modules/config_settings_public.py

# FILE: modules/config_settings_public.py (Hugging Face Demo - v2.0 - Verified Models)

import os
import logging
from huggingface_hub import hf_hub_download

logger = logging.getLogger("ZOTHEOS_Config")
if not logger.handlers:
    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)

# --- ✅ 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" # Smallest quantization for speed
    },
    "gemma": {
        "repo_id": "google/gemma-2b-it-gguf", # Using the official Google repository
        "filename": "gemma-2b-it.gguf"
    },
    "qwen": {
        "repo_id": "second-state/Qwen1.5-1.8B-Chat-GGUF",
        "filename": "Qwen1.5-1.8B-Chat-Q4_K_S.gguf"
    }
}

MODEL_PATHS = {}

logger.info("✅✅✅ RUNNING IN WEB DEMO MODE (Hugging Face Space) ✅✅✅")
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']}")
    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

# --- WEB-OPTIMIZED MODEL PARAMETERS ---
MODEL_SPECIFIC_PARAMS = {
    "_default": { 
        "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" },
    "gemma": { "chat_format": "gemma" },
    "qwen": { "chat_format": "chatml" }
}

# --- AGI-TIER INFERENCE & PROMPTS ---
INFERENCE_PRESETS = {"balanced": {"temperature": 0.7, "max_tokens": 512}}
DEFAULT_INFERENCE_PRESET = "balanced"
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. 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.")