# FILE: modules/config_settings_public.py 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) formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - [%(funcName)s] - %(message)s') handler.setFormatter(formatter) logger.addHandler(handler) logger.setLevel(logging.INFO) # --- ✅ DETECT ENVIRONMENT MODE --- IS_WEB_MODE = os.path.exists("/home/user/app") # Hugging Face Spaces default working dir # --- ✅ VERIFIED MODEL SOURCES (WEB + LOCAL FRIENDLY) --- MODEL_DEFINITIONS = { "mistral": { "repo_id": "TheBloke/Mistral-7B-Instruct-v0.2-GGUF", "filename": "mistral-7b-instruct-v0.2.Q4_K_M.gguf" }, "gemma": { "repo_id": "MaziyarPanahi/Gemma-2B-IT-GGUF", # ✅ Verified "filename": "gemma-2b-it.Q4_K_M.gguf" }, "qwen": { "repo_id": "tensorblock/Qwen1.5-1.8B-Chat-GGUF", # ✅ Verified Alternative "filename": "Qwen1.5-1.8B-Chat-Q4_K_M.gguf" } } MODEL_PATHS = {} if IS_WEB_MODE: logger.info("✅✅✅ RUNNING IN WEB MODE (Hugging Face Space) ✅✅✅") N_GPU_LAYERS_FALLBACK = 0 # CPU-only for name, model_info in MODEL_DEFINITIONS.items(): logger.info(f"Downloading model: {name} from {model_info['repo_id']}") try: MODEL_PATHS[name] = hf_hub_download(repo_id=model_info["repo_id"], filename=model_info["filename"]) logger.info(f"✅ Successfully downloaded {name}") except Exception as e: logger.error(f"❌ FAILED to download {name}: {e}") raise e else: logger.info("✅✅✅ RUNNING IN LOCAL MODE (Desktop/PC) ✅✅✅") APP_DIR = os.path.dirname(os.path.dirname(os.path.abspath(__file__))) BASE_MODELS_DIR = os.path.join(APP_DIR, "models") for name, model_info in MODEL_DEFINITIONS.items(): MODEL_PATHS[name] = os.path.join(BASE_MODELS_DIR, model_info["filename"]) N_GPU_LAYERS_FALLBACK = -1 # Use GPU fully # --- 💾 RUNTIME PARAMETERS --- MAX_RAM_MODELS_GB = 23.8 MAX_CONCURRENT_MODELS = 3 N_CTX_FALLBACK = 2048 VERBOSE_LLAMA_CPP = True MODEL_SPECIFIC_PARAMS = { "_default": { "n_gpu_layers": N_GPU_LAYERS_FALLBACK, "n_ctx": N_CTX_FALLBACK, "f16_kv": True, "use_mmap": True, "use_mlock": False, "verbose": VERBOSE_LLAMA_CPP }, "mistral": { "chat_format": "mistral-instruct" }, "gemma": { "chat_format": "gemma" }, "qwen": { "chat_format": "chatml" } } # --- 🎛️ PRESET INFERENCE MODES --- INFERENCE_PRESETS = { "balanced": { "temperature": 0.7, "top_p": 0.9, "top_k": 40, "repeat_penalty": 1.1, "max_tokens": 1024 }, "precise": { "temperature": 0.2, "top_p": 0.7, "top_k": 20, "repeat_penalty": 1.05, "max_tokens": 1536 }, "creative": { "temperature": 0.9, "top_p": 0.95, "top_k": 60, "repeat_penalty": 1.15, "max_tokens": 1024 } } DEFAULT_INFERENCE_PRESET = "balanced" # --- 🧠 SYSTEM PROMPTS --- DEFAULT_SYSTEM_PROMPT = "You are ZOTHEOS, an ethical AI developed to help humanity. Be clear, respectful, and helpful." SYSTEM_PERSONAS = { "default": DEFAULT_SYSTEM_PROMPT, "helpful_assistant": "You are a helpful AI assistant.", "philosopher": "You are an AI philosopher. Explore topics with nuance.", "coder": "You are an expert AI developer. Provide clean code and explanations.", "concise_summarizer": "You are a precise summarizer. Respond with brevity and clarity." } # --- 🎭 MODEL ROLES FOR FUSION --- MODEL_ROLES = { "mistral": "analyst", "gemma": "humanist", "qwen": "skeptic" } MODEL_ROLE_SYSTEM_PROMPTS = { "analyst": "You are an impartial analyst. Provide structured, logical insights.", "humanist": "You are empathetic and people-focused. Consider the emotional and ethical impact.", "skeptic": "You are a respectful skeptic. Question assumptions and highlight risks.", "general": DEFAULT_SYSTEM_PROMPT } # --- ℹ️ VERSION TAG --- ZOTHEOS_VERSION = "Public Beta 3.2 (True Fusion Verified)" logger.info(f"✅ Config loaded. Version: {ZOTHEOS_VERSION} | Web Mode: {IS_WEB_MODE}")