lyogavin/airllm
AirLLM 2.8T LLM inference with single 4GB GPU
A star delta has been computed for 61,707 of 62,640 tracked repositories in this period, and only those can be ranked in it.
A repository missing here has not been measured far enough back yet. That does not mean it did not move.
Star counts measured between 00:30 and 06:41 (BKK).
AirLLM 2.8T LLM inference with single 4GB GPU
Fine-tune LLMs from one YAML. Layer streaming trains an 8B model on a 4 GB laptop GPU.
Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)
Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.8, DeepSeek-V4, GLM-5.1, InternLM3, Llama4, ...) and 300+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, GLM5.3, Gemma4, Llava, Phi4, ...) (AAAI 2025).
Local AI music studio unifying ACE-Step 1.5 and YuE2-3B in one Vue interface — text-to-music generation, stem separation, MIDI transcription, and LoRA fine-tuning, with a built-in multitrack DAW.
A unified framework for training, fine-tuning, and evaluating World Action Models
The cryptography-based networking stack for building unstoppable networks with LoRa, Packet Radio, WiFi and everything in between.
Fizgig — LoRA & Fine-tune Studio for Klein 9B, Krea 2, MiniMax H3, Qwen Image 2.1, Anima & Z-Image Turbo and Ace-Step XL: train, fine-tune, profile, repair and extract LoRAs & LoKR for Image, Video & Music Models
The platform for continuously improving AI agents.
Imagination Studio by keys+Wasserman powered by TensorFold: an AI film studio for Mac and NVIDIA DGX Spark. Plan a film like a crew (Slate workflow by Sam Wasserman), render Qwen-Image-2.1 / FLUX3-K pictures and MiniMax H3 / FastH3 video with sound on TensorFold.
A high-performance framework for training LLMs, VLMs, diffusion, and embodied models on NVIDIA GPUs and Kunlun XPUs. Supports Pi0.5, GR00T-N1.7, FastWAM, Cosmos3, DreamZero, DeepSeek-V4, GLM-5.3, Kimi-K3 and more.
Large-scale LLM inference engine
Android application for Meshtastic
A browser panel for the chores around ComfyUI: models with Civitai covers, trigger words and base models; Model Check finds missing models when a workflow opens; search outputs by prompt or LoRA and drag one back to restore its workflow; saved workflows, combos, prompt editing, an activity log with undo, and character voices with Anomalous TTS.
The official firmware for Meshtastic, an open-source, off-grid mesh communication system.
Native Android messaging app using Bluetooth LE, TCP, or RNode (LoRa) over LXMF and Reticulum
Portable touchscreen SDR for M5Stack Tab5 + RTL-SDR dongles — spectrum, waterfall, FM/AM, Shortwave, ADS-B, FT8, P25, LoRa, RF tools, and Wi-Fi analysis without a PC.
Handheld T Display P4 & ESP32-P4 SDR scanners - RTL-SDR spectrum, P25 Phase 1, FM/AM, ADS-B and POCSAG, plus onboard LoRa/MeshCore, GPS and Flipper .sub capture. No PC.
Self-hosted decision model (selfjev-4b) with a Jev-compatible API: Python SDK, server, fine-tuning and RLCD.
🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning.
SD-Trainer. LoRA & Dreambooth training scripts & GUI use kohya-ss's trainer, for diffusion model.
Communicate Freely
A simple mesh network communications app powered by the Reticulum Network Stack.
[TMLR26]FreeFuse: Multi-Subject LoRA Fusion via Adaptive Token-Level Routing at Test Time
The cleanest, responsive AI workspace. Run highly tuned image and video workflows flawlessly from your desktop or your phone. Built on ComfyUI.
Easily fine-tune (SFT/RL), evaluate, and deploy Qwen, Gemma, or any open agentic LLM/VLM!
JevK5: open-weight alternative to TypeSafe Jev. Typed decisions with probabilities in one forward pass; Apache-2.0 weights and code.
The LoRaMesher library implements a distance-vector routing protocol for communicating messages among LoRa nodes.
A foundation model that generates synchronized video and audio in a single model
LXMF client for Android, Linux and macOS allowing you to communicate with people or LXMF-compatible systems over Reticulum networks using LoRa, Packet Radio, WiFi, I2P, or anything else Reticulum supports.
Showing 30 of 56 repositories