| accum_freq: 1 |
| aug_cfg: {} |
| batch_size: 16 |
| beta1: 0.9 |
| beta2: 0.98 |
| checkpoint_path: ./logs/2023_04_03-18_34_41-model_coca_ViT-L-14-lr_1e-05-b_16-j_4-p_amp/checkpoints |
| coca_caption_loss_weight: 2.0 |
| coca_contrastive_loss_weight: 1.0 |
| copy_codebase: False |
| csv_caption_key: title |
| csv_img_key: filepath |
| csv_separator: |
| dataset_resampled: False |
| dataset_type: webdataset |
| ddp_static_graph: False |
| debug: False |
| delete_previous_checkpoint: False |
| device: cuda:0 |
| dist_backend: nccl |
| dist_url: env:// |
| distill: False |
| distill_model: None |
| distill_pretrained: None |
| distributed: True |
| epochs: 1 |
| epochs_cooldown: None |
| eps: 1e-06 |
| force_custom_text: False |
| force_image_size: None |
| force_patch_dropout: None |
| force_quick_gelu: False |
| gather_with_grad: False |
| grad_checkpointing: False |
| grad_clip_norm: None |
| horovod: False |
| image_mean: None |
| image_std: None |
| imagenet_v2: None |
| imagenet_val: None |
| local_loss: False |
| local_rank: 0 |
| lock_image: False |
| lock_image_freeze_bn_stats: False |
| lock_image_unlocked_groups: 0 |
| lock_text: False |
| lock_text_freeze_layer_norm: False |
| lock_text_unlocked_layers: 0 |
| log_every_n_steps: 100 |
| log_level: 20 |
| log_local: False |
| log_path: ./logs/2023_04_03-18_34_41-model_coca_ViT-L-14-lr_1e-05-b_16-j_4-p_amp/out.log |
| logs: ./logs/ |
| lr: 1e-05 |
| lr_cooldown_end: 0.0 |
| lr_cooldown_power: 1.0 |
| lr_scheduler: cosine |
| model: coca_ViT-L-14 |
| name: 2023_04_03-18_34_41-model_coca_ViT-L-14-lr_1e-05-b_16-j_4-p_amp |
| no_set_device_rank: False |
| precision: amp |
| pretrained: laion2B-s13B-b90k |
| pretrained_image: False |
| rank: 0 |
| remote_sync: None |
| remote_sync_frequency: 300 |
| remote_sync_protocol: s3 |
| report_to: |
| resume: None |
| save_frequency: 1 |
| save_most_recent: False |
| seed: 0 |
| skip_scheduler: False |
| tensorboard: False |
| tensorboard_path: |
| torchscript: False |
| trace: False |
| train_data: /home/vision/seonghoon/caption_dataset/cc3m/{00000..00331}.tar |
| train_data_upsampling_factors: None |
| train_num_samples: 2339077 |
| use_bn_sync: False |
| val_data: None |
| val_frequency: 1 |
| val_num_samples: None |
| wandb: False |
| wandb_notes: |
| wandb_project_name: open-clip |
| warmup: 1000 |
| wd: 0.1 |
| workers: 4 |
| world_size: 4 |
| zeroshot_frequency: 2 |
|
|
