videollama3/constants.py

CONTROLLER_HEART_BEAT_EXPIRATION = 30
WORKER_HEART_BEAT_INTERVAL = 15

LOGDIR = "."

# Model Constants
IGNORE_INDEX = -100

# Image arguments
IMAGE_TOKEN_INDEX = -200
DEFAULT_IMAGE_TOKEN = "<image>"
DEFAULT_IMAGE_PATCH_TOKEN = "<im_patch>"
DEFAULT_IM_START_TOKEN = "<im_start>"
DEFAULT_IM_END_TOKEN = "<im_end>"
IMAGE_PLACEHOLDER = "<image-placeholder>"

# Video arguments
VIDEO_TOKEN_INDEX = -201
DEFAULT_VIDEO_TOKEN = "<video>"
NUM_FRAMES = 128
MAX_FRAMES = 768
NUM_FRAMES_PER_SECOND = 1

# Region arguments
REGION_TOKEN = "<REGION>"

# Audio arguments
AUDIO_TOKEN_INDEX = -202
DEFAULT_AUDIO_TOKEN = "<audio>"

# Stream arguments
STREAM_START_TOKEN = "<|stream_start|>"
STREAM_END_TOKEN = "<|stream_end|>"
STREAM_IMAGE_TOKEN = "<stream_image>"
STREAM_FPS = 2
STREAM_IMAGE_SIZE = 224
STREAM_DOWNSAMPLING = 4
STREAM_MAX_FRAMES = 400

MODAL_INDEX_MAP = {
    "<image>": -200,
    "<video>": -201,
    "<audio>": -202,
}

subimage_token_num=196