flickr30k_blip-base_v2-0.yaml

# @package _global_

# to execute this experiment run:
# python train.py experiment=example

defaults:
  - override /data: flickr30k
  - override /model: blip_base_pretrained
  - override /callbacks: default
  - override /trainer: default
  - override /logger: wandb

# all parameters below will be merged with parameters from default configurations set above
# this allows you to overwrite only specified parameters

tags: ["v1", "flickr30k", "blip_base"]

seed: 42

trainer:
  min_epochs: 5
  max_epochs: 10
  gradient_clip_val: 0.0

model:
  optimizer:
    _target_: torch.optim.AdamW
    _partial_: true
    lr: 0.000001
    weight_decay: 0.05

  scheduler:
    _target_: torch.optim.lr_scheduler.CosineAnnealingLR
    _partial_: true
    T_max: ${trainer.max_epochs}
    eta_min: 0.0000001

  net:
    _target_: transformers.AutoModelForVision2Seq.from_pretrained
    pretrained_model_name_or_path: Salesforce/blip-image-captioning-base
    cache_dir: models/huggingface

  processor:
    _target_: transformers.AutoProcessor.from_pretrained
    pretrained_model_name_or_path: Salesforce/blip-image-captioning-base
    cache_dir: models/huggingface

  compile: false

  hf_repo_id: tanthinhdt/blip-base_flickr30k

data:
  use_all_comments: False
  comment_number: 0
  padding: longest
  train_val_test_split: [0.7, 0.3]
  batch_size: 16
  num_workers: 8

logger:
  wandb:
    name: flickr30k_blip-base_v2-0
    id: null
    project: "IMCAP"
    tags: ${tags}