Use transformer subfolder config for Flux2 GGUF load

Pin Flux2 GGUF config to klein base repo and pass HF token

Make run_edit inputs backward-compatible with image+prompt only

Setup unsloth FLUX.2-klein-base-4B-GGUF local CPU Basic app

Use CPU-only PyTorch wheel to keep CPU Basic build lean

Switch Space to fully local CPU Basic image edit pipeline

Switch Space to remote Qwen image edit uncensored setup for CPU Basic

Handle model-load failures with graceful chat fallback

Reduce llama memory defaults and enable verbose errors

Add explicit send button to ChatInterface

Switch Space to CPU Basic defaults

Setup ZeroGPU Space files

initial commit