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

README.md

@@ -1,27 +1,25 @@
 ---
-title: Qwen3.6-35B-A3B HauhauCS CPU Basic
-emoji: ⚡
-colorFrom: blue
-colorTo: gray
+title: Qwen Image Edit Uncensored CPU Basic
+colorFrom: gray
+colorTo: blue
 sdk: gradio
 sdk_version: 5.29.1
 app_file: app.py
 pinned: false
 ---
 
-Space Gradio preparado para CPU Basic usando o modelo:
-`HauhauCS/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive`
+Space configurado para CPU Basic com edicao de imagem via endpoint remoto.
 
-Configuração rápida:
-1. Em **Settings > Hardware**, ativa **CPU Basic**.
-2. Faz push destes ficheiros para o teu Space.
-3. (Opcional) Em **Variables**, troca `MODEL_FILE` para outro GGUF do repositório.
-4. Se ficar lento ou sem memória, usa quantizações mais leves e reduz `N_CTX`.
+Modelo padrao:
+`ScottzillaSystems/qwen-image-edit-plus-nsfw-lora`
 
-Variáveis opcionais:
-- `MODEL_FILE` (default: `Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive-IQ2_M.gguf`)
-- `N_CTX` (default: `4096`)
-- `MAX_NEW_TOKENS` (default: `256`)
-- `TEMPERATURE` (default: `0.7`)
-- `TOP_P` (default: `0.9`)
-- `N_GPU_LAYERS` (default: `0`)
+Como funciona:
+- O Space roda so a interface em CPU Basic.
+- A inferencia de image edit roda no provider remoto do Hugging Face.
+
+Secrets obrigatorias:
+- `HF_TOKEN` com permissao de leitura (idealmente write tambem).
+
+Variaveis opcionais:
+- `MODEL_ID` (default: `ScottzillaSystems/qwen-image-edit-plus-nsfw-lora`)
+- `MAX_SIDE` (default: `1024`)

app.py

@@ -1,94 +1,62 @@
+import io
 import os
-import threading
-
 import gradio as gr
-from huggingface_hub import hf_hub_download
-from llama_cpp import Llama
-
-REPO_ID = "HauhauCS/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive"
-MODEL_FILE = os.getenv(
-    "MODEL_FILE",
-    "Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive-IQ2_M.gguf",
-)
-SYSTEM_PROMPT = os.getenv(
-    "SYSTEM_PROMPT",
-    "You are a direct, concise assistant."
-)
-N_CTX = int(os.getenv("N_CTX", "512"))
-MAX_NEW_TOKENS = int(os.getenv("MAX_NEW_TOKENS", "256"))
-TEMPERATURE = float(os.getenv("TEMPERATURE", "0.7"))
-TOP_P = float(os.getenv("TOP_P", "0.9"))
-N_GPU_LAYERS = int(os.getenv("N_GPU_LAYERS", "0"))
-
-_llm = None
-_llm_error = None
-_model_lock = threading.Lock()
-
-
-def get_llm() -> Llama:
-    global _llm, _llm_error
-    if _llm_error is not None:
-        raise RuntimeError(_llm_error)
-    if _llm is None:
-        with _model_lock:
-            if _llm is None:
-                try:
-                    model_path = hf_hub_download(repo_id=REPO_ID, filename=MODEL_FILE)
-                    _llm = Llama(
-                        model_path=model_path,
-                        n_ctx=N_CTX,
-                        n_gpu_layers=N_GPU_LAYERS,
-                        n_threads=max((os.cpu_count() or 4) // 2, 1),
-                        n_batch=128,
-                        verbose=False,
-                        chat_format="chatml",
-                        n_ubatch=32,
-                    )
-                except Exception as e:
-                    _llm_error = str(e)
-                    raise
-    return _llm
-
-
-def chat(message: str, history):
-    try:
-        llm = get_llm()
-    except Exception:
-        return "Olá! Não consegui carregar este modelo no hardware CPU Basic atual."
-    messages = [{"role": "system", "content": SYSTEM_PROMPT}]
-
-    for user_msg, assistant_msg in history:
-        if user_msg:
-            messages.append({"role": "user", "content": user_msg})
-        if assistant_msg:
-            messages.append({"role": "assistant", "content": assistant_msg})
+from PIL import Image
+from huggingface_hub import InferenceClient
+
+MODEL_ID = os.getenv("MODEL_ID", "ScottzillaSystems/qwen-image-edit-plus-nsfw-lora")
+MAX_SIDE = int(os.getenv("MAX_SIDE", "1024"))
+
+
+def _resize_for_api(img: Image.Image) -> Image.Image:
+    img = img.convert("RGB")
+    w, h = img.size
+    m = max(w, h)
+    if m <= MAX_SIDE:
+        return img
+    scale = MAX_SIDE / m
+    nw = max(64, int(w * scale))
+    nh = max(64, int(h * scale))
+    return img.resize((nw, nh), Image.Resampling.LANCZOS)
+
+
+def edit_image(image: Image.Image, prompt: str):
+    if image is None:
+        raise gr.Error("Carrega uma imagem primeiro.")
+    if not prompt or not prompt.strip():
+        raise gr.Error("Escreve a instrução de edição.")
+
+    token = os.getenv("HF_TOKEN") or os.getenv("HUGGINGFACEHUB_API_TOKEN")
+    if not token:
+        raise gr.Error("Falta HF_TOKEN nas Secrets do Space.")
+
+    client = InferenceClient(token=token)
+    base = _resize_for_api(image)
+
+    out = client.image_to_image(
+        image=base,
+        prompt=prompt.strip(),
+        model=MODEL_ID,
+    )
 
