KB-VQA-E / my_model /config /captioning_config.py
m7mdal7aj's picture
Update my_model/config/captioning_config.py
ccefb55 verified
import torch
# Configuration parameters
MODEL_TYPE = "i_blip"
MAX_IMAGE_SIZE = 1024
MIN_LENGTH = 150
MAX_NEW_TOKENS = 400
MODEL_PATH = "m7mdal7aj/captioner"
LOAD_IN_8BIT = False
LOAD_IN_4BIT = True
TORCH_DTYPE = torch.float16
DEVICE_MAP = "auto"
LOW_CPU_MEM_USAGE = True
SKIP_SPECIAL_TOKENS = True
PROMPT = 'Provide a comprehensive and detailed description of the following image. Focus on identifying and describing every element in the scene, including all people (along with their gender, age, color and any prominent feature), animals along with their breed and all objects, their count, their positions, and any actions or interactions taking place. Pay special attention to the positioning of limbs and hands, and any objects they might be holding or interacting with. Describe texts, colors, textures, setting, atmosphere, mood, and any indicators of the time of day, such as the quality of light, shadows. Ensure to capture both the obvious and subtle elements for a complete understanding of the image. Answer as if you were looking at the image'