Update modules/config_settings_public.py

modules/config_settings_public.py

@@ -1,4 +1,4 @@
-# FILE: modules/config_settings_public.py (Definitive Final Version)
+# FILE: modules/config_settings_public.py (Definitive Final Version 2 - Using TheBloke for Gemma)
 
 import os
 import sys
@@ -13,18 +13,17 @@ if not logger.handlers:
     logger.addHandler(handler)
     logger.setLevel(logging.INFO)
 
-# --- A more reliable way to detect if we are running in a Hugging Face Space ---
 IS_WEB_MODE = os.path.exists("/home/user/app")
 
-# --- Define Model Repositories and Filenames ONCE ---
+# --- ✅✅✅ SWITCHED GEMMA TO A MORE RELIABLE PROVIDER: TheBloke ✅✅✅ ---
 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": "google/gemma-2b-it-gguf",  # ✅✅✅ THE ONLY CHANGE IS THIS LINE (GGUF -> gguf) ✅✅✅
-        "filename": "gemma-2b-it.Q4_K_M.gguf"
+        "repo_id": "TheBloke/gemma-2B-it-GGUF", # Switched to TheBloke's repo
+        "filename": "gemma-2b-it.Q4_K_M.gguf"   # Using the same quality file
     },
     "qwen": {
         "repo_id": "Qwen/Qwen1.5-1.8B-Chat-GGUF",
@@ -32,14 +31,10 @@ MODEL_DEFINITIONS = {
     }
 }
 
-# --- Initialize MODEL_PATHS dictionary ---
 MODEL_PATHS = {}
 
-# --- Set up paths and GPU layers based on environment ---
 if IS_WEB_MODE:
     logger.info("✅✅✅ RUNNING IN WEB MODE (Hugging Face Space) ✅✅✅")
-    logger.info("Model paths will be resolved by hf_hub_download.")
-    
     for name, model_info in MODEL_DEFINITIONS.items():
         logger.info(f"Downloading model: {name} from repo: {model_info['repo_id']}")
         try:
@@ -48,20 +43,14 @@ if IS_WEB_MODE:
         except Exception as e:
             logger.error(f"❌ FAILED to download model {name}. Error: {e}")
             raise e
-
     N_GPU_LAYERS_FALLBACK = 0
-    logger.info("N_GPU_LAYERS_FALLBACK forced to 0 for CPU-only web environment.")
-
 else: # LOCAL MODE
     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")
-    logger.info(f"Models will be loaded from local directory: {BASE_MODELS_DIR}")
     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
-    logger.info("N_GPU_LAYERS_FALLBACK set to -1 for local GPU acceleration.")
-
 
 # --- Shared Configurations ---
 MAX_RAM_MODELS_GB = 23.8
@@ -76,25 +65,23 @@ MODEL_SPECIFIC_PARAMS = {
     "qwen": { "chat_format": "chatml" },
     "_default": {
         "f16_kv": True, "use_mmap": True, "use_mlock": False,
-        "verbose": VERBOSE_LLAMA_CPP,
-        "n_gpu_layers": N_GPU_LAYERS_FALLBACK,
-        "n_threads": N_THREADS_FALLBACK,
-        "n_ctx": N_CTX_FALLBACK
+        "verbose": VERBOSE_LLAMA_CPP, "n_gpu_layers": N_GPU_LAYERS_FALLBACK,
+        "n_threads": N_THREADS_FALLBACK, "n_ctx": N_CTX_FALLBACK
     }
 }
 
 INFERENCE_PRESETS = {
     "balanced": {"temperature": 0.7, "top_p": 0.9, "top_k": 40, "repeat_penalty": 1.1, "mirostat_mode": 0, "max_tokens": 1024},
     "precise": {"temperature": 0.2, "top_p": 0.7, "top_k": 20, "repeat_penalty": 1.05, "mirostat_mode": 0, "max_tokens": 1536},
-    "creative": {"temperature": 0.9, "top_p": 0.95, "top_k": 60, "repeat_penalty": 1.15, "mirostat_mode": 2, "mirostat_tau": 4.0, "mirostat_eta": 0.1, "max_tokens": 1024},
-    "passthrough": {}
+    "creative": {"temperature": 0.9, "top_p": 0.95, "top_k": 60, "repeat_penalty": 1.15, "mirostat_mode": 2, "mirostat_tau": 4.0, "mirostat_eta": 0.1, "max_tokens": 1024}
 }
 DEFAULT_INFERENCE_PRESET = "balanced"
 
-DEFAULT_SYSTEM_PROMPT = "You are ZOTHEOS, an ethical AI developed to help humanity. Provide clear, concise, and helpful responses. Be respectful and avoid harmful content."
+DEFAULT_SYSTEM_PROMPT = "You are ZOTHEOS, an ethical AI developed to help humanity. Provide clear, concise, and helpful responses."
 SYSTEM_PERSONAS = {
-    "default": DEFAULT_SYSTEM_PROMPT, "helpful_assistant": "You are a helpful AI assistant.", "philosopher": "You are an AI philosopher.",
-    "coder": "You are an expert AI programmer.", "concise_summarizer": "You are an AI tasked with providing very concise summaries."
+    "default": DEFAULT_SYSTEM_PROMPT, "helpful_assistant": "You are a helpful AI assistant.",
+    "philosopher": "You are an AI philosopher.", "coder": "You are an expert AI programmer.",
+    "concise_summarizer": "You are an AI tasked with providing very concise summaries."
 }
 MODEL_ROLES = {"mistral": "analyst", "gemma": "humanist", "qwen": "skeptic"}
 MODEL_ROLE_SYSTEM_PROMPTS = {
@@ -102,6 +89,6 @@ MODEL_ROLE_SYSTEM_PROMPTS = {
     "skeptic": "You are a critical evaluator and a respectful skeptic.", "general": DEFAULT_SYSTEM_PROMPT
 }
 
-ZOTHEOS_VERSION = "Public Beta 1.8 (Live)"
+ZOTHEOS_VERSION = "Public Beta 1.9 (TheBloke Gemma)"
 logger.info(f"Config settings loaded. Version: {ZOTHEOS_VERSION}")
 logger.info(f"Web Mode: {IS_WEB_MODE}")