Add model dropdown (presets w/ VRAM) + manual override; list HauhauCS GGUF models

New model_select dropdown with suggested VRAM in each label: huihui Qwen3-VL
4B(local)/8B/30B-A3B (transformers, auto-download) and HauhauCS Qwen3.5-9B /
Qwen3.6-35B-A3B Uncensored Aggressive (GGUF). model_path is now the manual
override (empty = use dropdown). agent_bridge gains --model-select/--model-path.
The HauhauCS models are GGUF-only (no safetensors) so the transformers backend
can't load them yet — selecting one returns a clear 'GGUF backend pending'
message until the backend is added.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

agent_bridge.py

@@ -48,7 +48,8 @@ def _http_json(url: str, payload: dict | None = None, timeout: int = 30):
 
 
 def _inject(graph: dict, prompt: str, negative: str, seed: int, run_tag: str, mode: str,
-            reference_description: str = "", profile: str = ""):
+            reference_description: str = "", profile: str = "", model_select: str = "",
+            model_path: str = ""):
     """Set the receptor's prompt/seed and the judge's mode/run_tag in-place.
 
     compare mode needs a receptor (to inject the prompt). describe mode is the first
@@ -71,6 +72,10 @@ def _inject(graph: dict, prompt: str, negative: str, seed: int, run_tag: str, mo
                 inputs["reference_description"] = reference_description
             if profile:
                 inputs["profile"] = profile
+            if model_select:
+                inputs["model_select"] = model_select
+            if model_path:
+                inputs["model_path"] = model_path
     if mode == "compare" and not found_receptor:
         raise SystemExit(
             f"[agent_bridge] no '{RECEPTOR_CLASS}' node in the workflow — add the "
@@ -119,6 +124,8 @@ def main(argv=None):
     ap.add_argument("--run-tag", default="")
     ap.add_argument("--profile", default="",
                     help="analysis profile on the judge (general/oral/penetration/handjob/solo)")
+    ap.add_argument("--model-select", default="", help="judge model dropdown label (overrides workflow)")
+    ap.add_argument("--model-path", default="", help="manual judge model path/repo (overrides dropdown)")
     ap.add_argument("--ref-desc", default="",
                     help="canonical reference text to anchor compare on (from the describe pass)")
     ap.add_argument("--ref-desc-file", default="",
@@ -143,7 +150,7 @@ def main(argv=None):
         graph = json.load(f)
 
     _inject(graph, args.prompt, args.negative, args.seed, args.run_tag, args.mode, ref_desc,
-            args.profile)
+            args.profile, args.model_select, args.model_path)
 
     client_id = uuid.uuid4().hex
     try:

nodes/qwen_judge.py

@@ -41,6 +41,26 @@ RECOMMENDED_MODELS = {
     "4b": "huihui-ai/Huihui-Qwen3-VL-4B-Instruct-abliterated",
 }
 
