INIT_PROMPT="You are an assitant"

# Max tokens for Generations
MAX_TOKENS = 4000

# Directory to your models (llama.cpp specfic settings)
DATA_DIR = /models

# Enable Expirmental Message Caches (Limited to single session)
# Cache will use ~1.4 GB or MORE of RAM. ONLY ENABLE IF YOUR SYSTEM CAN HANDLE THIS.
CACHE = 1

CACHE_TYPE = "ram"

# Set number of threads to use, currently, a standard thread will utilize 1 whole core
# I usually will set this between all cores I physcally have OR 2 cores less to allow for other processes.
N_THREADS = 8

# Always use MMAP unless you know what you are doing
USE_MMAP=1

# Only use MLOCK if you know what it does!
USE_MLOCK=0

# The higher the number the more hard core.
REPEAT_PENALTY=1

# GPU SPECIFIC SETTINGS BELOW

GPU=1

N_GPU_LAYERS=35 

PYTHONUNBUFFERED=1