Fine-tune LLMs from one YAML. Layer streaming trains an 8B model on a 4 GB laptop GPU.
Turkish voice design with VoxCPM2 on a 4 GB GPU: describe a voice, pick a variant, read Turkish text. Patched audio.cpp, local UI, 7 synthetic voices.
The most efficient one-page LoRA trainer for Anima 2B. Optimized for 6GB+ VRAM, featuring a smart dataset analyzer and real-time previews.
WeeLLM runs large diffusion models with as little as 4 GB of VRAM, without any quantization. It dynamically determines how many layers can fit within the available VRAM and streams the text encoder and transformer layers to the GPU layer by layer, enabling inference on hardware with limited VRAM. It supports both safetensors and GGUF models.
One-click Windows installer for Z-Image Turbo AI image generation. Optimized for low-VRAM GPUs (4GB+). Features Gradio web UI, automatic setup, and GGUF model support.
Train LoRAs for Qwen-Image 2.1, FLUX.2 Klein 9B, FLUX.3 Image Klein, Krea 2, Z-Image, MOSS-Image, ERNIE-Image, Ideogram 4, Anima, SDXL (Pony, Illustrious, NoobAI), LTX 2.3, MiniMax-H3, and more on NVIDIA GPUs from 4–8 GB VRAM. One install, one launcher, NF4 models. Windows, Linux and RunPod.
A ComfyUI Workflow for low vram users
Hierarchical RAG architecture scaling to 693K chunks on consumer hardware (4GB VRAM). Features 3-address routing, hybrid vector+graph fusion, and SetFit classification.
ComfyUI nodes for MiniMax H3: INT8 X2 VAE and asynchronous NVENC saving. 5-second 1088x1920 video with audio in about 60 s on an RTX 4070 12GB, about 30% faster than the FP16 X2 VAE.
"Adaptive Hybrid Quantization Framework for deploying 7B+ LLMs on low-VRAM devices (e.g., GTX 1050). Features surgical block alignment and Numba-accelerated inference.
Taiwanese Hokkien (Taigi) speech-to-text transcriber - MediaTek Breeze-ASR-26 with faster-whisper, tuned for RTX 3050 4GB low-VRAM GPUs. Gradio UI, CLI, Docker, SRT/VTT/TXT/JSON.
在 8 GB 显存的显卡上跑 35B MoE（Cyber-Tiel-Coder-35B-A3B）：专家权重流式上卡、未命中专家由 CPU 池就地计算。Strata 引擎的 fork（MIT）。RTX 4080S 实测 8 GB 档 57.7–63.5 tok/s、全常驻 134–136 tok/s。Build from THIS repo — upstream does not support this model.
Three MiniMax H3 ComfyUI workflows for 8GB laptop GPUs (20 / 8 / 4 steps). Includes the model list, launch flags, prompt-structure pitfalls, the resolution trap, and measured evidence for why you must restart ComfyUI before every run.
MiniMax H3 video generation on 2080Ti 22G (Turing sm_75) - field-tested handbook, workflows & FAQ
ComfyUI deployment kit for Qwen-Image-2.1 Uncensored (GGUF): ready-to-use t2i & image-edit workflows, ComfyUI-GGUF architecture patches, and a 14-second lossless Q8_0 quantization repair tool that fixes 'weight shape [136] vs normalized_shape [128]'. Runs on 8 GB VRAM. Weights NOT included.
Type a prompt or drop up to 10 photos: Qwen-Image-2.1 renders it on your 12 GB GPU in ~35 s. Desktop app + CLI, honest benchmark vs gpt-image-2.
SCAIL-2 (Wan 2.1) Low-VRAM motion transfer — endless videos from a single reference image. 8+ GB GPUs.
MiniMax-H3 本地部署·NVIDIA 显卡（8GB 显存起）：一键脚本 + 15+ 模型选型 + 9 套 ComfyUI 工作流，文生视频/图生视频/参考图锁角色 | Local MiniMax-H3 video generation on NVIDIA GPUs from 8GB VRAM
Modly extension for Hunyuan3D 2.1 patched for Windows AMD and modest PCs
To associate your repository with the low-vram topic, visit your repo's landing page and select "manage topics."