A hub of self-hosted projects and live demos.
Chat with Qwen3-30B-A3B — a Mixture-of-Experts LLM QLoRA-fine-tuned on Japanese visual-novel dialogue. Runs 4-bit on a single RTX 3090.