CPU-first, eager-execution tensor/autograd runtime and LLM inference engine written in pure Zig

Local-first, end-to-end encrypted agent-to-agent mesh protocol. Zero-config LAN + LoRa. Runs offline. No cloud. info@ironmesh.org

Free & Open Source AI Agent — Self-Hosted, Self-Learning, No Vendor API. Qwen2.5-7B + LoRA. 35 tools. Built on Islamic Epistemology (IHOS).

FDSpeech: Fréchet-distance-guided four-step VoxCPM2 speech synthesis

Ultra-fast TensorRT engine for ComfyUI. Features native Anima2B & Flux support with near-instant LoRA switching via NVIDIA Refit. Reforged f

Companion code for the Manning book 'LLM Customization and Fine-Tuning.' Adapt open-weights LLMs end to end: prompting and RAG, LoRA/QLoRA,

A user's facts become a few rows in a content-addressed memory table — not a rewrite of the model. Paper, code, data, and interactive site f

一个开箱即用、用于二分类任务的大语言微调模型框架。An out-of-the-box LLM fine-tuning framework for medical binary classification.

ComfyUI custom nodes for 180+ hosted AI models: Seedance, Kling, Wan, MiniMax H3, Vidu, LTX, Suno, GPT Image, Nano Banana, Seedream, FLUX Lo

Whisper-large-v3 fine-tune for Indonesian + Javanese + Sundanese ASR. Lhotse + k2 lattice rescoring pipeline.

五引擎语音合成：VoxCPM2 / IndexTTS 2.5 / 2.0 / OpenVoice / Step-Audio-EditX，声音克隆、声音设计、剧本配音、流式生成，中英日韩界面 | Multi-engine TTS (5 engines): voice cloning

Distil an expensive LLM API call on a narrow task into a small local model: capture traffic, curate, LoRA fine-tune with MLX on Apple Silico