| { | |
| "dataset_reader": { | |
| "type": "transformer_squad", | |
| "length_limit": 512, | |
| "transformer_model_name": "roberta-large" | |
| }, | |
| "model": { | |
| "type": "transformer_qa", | |
| "transformer_model_name": "roberta-large" | |
| }, | |
| "train_data_path": "https://allennlp.s3.amazonaws.com/datasets/squad/squad-train-v2.0.json", | |
| "validation_data_path": "https://allennlp.s3.amazonaws.com/datasets/squad/squad-dev-v2.0.json", | |
| "trainer": { | |
| "callbacks": [ | |
| "tensorboard" | |
| ], | |
| "grad_clipping": 1, | |
| "learning_rate_scheduler": { | |
| "type": "slanted_triangular", | |
| "cut_frac": 0.1, | |
| "num_epochs": 5 | |
| }, | |
| "num_epochs": 5, | |
| "optimizer": { | |
| "type": "huggingface_adamw", | |
| "eps": 1e-08, | |
| "lr": 2e-05, | |
| "parameter_groups": [ | |
| [ | |
| [ | |
| "bias", | |
| "LayerNorm\\.weight", | |
| "layer_norm\\.weight" | |
| ], | |
| { | |
| "weight_decay": 0 | |
| } | |
| ] | |
| ], | |
| "weight_decay": 0 | |
| }, | |
| "validation_metric": "+per_instance_f1" | |
| }, | |
| "vocabulary": { | |
| "type": "empty" | |
| }, | |
| "data_loader": { | |
| "batch_sampler": { | |
| "type": "bucket", | |
| "batch_size": 16 | |
| } | |
| }, | |
| "numpy_seed": 100, | |
| "pytorch_seed": 100, | |
| "random_seed": 100 | |
| } |