pi‑Qwen Demo with Custom LoRA
This demo showcases the pi‑Flow: Policy‑Based Few‑Step Generation via Imitation Distillation technique applied to the Qwen‑Image base model. In addition to the fast π‑Flow policy, this version also loads a user‑provided LoRA from lichorosario/dott‑3000‑qwen‑image‑lora to stylize the outputs.
Base model: Qwen‑Image. Fast policy: GMFlow. Code: https://github.com/Lakonik/piFlow.
64 2048
64 2048
4 50