Update modules/config_settings_public.py

modules/config_settings_public.py

@@ -1,40 +1,38 @@
-# FILE: modules/config_settings_public.py (Hugging Face Demo)
+# FILE: modules/config_settings_public.py (Hugging Face Demo - v2.0 - Verified Models)
 
 import os
-import sys
 import logging
 from huggingface_hub import hf_hub_download
 
 logger = logging.getLogger("ZOTHEOS_Config")
 if not logger.handlers:
-    handler = logging.StreamHandler(sys.stdout)
+    handler = logging.StreamHandler()
     formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - [%(funcName)s] - %(message)s')
     handler.setFormatter(formatter)
     logger.addHandler(handler)
     logger.setLevel(logging.INFO)
 
-# --- ✅ WEB-OPTIMIZED MODEL SOURCES ---
-# These models are smaller and faster, perfect for a CPU-based web demo.
+# --- ✅ VERIFIED WEB-OPTIMIZED MODEL SOURCES ---
+# These models are smaller, faster.
 MODEL_DEFINITIONS = {
     "mistral": {
         "repo_id": "TheBloke/Mistral-7B-Instruct-v0.2-GGUF",
-        "filename": "mistral-7b-instruct-v0.2.Q2_K.gguf" # Using a smaller Q2_K quantization for speed
+        "filename": "mistral-7b-instruct-v0.2.Q2_K.gguf" # Smallest quantization for speed
     },
     "gemma": {
-        "repo_id": "TheBloke/gemma-2b-it-GGUF",
-        "filename": "gemma-2b-it.Q4_K_M.gguf" # Using the 2B parameter version
+        "repo_id": "google/gemma-2b-it-gguf", # Using the official Google repository
+        "filename": "gemma-2b-it.gguf"
     },
     "qwen": {
-        "repo_id": "TheBloke/Qwen1.5-1.8B-Chat-GGUF",
-        "filename": "qwen1.5-1.8b-chat.Q4_K_M.gguf" # Using the 1.8B parameter version
+        "repo_id": "Qwen/Qwen1.5-1.8B-Chat-GGUF", # Using the official Qwen repository
+        "filename": "qwen1_5-1.8b-chat-q4_k_m.gguf"
     }
 }
 
 MODEL_PATHS = {}
 
-# This logic will only run when the code is on Hugging Face Spaces.
 logger.info("✅✅✅ RUNNING IN WEB DEMO MODE (Hugging Face Space) ✅✅✅")
-N_GPU_LAYERS_FALLBACK = 0  # Force CPU-only mode for Hugging Face free tier
+N_GPU_LAYERS_FALLBACK = 0  # Force CPU-only mode
 
 for name, model_info in MODEL_DEFINITIONS.items():
     logger.info(f"Downloading model for demo: {name} from {model_info['repo_id']}")
@@ -45,12 +43,12 @@ for name, model_info in MODEL_DEFINITIONS.items():
         logger.error(f"❌ FAILED to download {name}: {e}")
         raise e
 
-# --- ✅ WEB-OPTIMIZED MODEL PARAMETERS ---
+# --- WEB-OPTIMIZED MODEL PARAMETERS ---
 MODEL_SPECIFIC_PARAMS = {
     "_default": { 
-        "n_gpu_layers": N_GPU_LAYERS_FALLBACK, # Ensures CPU usage
-        "n_ctx": 4096,           # Smaller context window for lower RAM usage
-        "n_batch": 512,          # Standard batch size for CPU
+        "n_gpu_layers": N_GPU_LAYERS_FALLBACK,
+        "n_ctx": 2048, # Smaller context for lower RAM
+        "n_batch": 512,
         "verbose": True
     },
     "mistral": { "chat_format": "mistral-instruct" },
@@ -58,21 +56,15 @@ MODEL_SPECIFIC_PARAMS = {
     "qwen": { "chat_format": "chatml" }
 }
 
-# --- ✅ TIER INFERENCE PRESETS ---
-INFERENCE_PRESETS = {
-    "balanced": {"temperature": 0.7, "top_p": 0.9, "max_tokens": 1024, "repeat_penalty": 1.1 },
-    "precise": {"temperature": 0.2, "top_p": 0.7, "top_k": 20, "max_tokens": 1536, "repeat_penalty": 1.05 },
-    "creative": {"temperature": 0.9, "top_p": 0.95, "top_k": 60, "max_tokens": 1024, "repeat_penalty": 1.15 }
-}
+# --- AGI-TIER INFERENCE & PROMPTS ---
+INFERENCE_PRESETS = {"balanced": {"temperature": 0.7, "max_tokens": 512}}
 DEFAULT_INFERENCE_PRESET = "balanced"
-
-# --- ✅ TIER MODEL ROLES AND PROMPTS ---
-DEFAULT_SYSTEM_PROMPT = "You are ZOTHEOS, an ethical AI developed to help humanity. Be clear, respectful, and helpful. Respond only in English."
+DEFAULT_SYSTEM_PROMPT = "You are ZOTHEOS, an ethical AI. Respond only in English."
 MODEL_ROLES = {"mistral": "analyst", "gemma": "humanist", "qwen": "skeptic"}
 MODEL_ROLE_SYSTEM_PROMPTS = {
-    "analyst": "You are an impartial analyst. Provide structured, logical insights. Break down complex topics into clear, understandable points. Respond only in English.",
-    "humanist": "You are an empathetic and values-driven AI. Focus on the emotional, ethical, and societal impact. Consider the human element above all. Respond only in English.",
-    "skeptic": "You are a respectful but rigorous skeptic. Challenge the user's premise, identify potential risks, question assumptions, and explore alternative interpretations. Respond only in English.",
+    "analyst": "You are an impartial analyst. Provide structured, logical insights. Respond only in English.",
+    "humanist": "You are an empathetic AI. Focus on the emotional and ethical impact. Respond only in English.",
+    "skeptic": "You are a respectful skeptic. Question assumptions and highlight risks. Respond only in English.",
 }
 
 logger.info("✅ Hugging Face Demo Configuration Loaded Successfully.")