-    messages.append({"role": "user", "content": message})
+    if isinstance(out, Image.Image):
+        return out
+    if isinstance(out, bytes):
+        return Image.open(io.BytesIO(out)).convert("RGB")
 
-    try:
-        response = llm.create_chat_completion(
-            messages=messages,
-            max_tokens=MAX_NEW_TOKENS,
-            temperature=TEMPERATURE,
-            top_p=TOP_P,
-        )
-        return response["choices"][0]["message"]["content"]
-    except Exception:
-        return "Olá! O modelo não conseguiu gerar resposta neste hardware."
+    raise gr.Error("Resposta inesperada do endpoint de image edit.")
 
 
-with gr.Blocks(theme=gr.themes.Soft()) as demo:
-    gr.Markdown("# Qwen3.6-35B-A3B (HauhauCS) on CPU Basic")
-    gr.Markdown(
-        "Primeira resposta pode demorar (download + load do GGUF). Usa Enter ou o botão Enviar."
-    )
-    gr.ChatInterface(
-        fn=chat,
-        textbox=gr.Textbox(placeholder="Escreve a tua mensagem...", lines=3),
-        submit_btn="Enviar",
-        stop_btn="Parar",
-    )
+with gr.Blocks() as demo:
+    gr.Markdown("# Qwen Image Edit (CPU Basic frontend)")
+    gr.Markdown("Este Space corre em CPU Basic e usa inferência remota para editar imagens.")
+    with gr.Row():
+        input_img = gr.Image(type="pil", label="Imagem")
+        output_img = gr.Image(type="pil", label="Resultado")
+    prompt = gr.Textbox(lines=3, label="Instrução")
+    run = gr.Button("Editar")
+    run.click(edit_image, inputs=[input_img, prompt], outputs=output_img)
 
 
 if __name__ == "__main__":
-    demo.queue(default_concurrency_limit=1).launch(show_error=True)
+    demo.launch(show_error=True)

requirements.txt

@@ -1,3 +1,3 @@
 gradio==5.29.1
 huggingface_hub>=0.31.0
-llama-cpp-python==0.3.9
+Pillow>=10.0.0