+# Curated model dropdown (label shown in the node -> how to load it). The label
+# carries the suggested VRAM. `backend`: "transformers" = safetensors via this node
+# (auto-downloaded with snapshot_download); "gguf" = GGUF + mmproj (needs the GGUF
+# backend). `model_path` (manual) overrides this when non-empty.
+MANUAL_CHOICE = "(manual — use model_path below)"
+MODEL_PRESETS = {
+    "Qwen3-VL-4B abliterated (huihui) · local bf16 ~9GB": {
+        "repo": DEFAULT_MODEL_PATH, "backend": "transformers", "precision": "bf16"},
+    "Qwen3-VL-4B abliterated (huihui) · local fp8 ~5GB": {
+        "repo": DEFAULT_MODEL_PATH_FP8, "backend": "transformers", "precision": "fp8"},
+    "Qwen3-VL-8B abliterated (huihui) · bf16 ~17GB": {
+        "repo": "huihui-ai/Huihui-Qwen3-VL-8B-Instruct-abliterated", "backend": "transformers", "precision": "bf16"},
+    "Qwen3-VL-30B-A3B abliterated (huihui) · nf4 ~18GB (slow)": {
+        "repo": "huihui-ai/Huihui-Qwen3-VL-30B-A3B-Instruct-abliterated", "backend": "transformers", "precision": "nf4"},
+    "Qwen3.5-9B Uncensored Aggressive (HauhauCS) · GGUF Q4 ~8GB": {
+        "repo": "HauhauCS/Qwen3.5-9B-Uncensored-HauhauCS-Aggressive", "backend": "gguf", "precision": "Q4_K_M"},
+    "Qwen3.6-35B-A3B Uncensored Aggressive (HauhauCS) · GGUF Q4 ~22GB": {
+        "repo": "HauhauCS/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive", "backend": "gguf", "precision": "Q4_K_M"},
+}
+
 # Difference axes + a one-line definition each. Definitions are injected into the
 # prompt so the model fills the right axis (e.g. gender_mix = a count, not a position)
 # and the action/pose cluster is captured in detail. Fully configurable on the node;
@@ -665,7 +685,10 @@ class QwenVLImageJudge:
                 # Analysis profile: act-specialized axis set (distance-aware where it
                 # matters). `axes` below overrides it when non-empty.
                 "profile": (list(PROFILES.keys()), {"default": "general"}),
-                "model_path": ("STRING", {"default": DEFAULT_MODEL_PATH}),
+                # Curated model dropdown (label shows VRAM). model_path below overrides it.
+                "model_select": ([MANUAL_CHOICE] + list(MODEL_PRESETS.keys()),
+                                 {"default": list(MODEL_PRESETS.keys())[0]}),
+                "model_path": ("STRING", {"default": ""}),   # manual override (local dir / HF repo / alias)
                 "precision": (["bf16", "fp16", "fp8", "nf4"], {"default": "bf16"}),
                 "axes": ("STRING", {"default": "", "multiline": True}),
                 "max_new_tokens": ("INT", {"default": 1024, "min": 64, "max": 4096}),
@@ -687,7 +710,8 @@ class QwenVLImageJudge:
         }
 
     def judge(self, reference_image, mode, model_path, precision, axes,
-              max_new_tokens, temperature, swap_eval, profile="general", generated_image=None,
+              max_new_tokens, temperature, swap_eval, profile="general",
+              model_select=MANUAL_CHOICE, generated_image=None,
               keep_loaded=True, auto_download=True,
               report_dir="", run_tag="", prompt_used="", reference_description=""):
         # `axes` overrides the profile when provided; otherwise use the profile's axis set.
@@ -695,19 +719,36 @@ class QwenVLImageJudge:
         if not axis_list:
             axis_list = list(PROFILES.get(profile, PROFILES["general"]))
 
+        # Resolve the model: manual model_path overrides the dropdown preset.
+        if model_path.strip():
+            eff_repo, eff_precision = model_path.strip(), precision
+            eff_backend = "gguf" if eff_repo.lower().endswith(".gguf") else "transformers"
+        else:
+            preset = MODEL_PRESETS.get(model_select)
+            if not preset:
+                msg = "[QwenVLImageJudge] pick a model in model_select, or fill model_path."
+                print(msg); return (0.0, "{}", msg, msg, "")
+            eff_repo, eff_backend, eff_precision = preset["repo"], preset["backend"], preset["precision"]
+
+        if eff_backend == "gguf":
+            msg = (f"[QwenVLImageJudge] '{eff_repo}' is a GGUF model — the GGUF backend "
+                   f"is not enabled yet (pending Ollama vs llama-cpp-python choice). "
+                   f"Use a transformers (safetensors) model for now.")
+            print(msg); return (0.0, "{}", msg, msg, "")
+
         try:
-            resolved_path = _resolve_model_source(model_path, auto_download)
+            resolved_path = _resolve_model_source(eff_repo, auto_download)
         except Exception as e:  # missing model / download failure -> surface as score 0
             msg = str(e)
             print(msg)
             return (0.0, "{}", msg, msg, "")
 
         ref_pil = _tensor_to_pil(reference_image)
-        model, processor = _load_model(resolved_path, precision)
+        model, processor = _load_model(resolved_path, eff_precision)
 
         if mode == "describe":
             return self._describe(model, processor, ref_pil, axis_list, max_new_tokens,
-                                  temperature, resolved_path, precision, keep_loaded,
+                                  temperature, resolved_path, eff_precision, keep_loaded,
                                   report_dir, run_tag)
 
         if generated_image is None:
@@ -735,7 +776,7 @@ class QwenVLImageJudge:
                 raw_all = raw1 + "\n--- SWAPPED ---\n" + raw2
 
         if not keep_loaded:
-            _MODEL_CACHE.pop((resolved_path, precision), None)
+            _MODEL_CACHE.pop((resolved_path, eff_precision), None)
             del model
             torch.cuda.empty_cache()