63 lines
1.5 KiB
Bash
63 lines
1.5 KiB
Bash
|
# Discord Token
|
||
|
THE_TOKEN = ""
|
||
|
|
||
|
# The Channel IDs the bot will operate in seperated by commas
|
||
|
CHANNEL_IDS =
|
||
|
|
||
|
INIT_PROMPT="You are an assitant"
|
||
|
|
||
|
# Key for AbuseDB
|
||
|
ABUSE_KEY=
|
||
|
|
||
|
# When a message is too large for discord we chunk the response into seperate messages.
|
||
|
# To ensure we do not rate limit the bot we send these at a delay interval.
|
||
|
# DEFAULT: 3 a good setting is between 3 and 7 seconds.
|
||
|
OVERFLOW_DELAY=3
|
||
|
|
||
|
# Max Content to fetch from given URLs
|
||
|
MAX_CONTENT_LENGTH=8000
|
||
|
|
||
|
# Max tokens for Generations
|
||
|
MAX_TOKENS = 4000
|
||
|
|
||
|
# ROOT_IP is only used when running the bot without docker compose
|
||
|
ROOT_IP = 192.168.0.8
|
||
|
|
||
|
# PORT is only used when running the bot without docker compose
|
||
|
ROOT_PORT = 3000
|
||
|
|
||
|
# 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
|
||
|
|
||
|
# Custom Stuff internal to my use cases.
|
||
|
PATH_KEY=""
|
||
|
API_KEY=""
|
||
|
|
||
|
API_PATH=""
|