| { |
| "exp_name": "essen", |
| "seed": 0, |
| "batch_size": 512, |
| "per_gpu_batchsize": 128, |
| "eval_batch_size": 32, |
| "load_path": "", |
| "model_type": "two-tower", |
| "encoder": "google/electra-small-discriminator", |
| "pooler_type": "double", |
| "tokenizer": "bert-base-uncased", |
| "random_init_encoder": false, |
| "encoder_manual_configuration": false, |
| "hidden_size": 192, |
| "num_heads": 4, |
| "num_layers": 12, |
| "mlp_ratio": 4, |
| "drop_rate": 0.1, |
| "embedding_size": 96, |
| "image_encoder": "google/efficientnet-b2", |
| "random_init_vision_encoder": false, |
| "image_encoder_manual_configuration": false, |
| "image_encoder_hidden_size": 192, |
| "image_encoder_num_heads": 4, |
| "image_encoder_num_layers": 12, |
| "image_encoder_mlp_ratio": 4, |
| "image_encoder_drop_rate": 0.1, |
| "image_encoder_embedding_size": 128, |
| "image_size": 224, |
| "original_image_size": 224, |
| "patch_size": 16, |
| "image_only": false, |
| "train_transform_keys": [ |
| "imagenet" |
| ], |
| "val_transform_keys": [ |
| "imagenet" |
| ], |
| "text_encoder": "google/electra-small-discriminator", |
| "random_init_text_encoder": false, |
| "text_encoder_manual_configuration": false, |
| "text_encoder_hidden_size": 192, |
| "text_encoder_num_heads": 4, |
| "text_encoder_num_layers": 12, |
| "text_encoder_mlp_ratio": 4, |
| "text_encoder_drop_rate": 0.1, |
| "text_encoder_embedding_size": 64, |
| "max_text_len": 50, |
| "vocab_size": 30522, |
| "cross_layer_hidden_size": 256, |
| "num_cross_layers": 6, |
| "num_cross_layer_heads": 4, |
| "cross_layer_mlp_ratio": 4, |
| "cross_layer_drop_rate": 0.1, |
| "freeze_image_encoder": false, |
| "freeze_text_encoder": true, |
| "freeze_cross_modal_layers": false, |
| "whole_word_masking": true, |
| "mlm_prob": 0.15, |
| "draw_false_image": 1, |
| "draw_false_text": 0, |
| "get_recall_metric": false, |
| "vqav2_label_size": 3129, |
| "max_bb": 20, |
| "ref_res_head_layers": 2, |
| "optim_type": "adamw", |
| "learning_rate": 0.0001, |
| "weight_decay": 0.01, |
| "decay_power": 1, |
| "max_epoch": null, |
| "max_steps": 100000, |
| "warmup_steps": 0.1, |
| "end_lr": 0, |
| "lr_mult_head": 5, |
| "lr_mult_cross_modal": 5, |
| "fast_dev_run": false, |
| "val_check_interval": 1.0, |
| "test_only": false, |
| "num_gpus": 2, |
| "num_nodes": 1, |
| "num_workers": 12, |
| "precision": 32 |
| } |
